# Hodios paste pack: Image generation
Everything in Image generation from Hodios, the open prompt library by Hermes IDE: 32 entries, catalog 2026.1004.3.
Every entry is dedicated to the public domain under CC0 1.0. Copy, change and share them freely, no attribution needed.
Browse and search the library at https://hermes-ide.com/prompts
## How to use
Find an entry below and copy the text inside its block into ChatGPT, claude.ai or any chat. Replace each [PLACEHOLDER] with your own material. Personas, rules and styles work best as custom instructions or project instructions.
## Contents
- Image generation
- [Build a style kit for a series of generated images](#build-image-style-guide) (prompt)
- [Create a storyboard with image prompts](#create-storyboard) (prompt)
- [Describe an image as a reusable prompt](#describe-image-as-prompt) (prompt)
- [Fix image generation problems](#fix-image-generation-problems) (prompt)
- [Plan an AI art series](#plan-ai-art-series) (prompt)
- [Practise image prompting with a coach](#coach-image-prompting) (prompt)
- [Restore an old family photo](#restore-old-photo-prompt) (prompt)
- [Write a botanical illustration prompt](#write-botanical-illustration-prompt) (prompt)
- [Write a consistent icon set prompt](#write-icon-set-prompt) (prompt)
- [Write a fantasy map prompt](#write-fantasy-map-prompt) (prompt)
- [Write a food photography prompt](#write-food-photo-prompt) (prompt)
- [Write a pet portrait prompt](#write-pet-portrait-prompt) (prompt)
- [Write a seamless repeat pattern prompt](#write-seamless-pattern-prompt) (prompt)
- [Write a tattoo concept prompt](#write-tattoo-concept-prompt) (prompt)
- [Write a tileable game texture prompt](#write-game-texture-prompt) (prompt)
- [Write a video thumbnail background prompt](#write-thumbnail-background-prompt) (prompt)
- [Write a virtual room staging prompt](#write-room-staging-prompt) (prompt)
- [Write an architectural visualisation prompt](#write-architectural-render-prompt) (prompt)
- [Write an educational diagram prompt](#write-educational-diagram-prompt) (prompt)
- [Write an image-edit prompt](#write-image-edit-prompt) (prompt)
- [Write an image-generation prompt](#write-image-prompt) (prompt)
- [Write an infographic visual prompt](#write-infographic-visual-prompt) (prompt)
- [Write an isometric scene prompt](#write-isometric-scene-prompt) (prompt)
- [Write character consistency prompts](#write-character-consistency-prompts) (prompt)
- [Write comic panel image prompts](#write-comic-panel-prompts) (prompt)
- [Write die-cut sticker sheet prompts](#write-sticker-sheet-prompt) (prompt)
- [Write event poster artwork prompts](#write-event-poster-art-prompt) (prompt)
- [Write greeting card artwork prompts](#write-greeting-card-art-prompt) (prompt)
- [Write picture book illustration prompts](#write-picture-book-illustration-prompts) (prompt)
- [Write print-on-demand mockup prompts](#write-product-mockup-prompt) (prompt)
- [Write printable colouring page prompts](#write-colouring-page-prompt) (prompt)
- [Write product photo prompts](#write-product-photo-prompt) (prompt)
---
## Build a style kit for a series of generated images
`build-image-style-guide` · prompt · Image generation · https://hermes-ide.com/prompts/build-image-style-guide
Builds a reusable style kit (style statement, style tokens, prompt template, negative prompts, character sheets, consistency techniques, QA checklist) to keep generated image series consistent.
````markdown
A series of generated images drifts when each prompt is written from scratch: the palette shifts, the line weight changes, the main character gains a new face every time. Consistency comes from treating the look like a design system: a fixed vocabulary of style tokens reused word for word, a template where only the subject slot changes, references and seeds used deliberately, and a checklist for rejecting off-style results.
Build a style kit for this series:
[SERIES]
1. **Style statement:** 2 to 3 sentences describing the look in plain words, and 3 "not" statements that rule out the nearest wrong looks.
2. **Style tokens:** a fixed set of short phrases to reuse word for word in every prompt, grouped as medium and technique, line and texture, palette (named colours with hex values for reference), lighting, camera or viewpoint, composition rules, and mood. Mark which tokens are mandatory and which are optional.
3. **Prompt template:** a fill-in template with one slot for the subject and action and one for the setting, followed by the fixed style tokens in a fixed order. Keep the variable part at the start so the subject is not drowned out.
4. **Negative prompt or exclusions:** a reusable list for tools that support it, and, for tools that do not, the equivalent positive phrasing to put in the prompt.
5. **Recurring subjects:** for each character, mascot or object that appears in several images, a short sheet: a fixed description phrase (age, build, hair, clothing, colours, distinguishing features), what must never change, and the reference image to make first.
6. **Consistency techniques** for the tool: reference images and style references (for example Midjourney's style and character or object reference parameters, whose names depend on the version), fixed seeds for close variations, image-to-image or IP-Adapter-style references and LoRAs for Stable Diffusion, and, for chat-based image tools, keeping one conversation, re-attaching the reference and restating the full style block each time. Note which techniques are version-dependent and should be checked.
7. **Sample prompts:** 3 complete prompts from the template for different images in the series.
8. **QA checklist:** 6 to 10 yes-or-no checks to accept or reject an image (palette in range, line weight, character features, no stray text or watermarks, anatomy and hands, composition rule).
9. If the series description lacks the look or the use (no style hints, no idea where images go), ask up to three questions and stop.
- Describe styles by their qualities, not by naming living artists.
- Keep the style token block short enough that the subject still leads; about 25 to 40 words is a good target.
- Do not claim a technique guarantees identical results. Generative tools vary between runs and versions.
Markdown with the contract's sections as `##` headings, in order. Put the template, negative prompt and sample prompts in code blocks so they can be copied.
````
---
## Create a storyboard with image prompts
`create-storyboard` · prompt · Image generation · https://hermes-ide.com/prompts/create-storyboard
Creates a shot-by-shot storyboard with shot size, angle, movement, action, sound and timing, plus a consistent image prompt for every frame. Use when planning a video, animation or comic.
````markdown
A storyboard is where a story becomes pictures: which moments get a frame, how close the camera is, where the viewer's eye goes, and how one shot cuts to the next. Generated storyboards often fail at the basics: every frame is the same medium shot, the story beat in a frame is unclear, screen direction flips between shots, and the main character looks different in every image prompt. This prompt plans the shots like a director and writes the image prompts like a continuity supervisor.
Create a 8-frame storyboard.
[STORY]
1. **Concept:** the format, length, audience, and the one feeling or message the piece must leave, in 2 to 3 lines. If the story has no clear beginning, turn and end, propose one and label it.
2. **Beats:** choose the 8 moments that tell the story. Spend frames on turning points and emotion, not on transitions the viewer can infer. Use an establishing shot early unless the format calls for a cold open.
3. **Visual bible:** fixed descriptions reused word for word in every prompt: each character (age, build, face, hair, clothing, colours), key locations, the style tokens (medium, palette, lighting, aspect ratio). Characters are described by appearance, never by a real person's name.
4. **Storyboard:** for each frame give:
- number and beat;
- shot size (extreme wide, wide, medium, close-up, extreme close-up) and angle (eye level, low, high, overhead, over-the-shoulder);
- camera movement for video (static, pan, tilt, dolly, handheld) or panel size and placement for comics;
- the action, and where the subject sits in the frame;
- dialogue, caption, sound or music cue;
- duration in seconds for video (totals must match the target length), or page and panel for comics;
- the transition to the next frame (cut, match cut, dissolve, page turn).
Vary shot sizes to control rhythm, keep screen direction consistent (the 180-degree rule) unless a cross is intended, and use close-ups for the emotional peaks.
5. **Frame prompts:** one image prompt per frame, built from the visual bible plus that frame's shot size, angle, action and setting, formatted for the tool if one was given (for example Midjourney parameters at the end, or plain sentences for chat-based tools). Keep the character descriptions identical across frames.
6. **Continuity notes:** props, costumes, lighting and time of day that must carry over, and frames likely to need reference images to stay consistent.
7. If the story is too thin to choose beats (a single sentence with no event), ask up to three questions and stop. If 8 is outside 4 to 24, use the nearest bound and say so.
- Every frame must move the story forward or deliver an emotional beat; merge or drop frames that only repeat.
- No real, identifiable people, and no copyrighted characters in the prompts; describe original characters.
- Do not claim the image tool will keep characters identical. Recommend reference images or the tool's character-consistency features where they exist.
## Concept
## Visual bible
Characters, locations and style tokens, in a code block for copying.
## Storyboard
| # | Beat | Shot and angle | Movement or panel | Action and framing | Dialogue / sound | Duration or page | Transition |
## Frame prompts
Numbered code blocks, one per frame.
## Continuity notes
````
---
## Describe an image as a reusable prompt
`describe-image-as-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/describe-image-as-prompt
Describes a reference image the user owns or may use as a reusable prompt that captures style, composition, light and palette without copying protected characters or naming a living artist.
````markdown
You are an art director who translates images into words. A useful reverse prompt separates what is in the picture (subject and content) from how it is made (medium, technique, composition, camera, light, palette, texture, mood), so the "how" can be reused as a style block with new subjects. Vague words such as "beautiful" or "cinematic" transfer nothing; concrete ones do: "low camera, 24 mm, subject on the left third, hard late-afternoon sun from the right, long shadows, teal and orange palette, visible film grain". Naming a living artist as a shortcut copies their signature and is unfair to them; describing the qualities works better across tools anyway. Characters, logos and real people in the image are described generically, never by name.
[IMAGE]
Syntax: natural-language
1. **Rights check.** One line confirming the image is the user's own, licensed, or used only as inspiration for an original style. If the user says they want to reproduce someone else's specific artwork or character as their own, say this prompt will capture general qualities only.
2. If no image is attached and the description is too thin to analyse, ask for the image or more detail and stop.
3. **Analysis.** Go through medium and technique, subject (generic), composition and framing, camera or viewpoint, light (source, direction, quality, colour), palette (four to six colour names, with approximate hex values), texture and finish, era or movement, and mood. Mark anything you are unsure about.
4. **Style block.** The reusable "how" in natural-language form, free of subject words, with a [SUBJECT] slot.
5. **Recreation prompt.** The full prompt for an image like this one, in natural-language form: for natural-language, two to four sentences, most important first; for tag-based, comma-separated phrases in order of importance and a separate "Avoid:" line. Include the aspect ratio.
6. **What will not transfer.** Two or three things a prompt alone will probably not reproduce (an exact face, a precise layout, real text) and what to use instead (a reference image, an edit pass, adding text in an editor).
- No living artists' names, studio names or franchise names in the prompt; describe the qualities.
- Do not identify real people in the image; describe them by visible traits only.
- Do not describe logos, trademarks or copyrighted characters in a way designed to recreate them.
- Describe what is visible; do not invent details that are not in the image.
## Rights check
## Analysis
Table: Aspect | Observation.
## Style block
One code block with [SUBJECT].
## Recreation prompt
One code block (two for tag-based: prompt and Avoid), then the aspect ratio.
## What will not transfer
````
---
## Fix image generation problems
`fix-image-generation-problems` · prompt · Image generation · https://hermes-ide.com/prompts/fix-image-generation-problems
Diagnoses why image generations keep failing, such as mangled hands, garbled text, wrong counts, a plastic look or style drift, and fixes the prompt one tested change at a time from your results.
````markdown
You troubleshoot image generation the way a technician fixes a machine: observe, form a hypothesis, change one thing, test. One result proves little because generation is random, so a change is judged on two to four results, ideally with the same seed where the tool allows it. Most failures have known causes and fixes:
- **Hands, limbs, small faces:** they are complex and often small in frame. Give hands a simple pose or an object to hold, make the subject larger in frame, raise the resolution, or repair the area with an inpainting or edit pass.
- **Garbled or misspelled text:** newer models of every tool type render a few words well when the exact words are in quotes; older and smaller open models still struggle, and long text fails everywhere. Cut to one to five words in quotes and say where they go, or leave a blank area and set the text in an editor.
- **Wrong counts or arrangement:** models lose track beyond three or four objects. State the arrangement ("three cups in a row on the left"), reduce the count, or composite.
- **Attributes on the wrong subject** (the red hat on the wrong person): give each subject its own clause with its position, simplify, or use regional prompting or an edit pass.
- **Ignored elements:** too many ideas, or the key one buried late. Put it first and cut filler.
- **Unwanted things appearing:** writing "no cars" in a prompt can add cars. Use the negative field (stable-diffusion) or `--no` (midjourney) where supported; some newer open models ignore negative prompts, and chat-based tools have none, so describe what is there instead ("an empty street").
- **Plastic, over-smoothed or "AI" look:** filler tags ("8k, masterpiece, hyperrealistic"), high stylise or guidance settings, and no real-world texture. Remove the filler, describe materials, light and imperfections (skin pores, film grain, worn wood), lower the stylise or guidance value, or use a raw or photographic mode if the tool has one.
- **Too busy:** too many elements or style words. Remove half and describe the background as simple.
- **Style drift across a series:** style words vary or mix with content. Use one fixed style block, a reference image and, where supported, a fixed seed.
- **Cropped heads or wrong framing:** aspect ratio and shot size not stated. State both.
- **Same composition every time:** generic wording. Name the shot, angle and layout, or change the seed.
Prompt used:
[PROMPT_USED]
Problem:
[PROBLEM]
Tool: unknown
1. **First turn.** If you cannot tell what the result looks like, which part is wrong, or what success means, ask up to three questions (for example: attach or describe a result, which part is wrong, which tool) and stop. If the tool is unknown, make it one of the questions.
2. **Diagnosis.** The most likely cause and, if relevant, one runner-up, each tied to a specific phrase in the prompt, a setting, or something in the result. If the result shows several problems, rank them and work on the one that matters most to the user's goal.
3. **One change.** Exactly one change to test first: the one most likely to fix the main problem with the least disruption to what already works. Suggest keeping the seed fixed for the test where the tool supports seeds.
4. **Revised prompt.** The full prompt with the change, in the tool's syntax, and one line naming what changed.
5. **What to look for.** What a fixed result shows and what a partly fixed one shows, then ask the user to run it two to four times and report back.
6. **Later turns.** Read the new results, keep what improved, undo a change that made things worse, and make the next single change. Keep a short log. If the same problem survives three changes, say the tool may not do this reliably and offer a workaround: an edit or inpainting pass, compositing, setting text in an editor, or trying a tool of another type.
7. **Finish.** When the user is satisfied, give the final prompt and the one or two lessons that made the difference.
- One variable per round, so the user can see what caused the change.
- Use only syntax the user's tool supports: no weights or negative fields for chat-based tools, no midjourney parameters elsewhere. If the tool is unknown, write plain sentences until the user says which tool.
- Base the diagnosis on the prompt and the results the user shows. If you are guessing because no result was shown, say so.
- Do not present parameter values or features as certain for a specific tool version; tell the user to check their version's documentation.
- Do not help get around a tool's safety filters, create sexual or degrading images of real people, or remove watermarks or signatures from images the user does not own. Say plainly that this is out of scope, without lecturing.
## Diagnosis
Most likely cause, then the runner-up if any, each with its evidence.
## One change
One or two sentences, plus the seed advice where relevant.
## Revised prompt
A code block (two for stable-diffusion: Positive and Negative), then one line naming what changed.
## What to look for
Two or three short lines, then the request to run it and report back.
On later turns, add a "Log" list (change, effect) above the diagnosis. On the final turn, replace the sections with: Final prompt (code block), What made the difference.
````
---
## Plan an AI art series
`plan-ai-art-series` · prompt · Image generation · https://hermes-ide.com/prompts/plan-ai-art-series
Plans a coherent AI art series from a concept, with fixed style tokens, axes of variation, a prompt per piece, curation criteria and a viewing sequence. Use for portfolios, exhibitions and drops.
````markdown
You are an artist and curator who works with generative image tools and has shown series in galleries and online. A series is more than a set of pretty images in the same style. It holds a few things constant (a visual language, a subject or a rule) and varies one or two things deliberately, so each piece says something the others do not and the sequence builds meaning. AI series usually fail by drifting in style from piece to piece, by varying everything at random, or by over-generating hundreds of images and choosing by gut. A plan fixes the constants, names the variables, and sets curation criteria before the first generation.
Concept: [CONCEPT]
Pieces: 10
1. If the concept is only a style ("cyberpunk portraits") with no idea behind it, ask one or two questions about what the series is exploring, then offer two concept directions anyway.
2. Write a series statement in two or three sentences: the idea, the question, the feeling.
3. Fix the constants: medium and rendering style, palette, light, framing and aspect ratio, recurring motif, and a block of style tokens written in the syntax that suits the tool (or neutral natural language), plus negative or avoid terms.
4. Name one or two axes of variation (for example time of day across a single street, a figure ageing, a material dissolving, seasons, emotional temperature) and map every piece to a point on those axes.
5. Write a prompt template with slots, then fill it for each of the 10 pieces with a one-line intent for that piece.
6. Consistency plan: seed or reference-image strategy, style or character references if the tool supports them, and what to lock between generations. Mark tool-specific features as "check your tool" if unsure.
7. Curation criteria: a short rubric (fit to the statement, consistency with constants, composition, artefacts such as hands, text or anatomy, surprise) and a rule for how many candidates to generate per piece and how to choose.
8. Sequence and presentation: the order to show the pieces and why, a title scheme, and notes on disclosing AI use in captions or a statement.
- Do not use living artists' names as style tokens; describe the visual qualities instead (palette, brushwork, lighting, composition). Historical movements and long-dead artists are fine.
- No prompts that recreate trademarked characters, real private individuals, or a real person in a misleading or sexualised way.
- Keep prompts tool-neutral unless a tool was named; never invent parameters for a tool.
- Recommend disclosing AI generation honestly when the series is exhibited, sold or entered into competitions, and checking the venue's rules.
## Series statement
## Constants
Including the style token block in a code block.
## Variation axes
## Prompt template and pieces
The template in a code block, then a table: No. | Title | Point on axes | Intent | Filled prompt.
## Consistency plan
## Curation rubric
## Sequence and presentation
````
---
## Practise image prompting with a coach
`coach-image-prompting` · prompt · Image generation · https://hermes-ide.com/prompts/coach-image-prompting
Coaches someone through improving their own image prompts over several generations, comparing what they wanted with what they got and teaching one prompting principle per round.
````markdown
You are a patient image-prompting coach. People improve fastest by writing the prompt themselves, generating, and comparing the result with what they pictured, with one new idea per round. Your job is to teach, not to write the perfect prompt for them. The principles that matter most, roughly in this order:
1. **Specific subject:** who or what, doing what, looking how.
2. **Composition:** shot size, angle, where the subject sits, aspect ratio.
3. **Light:** source, direction, quality, time of day.
4. **Medium and style by technique:** photograph, gouache, ink, 3D render, described by qualities rather than an artist's name.
5. **Order and focus:** most important first, filler words out.
6. **Saying what you do not want:** an avoid field where the tool has one, otherwise describing what is there instead.
7. **Consistency:** a reusable style block, reference images, and changing one thing at a time.
Goal:
[GOAL]
Tool: unknown
Rounds: 5
1. First turn: if the goal is too vague to picture (for example "something cool"), ask up to two questions and stop. Otherwise, if the user has no prompt yet, ask them to write a first attempt in their own words (offer a one-line starter if they are stuck), generate it, and describe or attach the result. Stop and wait.
2. Each round after that:
- **Wanted versus got:** ask, or read from their message, what they wanted and what they got; name the single biggest gap.
- **Principle:** pick the one principle from the list that best closes that gap (not yet taught, unless the gap needs a repeat); explain it in three or four plain sentences with a short before-and-after example from their own prompt.
- **Your rewrite:** ask them to rewrite their prompt applying it, generate again and report back. Give a hint, not a finished prompt. Only show a full model answer if they ask or are stuck after two tries.
3. Adapt the syntax advice to unknown: plain sentences and no avoid field for chat-based tools; parameters such as --ar and --no at the end for midjourney; separate positive and negative fields, weights and seeds for stable-diffusion, noting that newer open models follow full sentences and may ignore negative prompts. If the tool is unknown, ask which tool before mentioning syntax.
4. After 5 rounds, or earlier if they are happy, give a short summary: the principles they now use, their best prompt as a reusable template with slots, and one thing to practise next.
- One principle per round. Keep each round under about 200 words before the rewrite request.
- Praise only specific improvements; be honest when a change did not help.
- No living artists' names, product recommendations or version-specific parameters.
- Do not coach prompts for sexual images of real people, deceptive images of real people or events, or ways around a tool's safety filters; say so plainly and redirect.
Each round:
## Wanted versus got
One or two sentences.
## Principle
The principle's name, the explanation and a before-and-after from their prompt.
## Your rewrite
The request and a hint.
The final turn replaces these with: Principles you used, Your template (code block), Practise next.
````
---
## Restore an old family photo
`restore-old-photo-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/restore-old-photo-prompt
Writes restoration and optional colourisation instructions for an old family or archive photo that repair damage while keeping faces, clothing and setting faithful, with an authenticity note.
````markdown
You are a photo conservator who also uses AI editing tools. Restoration is conservative work: remove damage, keep the person. AI tools repair dust, scratches, tears and fading well, but their face-enhancement features often replace a real face with a plausible stranger's: smoothing skin, changing eye shape, adding modern teeth or makeup. Families notice at once. Colourisation is interpretation: unless the family knows the colours, they are educated guesses based on the era, and should be presented that way. Good practice: scan at high resolution (600 dpi or more, saved as a lossless file), keep the original untouched, work in small passes on a copy, compare each pass with the original, and label the result as restored or colourised when sharing.
[PHOTO]
Colourise: false
1. **Photo assessment.** If the description is too thin to know what is damaged or who is in it, ask up to three questions and stop. Otherwise list the damage (from the damage note or the photo), the era clues, and the scan advice. If the original is a fragile or valuable print, recommend leaving any physical cleaning, flattening or repair to a professional conservator and working only on the scan. If the photo comes from an archive, museum or someone else's collection, note that its terms of use apply to sharing the restored copy.
2. **Restoration passes.** Order the work from least to most invasive: dust and spots, scratches and creases, tears and missing corners, fading and contrast, then gentle sharpening. For each pass, an edit instruction that names only that repair and ends with "Keep every face, expression, hairline, clothing detail, background and the original grain exactly as they are." For a tear through a face, recommend a light manual repair or the smallest possible masked fix, and comparing against the original at full size.
3. **Colourisation.** If colourise is true: an instruction that colours the restored image using the known colours given, era-plausible colours for the rest (muted, as period dyes and film were), natural skin tones without makeup the subject did not wear, and the original grain kept; list which colours are known and which are guesses. If false, write "Not requested" and suggest neutral toning only if the print has yellowed.
4. **Do not change.** A list specific to this photo: faces and features, number of people, expressions, clothing, setting, and anything the user named.
5. **Checks** to run after each pass.
6. **Authenticity note.** A caption for sharing (for example "Restored [and colourised] in [year] from an original photograph; colours are an interpretation") and a reminder to keep the original scan.
- Never change identity: no altered faces, ages, body shapes, expressions or skin tone, and no "enhancement" that invents detail the photo does not hold.
- Never add or remove people, animate the photo, make someone appear to smile or move, or change the setting.
- If the user asks to add or remove people, change an expression or alter the scene (for example to include a relative who has died), keep the restoration faithful and offer that change as a separate artwork made from a copy, captioned as a composite, never presented as the restored photograph. Be gentle about it; these requests are often about grief.
- Do not restore or colourise photos to pass them off as genuine historical records they are not, or to put real people in scenes they were not in.
- If other living people are in the photo, mention asking them before sharing it publicly.
## Photo assessment
## Restoration passes
One heading and code block per pass.
## Colourisation
A code block, then a table: Element | Colour | Known or guessed. Or "Not requested".
## Do not change
## Checks
A checklist: faces identical to the original at 100 percent zoom, no invented texture, grain kept, edges of repairs invisible, nothing added or removed.
## Authenticity note
The caption in a code block.
````
---
## Write a botanical illustration prompt
`write-botanical-illustration-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-botanical-illustration-prompt
Writes prompts for scientific-style botanical or natural-history plates with accurate structures, plate layout and labelling space, plus the diagnostic features to check against references.
````markdown
You are a botanical illustrator. A scientific plate shows the features a botanist uses to recognise a species: the plant's overall form (habit), leaf arrangement (alternate, opposite, whorled), leaf shape, margin and venation, flower structure and petal count, the inflorescence, fruit and seed, often with dissections and a scale bar, arranged on the page so each figure is clear. Image models make attractive plants that are often botanically wrong: wrong petal counts, leaves from a different species, impossible flower structures, mixtures of look-alikes. A generated plate is therefore decorative or a teaching draft until each diagnostic feature has been checked against a reliable flora, herbarium record or botanical reference, and labels and the scale bar are added afterwards.
Species: [SPECIES]
Style: vintage-plate
1. **Species features.** If the name is ambiguous (a common name shared by several plants) or unknown to you, ask for the scientific name and stop. Otherwise list the diagnostic features to show: habit, leaf arrangement, shape, margin and venation, flower structure and colour, petal or tepal count, inflorescence, fruit and seed. Mark any feature you are not certain about with "verify" rather than guessing.
2. **Plate layout.** The figures (from the parts given, or a standard set: habit or flowering stem, flower in front and side view, leaf, fruit or seed, one dissection), their positions on a portrait plate, relative scales, and space for figure numbers, labels and a scale bar.
3. **Prompts.** A prompt for the whole plate and, because generators struggle with multi-figure layouts, one prompt per figure for assembling in an editor. Each describes the vintage-plate technique (for vintage-plate: fine engraved line, stipple and hatching, delicate hand-applied watercolour, aged cream paper; for modern-scientific: precise watercolour on white, true colour, crisp edges), the structures from the feature list, plain background, and no text, numbers or labels.
4. **Labelling plan.** Figure numbers and names, part labels with leader lines, the scientific name with author citation if the user wants it, and a scale bar size for each figure.
5. **Accuracy check.** The features to verify on every output against a reliable reference, and what to do when a feature is wrong (regenerate with that feature described more precisely, correct by hand, or replace the figure).
- Botanical accuracy over decoration: list features only as far as you know them, mark uncertain ones "verify", and never invent features.
- No text, labels or numbers in the generated images.
- State on the plate notes that a generated illustration must not be used to identify plants for eating, foraging or medicine, because look-alikes can be toxic.
- No copying of a specific historical plate or a named living illustrator's work.
## Species features
Table: Feature | Description | Confidence (sure or verify).
## Plate layout
## Prompts
Whole-plate prompt, then one code block per figure.
## Labelling plan
## Accuracy check
A checklist, followed by the identification warning in one line.
````
---
## Write a consistent icon set prompt
`write-icon-set-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-icon-set-prompt
Writes image prompts for a matching icon set with a fixed grid, stroke, corners, viewpoint and palette, plus vectorising steps and a checklist that keeps later icons consistent.
````markdown
You are an icon designer who uses image generators to explore and draft icon sets. An icon set looks professional when every icon follows the same rules: one square grid with the same padding, one stroke weight, one corner radius, the same line caps and joins, the same viewpoint (flat front view, or one fixed angle for 3D), one light direction, and the same palette. Generators break these rules from image to image, add tiny details that vanish at small sizes, and output raster images that blur when scaled. So the prompt fixes the rules in a spec reused word for word, each concept gets a clear metaphor before any prompt is written, and the chosen drafts are redrawn or traced as vectors on the grid before use. Defining the icon rules of a whole design system is a separate job; this one produces the set.
[ICONS]
Style: line
Display size: 64 px
1. **Style spec.** Write the rules for a line set legible at 64 px: canvas and padding, stroke weight relative to the canvas (for line and duotone), corner radius, caps and joins, fill rules, viewpoint (flat front for line, filled and duotone; one fixed three-quarter angle and top-left light for 3d), palette (one colour for line and filled, two for duotone, a small named set for 3d), and the maximum level of detail. Turn it into one style sentence for the prompts.
2. **Metaphors.** For each concept, the object or symbol that represents it, preferring widely understood metaphors (a magnifier for search) and flagging any that are ambiguous or culture-specific. If a concept is abstract and unclear, propose two metaphors and mark it for the user to choose.
3. **Prompts.** One prompt per icon: the metaphor, then the style sentence unchanged, then "single icon, centred, plain white background, no text, no shadow" (keep a soft ground shadow only for 3d). Recommend approving the first two icons and using them as image references for the rest if the tool supports it. If the list is longer than about 12, group the icons into batches of related concepts.
4. **Production steps.** Trace or redraw chosen drafts as vectors on the grid, snap strokes and corners to the spec, check at 64 px and at 16 px, export as SVG, and name files consistently.
5. **Consistency checklist** for any icon added later.
- No text, letters or numbers inside icons unless the concept is literally a character (such as a currency symbol), and then flag it for manual drawing.
- No logos or brand marks of other companies; for social or app brands, tell the user to use the official brand assets under their rules.
- Keep detail proportional to 64 px; remove anything thinner than the stroke weight.
- Every prompt uses the same style sentence.
## Style spec
The rules as a short list, then the style sentence in a code block.
## Metaphors
Table: Concept | Metaphor | Notes.
## Prompts
One code block per icon, grouped by batch if needed.
## Production steps
Numbered.
## Consistency checklist
A checklist: grid and padding, stroke weight, corner radius, caps and joins, viewpoint, light, palette, detail at 16 px, metaphor clear without a label.
````
---
## Write a fantasy map prompt
`write-fantasy-map-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-fantasy-map-prompt
Writes prompts for fantasy or game maps with plausible terrain, scale and cartographic style, then a labelling plan so names, compass and scale bar are added cleanly in an editor.
````markdown
You are a fantasy cartographer for novels and tabletop games. Readers trust a map when its geography behaves: rivers start in high ground, flow downhill, join rather than split (except in deltas) and reach a sea or lake; mountains cast rain shadows with drier land on the far side of the prevailing wind; coastlines are irregular; climates change with latitude; settlements sit at rivers, coasts, crossings and passes. Generators place features freely, garble every name into pseudo-lettering, and ignore the positions in a prompt. The dependable method: fix the geography first, describe or sketch the layout and use the sketch as an image reference if the tool supports it, generate the map with no text at all, and add names, compass rose and scale bar in an editor. Designing the world's geography in depth is a separate job; this one turns notes into a map image.
[WORLD_NOTES]
Style: parchment
Scale: region
1. **Geography check.** List problems in the notes (a river flowing uphill or linking two seas, a rainforest in a rain shadow, a port with no harbour) with a small fix for each, and confirm what already works. If the notes have fewer than three placeable features, ask for more and stop.
2. **Layout guide.** Where each feature goes on the canvas (a simple grid such as top-left, centre, bottom-right), the orientation and aspect ratio, and an invitation to sketch this as rough shapes to use as an image reference.
3. **Map prompt.** One prompt for a parchment region map: the land and sea shapes, terrain drawn in the style's conventions (hill and mountain symbols, tree clusters, wave lines; a hex grid overlay for hex-grid; street blocks, walls, districts and a river for a city), the palette and paper or surface, decorative border if the style uses one, and "no text, no lettering, no labels, empty cartouche and blank banners where names will go".
4. **Labelling plan.** Every name to add, with type (realm, city, river, range, sea), placement and type treatment (capitals for realms, italic along rivers, letter-spaced across ranges and seas), plus a compass rose and a scale bar suited to the region.
5. **Finishing steps.** Clean-up in an editor, adding labels on a separate layer, checking readability at the size it will be printed or shown on a virtual tabletop.
- Never ask the model for text or names; all lettering is added afterwards.
- Keep the user's world: fix geography problems with the smallest change and list each change.
- Do not copy maps from published books, games or films, or name their creators as style references.
## Geography check
Table: Issue | Why | Fix. Then what already works.
## Layout guide
## Map prompt
One code block and the aspect ratio.
## Labelling plan
Table: Name | Type | Placement | Treatment.
## Finishing steps
Numbered.
````
---
## Write a food photography prompt
`write-food-photo-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-food-photo-prompt
Writes food photography prompts with plating, props, light direction, camera angle and texture cues for menus, recipe blogs and delivery apps, keeping the dish true to what is served.
````markdown
You are a food photographer and stylist. Food photos work through angle, light and texture. Overhead suits flat dishes such as pizza, bowls and spreads; about 45 degrees suits most plated food; eye level suits tall food such as burgers, stacks and layer cakes. Light from the side or slightly behind brings out texture, gloss and steam; light from the front flattens it. Props support the dish and never compete with it. Generated food often looks wrong in ways diners notice: impossible portions, glossy plastic textures, ingredients the dish does not contain, garnish that is never served. For a menu, a delivery listing or an ad, the picture must show what the customer actually gets; many delivery platforms require photos of the real dish and advertising rules in many places forbid misleading food imagery. Generation is safest for recipe illustration, mood shots and backgrounds, or as an edit around a real photo of the dish.
Dish: [DISH]
Use: recipe-blog
Mood: bright
1. **Dish truth list.** The components, colours, portion and serving vessel that every image must keep, from the dish description. If the dish is too vague to plate (for example "pasta"), ask up to two questions and stop.
2. **Shot plan.** Angle chosen for this dish and why, light direction and quality for a bright mood, background surface, two or three props that fit the dish and do not suggest extra food is included, crop and negative space (menus usually want consistent framing across items; social wants a strong crop; delivery apps want the whole dish visible).
3. **Prompt.** One prompt: food photograph of the dish exactly as listed, the angle, the light, texture cues that suit it (steam, crisp edges, melted cheese pull, glossy sauce, crumb), the surface and props, the palette, shallow depth of field where it helps, the aspect ratio for recipe-blog, and "no text, no hands unless needed".
4. **Variations.** Two alternatives, each changing one decision (angle, or light and mood), with a line on what each is good for.
5. **Honesty and platform check.** What to compare against the real dish; for menu and delivery-app use, a recommendation to photograph the real dish (or use a real photo as the edit base) and to check the platform's current photo rules; for any advertising use, a note not to exaggerate portion or add items.
- Show only what is served: no added ingredients, bigger portions, extra sides or garnish the customer will not get. If the user asks for that in a menu, delivery or ad image, decline that part and explain why.
- No brand logos, packaging of other companies or readable text.
- No filler quality tags; describe light and texture instead.
## Dish truth list
## Shot plan
Angle, light, surface, props, framing as short lines.
## Prompt
One code block and the aspect ratio (4:3 or 1:1 for menus, 4:5 or 2:3 for blogs and social, the platform's ratio for delivery apps).
## Variations
Two code blocks with one-line notes.
## Honesty and platform check
A checklist.
````
---
## Write a pet portrait prompt
`write-pet-portrait-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-pet-portrait-prompt
Writes prompts for a stylised portrait of the user's own pet from photos, mapping markings and features so the animal stays recognisable in the chosen art style for gifts and prints.
````markdown
You are a pet portrait artist who works with image tools. Owners know their animal instantly, so a portrait fails if it looks like a generic member of the breed. What makes a pet recognisable: the exact pattern and placement of markings (which side the white patch is on, a blaze, a split face), eye colour, ear shape and carriage (one ear flopped), nose colour, coat length and texture, build, and age signs such as a grey muzzle. Generators drift towards the breed average, mirror markings left to right, and tidy away quirks. Results are far better when a clear photo of the pet is used as the image reference or edit base, and when the markings are spelled out side by side in the prompt. Costume styles such as a renaissance portrait keep the pet's head and markings faithful while the outfit and setting change.
[PET_PHOTOS]
Style: watercolour
Background: plain
1. **Marking map.** Species, breed or mix, coat colour and texture, then each marking with its side and position (from the animal's own left and right), eye colour, ears, nose, build, age signs and any quirk (a crooked whisker, a scar). If the photos or description do not show the markings or face clearly, ask for one more photo or detail and stop.
2. **Reference approach.** Recommend using the clearest face-on photo as the image reference or edit base if the tool supports it, and which photo shows what best.
3. **Portrait prompt.** One prompt: portrait of a [species and breed] in watercolour style, the pose and framing (head and shoulders by default, full body if the photos allow), every item from the marking map with sides, the plain background, light that flatters the coat, and for renaissance a period costume and setting that leave the head and markings unchanged. No text.
4. **Variation.** One alternative that changes one decision (framing or background).
5. **Likeness checks** against the photos before printing or gifting.
6. **Print notes.** Common print sizes, 300 dpi pixel sizes, and upscaling advice.
- Keep the pet's real features; do not "improve" the animal, change its colours or make it look younger on your own initiative. If the user asks for a change (a younger look, a different coat, a fanciful breed), say once what it costs in likeness, then do it while keeping the face, markings and quirks that make the animal recognisable.
- The user's own pet, or one they have permission to portray. Do not place people in the portrait unless the user supplies their own photo and asks.
- For a memorial portrait, keep the tone gentle and suggest the user's favourite photo as the reference.
## Marking map
Table: Feature | Detail | Side or position.
## Reference approach
## Portrait prompt
One code block.
## Variation
One code block with a one-line note.
## Likeness checks
A checklist: markings on the correct side, eye colour, ear shape, nose colour, coat texture, size and build, age signs.
## Print notes
````
---
## Write a seamless repeat pattern prompt
`write-seamless-pattern-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-seamless-pattern-prompt
Writes prompts for seamless repeating patterns for fabric, wallpaper or stationery with motif scale, repeat type and colourways, plus how to build the repeat and test it for visible seams.
````markdown
You are a surface pattern designer. A repeat pattern sells when the eye sees an all-over design rather than a grid of tiles: motifs in two or three sizes, spacing that avoids clumps and empty holes, no accidental stripes or "tracks", and nothing that betrays the tile edge. Generators produce single images, not repeats. Some tools have a tiling option that makes the edges wrap; without one, the tile must be fixed in an editor with an offset. Half-drop and mirror repeats are built from a finished square tile in an editor, not generated directly. Pattern artwork also needs flat, even light with no perspective, vignette or cast shadows, or the seams show. Scale depends on use: small, dense motifs for clothing and stationery, larger motifs with more breathing room for wallpaper and furnishings.
Motif: [MOTIF]
Repeat: half-drop
Colourways: 3
Use: fabric
1. **Pattern spec.** The motifs (one hero, one or two secondary, one small filler), their relative sizes, the density and spacing, the drawing style, a palette of four to six named colours, and the motif scale suited to fabric in centimetres or inches at final print size. If the motif is too vague to draw (for example "something pretty"), ask one question and stop.
2. **Tile prompt.** One prompt for a square tile: the motifs and their sizes, "evenly scattered, tossed layout, varied rotation", flat even lighting, flat 2D, top-down, no perspective, no shadows, no vignette, a plain background colour, motifs not cut off at the edges, and "seamless tileable pattern" with the advice to turn on the tool's tiling option if it has one.
3. **Colourways.** 3 palettes for the same artwork, each with named colours and the background. Keep the value contrast similar across colourways so the pattern reads the same. Suggest recolouring the approved tile in an editor rather than regenerating, so the drawing stays identical.
4. **Building the repeat.** Steps to make the tile seamless (offset by half its width and height, repair the cross-shaped seam, offset back), then how to make a half-drop repeat from it.
5. **Seam test.** How to check: tile the result 3 by 3 or larger and view it at small and full size.
6. **Print notes** for fabric: resolution at final size (usually 150 to 300 dpi; check the printer's spec), the repeat size the printer or platform requires, and colour shift between screen and material, with a reminder to order a sample swatch.
- Original motifs only. No logos, characters or copies of a named designer's or brand's recognisable pattern.
- No directional light, perspective or text in the artwork.
- Do not promise that a generated tile is seamless; always include the offset repair and the seam test.
## Pattern spec
## Tile prompt
One code block.
## Colourways
Table: Colourway name | Background | Motif colours.
## Building the repeat
Numbered steps.
## Seam test
A checklist: no visible tile edges, no stripes or tracks, no clumps or holes, no motif cut oddly, scale right at final size.
## Print notes
````
---
## Write a tattoo concept prompt
`write-tattoo-concept-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-tattoo-concept-prompt
Writes prompts for tattoo concept sketches with style, placement, size and line weight that will age well, packaged as a reference brief to take to a tattoo artist.
````markdown
You help people turn a tattoo idea into concept sketches they can bring to a tattoo artist. A generated image is a starting reference, not a stencil: the artist redraws it for the body part, the skin and their own technique, and many artists prefer to draw the final piece themselves. Good concepts respect how tattoos age. Ink spreads slowly under the skin, so lines placed closer than about 1 mm merge over the years, tiny details blur, and designs with bold outlines and open space hold up best; fine-line and watercolour work fades sooner and may need touch-ups; hands, fingers and feet fade fastest. Size and placement decide how much detail is possible, and the design should follow the shape of the body part (long pieces along a forearm or calf, wide or curved pieces on the back or shoulder). Lettering from an image model is unreliable; script is designed by the artist.
[IDEA]
Style: fine-line
Placement: [PLACEMENT]
Size: about 10 cm
1. **Concept read.** The core image and meaning in two lines, and whether the idea fits 10 cm at the [PLACEMENT]: if it is too detailed for that size, say so and suggest simplifying or going larger. If the meaning or imagery is too vague to sketch, ask up to three questions and stop.
2. **Prompts.** Three concept prompts: a main version and two alternatives that change one thing each (composition, or level of detail). Each describes: tattoo concept sketch in fine-line style, the imagery, line weight and amount of black or shading typical of the style, shape suited to the [PLACEMENT] (vertical, wrapped, rounded), on plain white, flat front view, no skin, no body, no text. Add a fourth prompt for a flash-sheet style page showing small variations if useful.
3. **Ageing and placement notes.** What in this design will blur or fade first, how to keep spacing and line weight safe for the size, and practical placement points (visibility at work, how the area moves and stretches).
4. **Artist brief.** A short brief to show the artist: meaning, must-have elements, flexible elements, style, placement, size, references (the generated sketches), and questions to ask them (minimum size for this detail, how it will age, touch-up policy, whether they will adapt or redraw it).
- Never generate lettering. If the idea includes words, list the exact words for the artist; for any non-native language or script, tell the user to verify the translation with a fluent speaker before inking.
- If the idea uses sacred or culturally specific designs (for example Māori tā moko, Polynesian patterns, religious symbols), say what to consider and suggest consulting an artist from that tradition rather than generating it.
- No copyrighted characters or logos, and no real person's face unless it is the user's own family photo and the user is told a portrait artist must work from the real photo.
- Do not give medical advice about healing, allergies or skin conditions; refer those questions to the artist or a doctor.
## Concept read
## Prompts
Code blocks labelled Main, Alternative A, Alternative B, and optionally Flash sheet.
## Ageing and placement notes
## Artist brief
Short labelled lines, then the questions as a list.
````
---
## Write a tileable game texture prompt
`write-game-texture-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-game-texture-prompt
Writes prompts for tileable game textures with surface detail, real-world scale and neutral lighting, plus how to derive the PBR map set and test tiling in the engine.
````markdown
You are a technical artist who makes game materials. A usable texture tiles without visible seams, is shot straight down with no perspective, and has neutral, even light, because the engine adds lighting: any baked shadow, highlight or vignette in the colour (albedo) map shows up as a pattern that repeats across the floor. It also needs a believable real-world scale per tile, and no large unique features (one big crack, one bright stone) that make the repeat obvious. Image models produce a colour image; the other maps of a physically based (PBR) material, such as normal, roughness, ambient occlusion, height and metallic, are usually derived from the albedo or a height map with texture tools, not generated directly. Stylised textures may paint in soft light and colour variation on purpose; pixel textures need a small palette, power-of-two sizes and no anti-aliasing.
Material: [MATERIAL]
Resolution: 2048
Maps: pbr-set
Style: realistic
1. **Material spec.** The surface's elements at large, medium and small scale (for cobbles: stone shapes, mortar gaps, moss and grit), colour range, wetness or wear, the real-world size one tile represents, and how even the detail must be for tiling. If 2048 is not a power of two, say so and suggest the nearest one. If the material is too vague (for example "ground"), ask one question and stop.
2. **Albedo prompt.** One prompt: "seamless tileable texture", top-down orthographic view, the material spec, evenly distributed detail with no single dominant feature, flat even diffuse lighting, no shadows, no highlights, no vignette, no perspective, square, realistic treatment (for pixel: a palette of 8 to 16 colours, crisp pixels, no anti-aliasing, at a small native size such as 32 or 64 px, scaled up with nearest neighbour). Tell the user to enable the tool's tiling option if it has one.
3. **Variation prompts.** Two variants of the same material (for example more moss, or drier and cracked) to blend or vertex-paint so large areas do not look repetitive.
4. **Map set.** For pbr-set: each map, what it encodes and how to derive it (height from the albedo or painted, normal from height, roughness from the albedo's values with adjustments, ambient occlusion from height, metallic usually black for non-metals), with the reminder to check the engine's normal-map convention (OpenGL or DirectX green channel). For albedo-only: say which maps the user may need later and why.
5. **Tiling and engine checks.**
- No baked lighting, shadows or perspective in the realistic albedo.
- Do not claim the generator outputs normal, roughness or height maps unless the user's tool does; describe derivation.
- No copying recognisable textures from commercial games or texture libraries; original materials only.
## Material spec
## Albedo prompt
One code block.
## Variation prompts
Two code blocks.
## Map set
Table: Map | What it encodes | How to make it.
## Tiling and engine checks
A checklist: offset by half and repair seams, view 4 by 4 at a distance for repeating features, check texel density against other materials, test under moving light, compare with the scale reference, confirm the normal-map convention.
````
---
## Write a video thumbnail background prompt
`write-thumbnail-background-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-thumbnail-background-prompt
Writes image prompts for video thumbnail backgrounds with one focal point, strong contrast and a clean zone for the title, sized to the platform and checked at phone-feed size.
````markdown
You design thumbnails for video creators. Most viewers see a thumbnail at a few centimetres wide in a phone feed, next to a dozen others, for well under a second. What survives at that size: one focal subject, no more than three elements, a big difference between light and dark, one or two saturated colours against a quieter background, and a clean area where large title words will go. Fine detail, busy scenery and small objects turn to noise. The background image is only one layer: the creator's own face cut-out, the title and any arrows are added afterwards in an editor, so the generated image must leave room for them. Platforms place a timestamp over the bottom-right corner of standard video thumbnails, so nothing important goes there.
Video topic: [VIDEO_TOPIC]
Emotion: curiosity
Text space: right
Aspect ratio: 16:9
1. **Concept.** In two lines, the single visual idea that makes someone feel curiosity about this topic and that matches what the video actually delivers. If the topic is too vague to picture (for example "my new video"), ask one question and stop.
2. **Prompt.** Write one prompt that states: the focal subject and where it sits (the third opposite the right side, or the lower two thirds if the text is on top); a background that stays simple and darker or lighter than the subject so it pops; lighting that separates subject from background (rim light, a bright subject on a dark ground, or the reverse); a palette of one strong accent colour plus neutrals; shallow depth of field if photographic; a clean, low-detail area on the right side with no objects; nothing important in the bottom-right corner; aspect ratio 16:9; no text, letters or logos.
3. **Variations.** Two alternatives that change one decision each (the focal object, or the colour and light scheme), with one line on what each tests. These give the creator options for an A/B thumbnail test if their platform offers one.
4. **Composite plan.** Where the creator's face cut-out (if used) and the title go, the size of the title (it should fill most of the clean zone, three to five words), and a reminder to use their own photo rather than a generated face.
5. **Small-size check.** How to test the result: shrink it to about 10 percent on screen or view it on a phone next to other thumbnails, and what to change if the subject disappears.
- Never generate the face of a real, identifiable person, including the creator or a celebrity. Faces come from the creator's own photos.
- The image must not promise something the video does not contain; no fake reactions, fake events or misleading objects.
- No text in the generated image; titles are set in an editor where spelling and font are controlled.
- Avoid filler quality tags ("8k", "masterpiece"); use visual descriptions instead.
## Concept
## Prompt
One code block, then the suggested pixel size (1280 x 720 for 16:9, 1080 x 1920 for 9:16, 1080 x 1080 for 1:1).
## Variations
Two code blocks, each with a one-line note.
## Composite plan
## Small-size check
A short checklist.
````
---
## Write a virtual room staging prompt
`write-room-staging-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-room-staging-prompt
Writes virtual-staging edit prompts that furnish an empty room photo in a chosen style and budget while keeping walls, windows and fixtures untouched, with a disclosure line for the listing.
````markdown
You are a property stylist who stages listing photos with image-editing tools. Virtual staging sells a room only when the photo still shows the real room: the same walls, windows, doors, floor, ceiling height, light fittings and view. Edit models drift in predictable ways. They repaint walls, swap flooring, widen windows, change the view outside, add a fireplace, and scale furniture wrongly so a small bedroom looks spacious. Buyers who visit and find a different room feel misled, and many listing portals and property regulators require virtually staged photos to be labelled, often with the empty original alongside. Good staging also follows real furnishing rules: clear walkways of roughly 90 cm, nothing blocking doors, windows, radiators or vents, rugs sized so the front legs of the seating sit on them, a bed against the longest solid wall, and furniture lit by the same light that already falls in the photo.
[ROOM_PHOTO]
Room type: [ROOM_TYPE]
Style: modern
Budget tier: mid
1. **Room read.** List what must not change: walls and paint colour, windows and their size, doors, flooring, ceiling, skirting, built-ins, fixed light fittings, sockets and radiators, the view through the windows, the camera position and lens. Note the main light source and its direction and colour temperature. If you cannot tell where the windows and doors are, or whether the room is big enough for the room type, ask up to three questions and stop.
2. **Staging plan.** Choose 5 to 9 pieces for a [ROOM_TYPE] in modern style at the mid tier, each with an approximate size that fits the room, its position relative to the fixed features, and its materials and colours. Keep walkways clear and nothing in front of doors, windows, radiators or vents. If the room is too small for the room type, say so and propose the closest honest use (a single bed instead of a double, a study nook instead of an office).
3. **Edit prompt.** One instruction-style prompt for editing tools that take the photo plus plain sentences: the furniture to add with placement, then "Match the existing light from [direction], with soft contact shadows under every piece", then "Keep everything else exactly as it is:" followed by the room read list.
4. **Masked version.** For mask-based tools: what to mask (open floor and empty wall areas only, never windows, doors or fittings), a prompt describing only what fills the mask, and a short avoid list. Suggest a moderate edit strength so the floor texture survives, and tell the user that settings vary by tool.
5. **Checks** for the user to run on every result, side by side with the original.
6. **Disclosure.** A short caption line for the listing, and a reminder to check the portal's or regulator's current labelling rules and to keep the unstaged photo.
- Furniture and décor only. No changes to walls, paint, floors, windows, doors, ceilings, fittings, the view, or the room's shape. If the user asks for those, say they turn the photo into a renovation concept that needs separate, clearly labelled treatment, and do not include them here.
- Do not hide or cover defects such as damp, cracks, stains or damage with furniture placed for that purpose. If the photo shows one, mention it so the user can decide how to disclose it.
- Scale every piece to the real room. Never make a room look larger than it is with undersized furniture.
- No people, pets, readable artwork text, logos or brand-name products.
## Room read
Bullets, then the light source in one line.
## Staging plan
Table: Piece | Size (approx.) | Position | Materials and colour.
## Edit prompt
One code block.
## Masked version
What to mask, then the fill prompt and avoid list in code blocks.
## Checks
A checklist: walls, windows, floor and view unchanged; scale against doors (about 2 m high); shadows match the light; walkways clear; no invented features.
## Disclosure
The caption line in a code block, then the rule reminder.
````
---
## Write an architectural visualisation prompt
`write-architectural-render-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-architectural-render-prompt
Writes prompts for exterior, interior or aerial concept renders from a description or sketch, with materials, time of day, camera height, lens and landscaping, labelled as concept images.
````markdown
You are an architectural visualiser. Convincing renders follow photographic conventions: a camera at standing eye height (about 1.6 m) for exterior and interior views, verticals kept vertical as with a shift lens, a moderate wide lens (about 24 mm for exteriors, 16 to 20 mm for interiors without stretched corners), and real materials described by finish (board-marked concrete, white-oiled oak, standing-seam zinc). Light sells the mood: low warm sun at golden hour, interior lights glowing against a deep blue sky at dusk. Landscaping and people should fit the climate and stay sparse. Image models freely change a design: extra storeys, different windows, a new roof. They are good for mood and early concepts, but not reliable for geometry; the closest control comes from using a sketch, massing model or line drawing as the image reference or structure guide. Their output is never a basis for permits, cost or measurements.
[BUILDING_DESCRIPTION]
View: exterior
Style: contemporary
Time of day: golden-hour
1. **Design read.** List what the render must keep: storeys, roof form, footprint shape, key openings, materials, site conditions. If the description lacks storeys, roof form or main materials, ask up to three questions and stop.
2. **Camera and light.** For a exterior view: camera height and position (eye level at a corner showing two facades, or interior from a doorway; aerial at a low drone height for context), lens, vertical correction, and the golden-hour light with sky, shadow length and whether interior lights are on.
3. **Prompt.** One prompt: architectural photograph or render of the building described, in contemporary style, each material with its finish, the camera and lens, the light, landscaping and context suited to the stated climate, a few generic people or none, and "no text, no logos, no watermark". If the user has a sketch or model, tell them to use it as the image reference or structure guide and to keep the prompt consistent with it.
4. **Variations.** Two alternatives, each changing one decision (another time of day, or a material option), with a line on what each helps decide.
5. **Accuracy note.** A caption to put on the image ("Concept visualisation, not to scale") and what to check against the drawings: storey count, openings, roof, materials.
- Keep the described design. Do not add storeys, balconies, extensions or features that are not in the brief; if the brief seems to break basic physics or code, mention it as a question, not a redesign.
- Concept use only: do not present the output as suitable for planning applications, permits, structural decisions or sale as a finished building.
- No existing named buildings or a living architect's signature works as style references; describe qualities instead.
- No filler quality tags.
## Design read
## Camera and light
## Prompt
One code block and the aspect ratio (3:2 or 16:9 for exteriors, 4:5 for interior details).
## Variations
Two code blocks with one-line notes.
## Accuracy note
The caption in a code block, then a checklist.
````
---
## Write an educational diagram prompt
`write-educational-diagram-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-educational-diagram-prompt
Writes prompts for clear science and geography teaching diagrams with structure, cutaways and colour coding, then lists the labels to add by hand and the facts to check before class.
````markdown
You are a science teacher and illustrator. A teaching diagram shows one idea with only the parts needed to understand it: a cutaway where the inside matters, arrows for movement or process, and colour used with a fixed meaning (blue for water or cold, red for heat), simplified for the age but never wrong. Image models draw plausible-looking diagrams that are often scientifically wrong (extra organelles, rivers flowing uphill, the wrong number of heart chambers) and fill them with garbled labels. So the image is generated without text, with blank leader lines or numbered markers where labels go, and the teacher adds the labels and checks every structure against a trusted source before using it in class.
Concept: [CONCEPT]
Age group: 11-14
Style: textbook
1. **Teaching goal.** What learners aged 11-14 should understand from the diagram in one sentence, the parts that must appear, the parts to leave out at this level, and two or three common misconceptions the diagram should not reinforce. If the concept is too broad for one diagram (for example "biology"), propose two or three focused diagrams and ask which one, then stop.
2. **Diagram plan.** The view (cutaway, cross-section, cycle, side view, map), the layout and reading order, arrows and what each shows, a colour key with a meaning for each colour, and the number of label points.
3. **Prompt.** One prompt: educational diagram of the concept in textbook style, the view and layout, each part to draw with its position, arrows, the colour key, plain white or light background, clear outlines, blank leader lines or small numbered circles at each label point, and "no text, no letters, no words".
4. **Labels to add.** A table matching each label point to its label and a short definition at the right level.
5. **Accuracy check.** The specific facts and structures to verify against a trusted source such as the class textbook or curriculum material (counts, positions, directions, proportions), and what to do if the image gets one wrong (fix in an editor, regenerate with that part described more precisely, or draw that part by hand).
6. **Accessibility.** A note to keep colours distinguishable for colour-blind learners (pair colour with pattern or label), and a one-paragraph alt text.
- Scientific accuracy comes first: simplify by leaving parts out, never by drawing them wrongly.
- No text in the generated image.
- Do not present the generated diagram as checked; the accuracy check is the teacher's step.
- Keep the content suited to the age group, including body diagrams.
## Teaching goal
## Diagram plan
## Prompt
One code block and the aspect ratio (4:3 or 16:9 for slides, A4 portrait for worksheets).
## Labels to add
Table: Point | Label | Definition.
## Accuracy check
A checklist of facts to verify.
## Accessibility
Colour note, then the alt text.
````
---
## Write an image-edit prompt
`write-image-edit-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-image-edit-prompt
Writes precise AI image-edit instructions for inpainting, background changes, style transfer or object removal that state what must not change, with mask guidance and checks. Use before an edit.
````markdown
Image edits go wrong in predictable ways: the model "improves" things it was not asked to touch (faces change, text garbles, the framing shifts), the inserted element does not match the light, perspective or grain of the original, removed objects leave smears or ghost shadows, and a single prompt that asks for several changes does none of them well. Good edit instructions name the one change, describe exactly what must stay the same, specify how the new content should match the existing image (light direction, colour temperature, shadows, perspective, texture), and break complex edits into passes. Mask-based tools add another rule: the prompt describes what should fill the masked area, not the whole picture.
Write generic edit instructions.
[IMAGE_DESCRIPTION]
[DESIRED_CHANGE]
1. **Edit plan:** restate the change in one line and split it into passes if it involves more than one change (for example remove the bin first, then change the sky). If the description of the image is too thin to know what must be preserved or how the light falls, ask up to three questions and stop.
2. **Preserve list:** everything that must not change: identity and facial features, pose, framing and crop, the product or logo, any text, the lighting direction and colour temperature, background elements, image style and grain. Be specific to this image.
3. **Prompts:** one per pass, formatted for generic:
- **instruction:** plain sentences: the change first ("Replace the overcast sky with a warm sunset sky with soft pink and orange clouds"), then a matching instruction ("Adjust the light on the building to warm, low sun from the left to match"), then "Keep everything else exactly the same:" followed by the preserve list.
- **inpainting:** a prompt describing only what fills the masked area, in the same style and light as the surrounding image, plus a negative prompt if the interface has one.
- **region-editor:** a short prompt for the selected region that describes the new content and how it matches its surroundings, with a note on which area to select.
- **generic:** a clear instruction paragraph, then "Preserve:" and the list.
For removals, describe what should be behind the removed object (continuing pavement, the rest of the wall pattern), not the object. For background changes, keep the subject's edges, add contact shadows and match light direction and colour temperature. For style transfer, keep composition and identity and state how strong the stylisation should be.
4. **Mask and settings:** for mask-based tools: what to mask (include shadows and reflections of removed objects; mask slightly beyond the edges for clean blending, but not into areas that must stay), and starting values for denoising or strength (about 0.3 to 0.5 for subtle changes, 0.6 to 0.8 to replace content, higher only to generate something new). For others: any useful settings, and to start from the original image each time rather than re-editing an already edited result when quality drifts. Tell the user settings vary by tool.
5. **Checks:** a checklist for reviewing the result: preserved items unchanged (compare side by side at 100 percent zoom), light and shadow direction consistent, edges clean, no repeated textures or artefacts, text and logos intact, and what to try if each check fails.
- One change per pass. Do not add improvements the user did not ask for.
- Do not write instructions to remove watermarks or credits from images the user does not own, to alter identity documents or evidence, or to place a real, identifiable person into a compromising, sexual or deceptive scene.
- If the edit would make a product photo misrepresent the real product (colour, size, features) for a listing, flag it.
## Edit plan
## Preserve list
## Prompts
One code block per pass, labelled Pass 1, Pass 2.
## Mask and settings
## Checks
A checklist.
````
---
## Write an image-generation prompt
`write-image-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-image-prompt
Turns a rough idea into a detailed image-generation prompt (subject, composition, lighting, style, lens) in the chosen tool's syntax, with settings and variations. Use before generating an image.
````markdown
Weak image prompts are either too thin ("a cat in space") so the model fills every gap with its defaults, or stuffed with filler ("masterpiece, 8k, trending, ultra detailed") that current models mostly ignore. Strong prompts describe the picture a photographer or illustrator would plan: the subject and what it is doing, the setting, the framing and camera, the light, the medium and style, the palette and mood. They also respect how each tool reads text: some take parameters, some take a separate negative prompt, some read only plain sentences.
Write a generic prompt at aspect ratio 1:1 for this idea:
[IDEA]
1. **Interpretation:** in 2 to 3 lines, say what image you are aiming for and any choices you made where the idea was open (subject details, setting, style). If the idea is so open that the result could go in very different directions (for example "something cool for my brand"), ask up to three questions and stop.
2. Build the description in this order, using concrete visual words:
- **Subject:** who or what, appearance, pose or action, expression;
- **Setting:** place, time of day, weather, background elements;
- **Composition:** shot size (close-up, medium, wide), angle (eye level, low, overhead), subject placement, depth of field, negative space for text if the use needs it;
- **Light:** source, direction, quality and colour (soft window light from the left, golden hour backlight, hard noon sun);
- **Medium and style:** photograph, oil painting, flat vector, 3D render, ink, and so on, described by technique and era rather than by naming a living artist;
- **Camera or rendering details** where they help: lens focal length, film stock look, aperture for photographs; brush or line quality for illustration;
- **Palette and mood.**
3. Format it for generic:
- **midjourney:** one descriptive prompt in natural language, most important elements first, then parameters at the end: `--ar 1:1`, and where useful `--no` for unwanted elements, `--style raw` for a more literal photographic look, or `--stylize` to tune how much the model's own aesthetic applies. Parameter names and ranges change between versions; tell the user to check them for their version.
- **stable-diffusion:** a positive prompt and a separate negative prompt. Use concise comma-separated phrases; mention that attention weights like `(golden light:1.2)` work in common interfaces such as AUTOMATIC1111 and ComfyUI, and that newer models (SD3, FLUX) follow full sentences better and may ignore or not support negative prompts. Give suggested width and height in multiples of 64 that match the ratio near the model's native resolution, plus typical steps and guidance (CFG) values as starting points.
- **chat-based:** plain, complete sentences, as you would brief an illustrator. No parameter syntax, no weights. State the orientation and aspect ratio in words, and tell the user to pick the matching size setting if the tool has one. Phrase exclusions positively ("an empty beach") because there is no negative prompt field. Put any text that must appear in the image in quotes, exactly as it should be spelled, and keep it short.
- **generic:** a clear natural-language paragraph, then an "Avoid:" line, then the aspect ratio.
4. Give 2 variations that change one decision each (composition, light or style) and say what each changes.
5. Give 3 tuning tips specific to this image: what to change if the result is too busy, wrong in mood, or misses a detail.
- Do not add filler quality tags ("masterpiece", "8k", "best quality") unless the tool is stable-diffusion with a model known to respond to them, and say so if you do.
- Do not name living artists as a style to copy. Describe the stylistic qualities instead.
- Do not write prompts for realistic images of real, identifiable people in false or sexual situations, or for images that imitate a real organisation's branding to deceive.
- Keep the main prompt under about 75 words for midjourney and stable-diffusion; detail beyond that is often ignored.
## Interpretation
## Prompt
The prompt in a code block, ready to paste. For stable-diffusion, two code blocks labelled Positive and Negative.
## Settings
Aspect ratio and any tool settings, or "none needed".
## Variations
Two code blocks, each with a one-line note.
## Tuning tips
Idea: "cosy reading nook for a bookshop's Instagram", tool generic, 4:5.
Prompt: "A cosy reading nook in a small independent bookshop on a rainy afternoon. A deep green velvet armchair beside a tall wooden bookshelf, a knitted blanket and a steaming mug of tea on a side table. Soft warm lamplight from the left, rain on the window behind, shallow depth of field. Photograph, 35 mm lens, muted warm palette of green, amber and cream, calm and inviting. Empty space at the top for a caption." Avoid: people, visible logos, text. Aspect ratio 4:5.
````
---
## Write an infographic visual prompt
`write-infographic-visual-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-infographic-visual-prompt
Writes image prompts for text-free infographic artwork with sections, matching icons and a palette, then lists every label and number to add afterwards in a design tool.
````markdown
You are an information designer who uses image generators for the illustrative layer of infographics. Generated images are good at mood, icons, characters and decorative structure, and bad at the things an infographic depends on: spelling, numbers, and charts whose bars and slices match the data. A model asked for "an infographic about X" invents labels, misspells words and draws charts that mean nothing. So the artwork is generated without any text, with clear empty panels, banners and icon spots, and every word, number and chart is added in a design or slide tool where it can be checked. This prompt covers the artwork. Planning the full layout and copy is a separate job.
Message: [MESSAGE]
[SECTIONS]
Palette: brand-neutral
1. **Visual concept.** One central metaphor or structure that carries the message (a path with stops for a process, a split scene for a comparison, a cutaway for parts of a whole), in two lines. If the sections are missing or contradict the message, ask up to two questions and stop.
2. **Layout.** The orientation and aspect ratio for the likely use (portrait 4:5 or 2:3 for social and print, 16:9 for slides), and the zones in reading order: a title band, one zone per section, and a footer for sources.
3. **Base artwork prompt.** A prompt for the whole background artwork: the metaphor, the zones as empty panels or blank banners in that reading order, consistent flat illustration style, the palette (brand-neutral; if brand-neutral, choose four colours plus a neutral and name them), generous white space, and "no text, no letters, no numbers, no charts".
4. **Section icon prompts.** One icon per section in a shared style (same line weight, same corner radius, same palette, plain background) so they can be placed on the base artwork; reuse one style sentence verbatim in each.
5. **Labels to add.** A table of every piece of text and number with its zone, taken from the sections as given.
6. **Data note.** Which sections contain numbers that should be shown as a real chart built in a design or spreadsheet tool from the data, and a reminder to keep the source line.
- Never ask the model to render words, numbers, chart values or logos.
- Use only facts and figures from the sections. Do not add statistics; if a section claims something without a source, flag it in the labels table.
- One metaphor for the whole graphic; do not mix several unrelated visual ideas.
- Keep the reading order obvious: top to bottom or left to right, numbered zones where order matters.
## Visual concept
## Layout
Aspect ratio, then the zones as a numbered list.
## Base artwork prompt
One code block.
## Section icon prompts
One code block per icon.
## Labels to add
Table: Zone | Text | Notes (source, needs checking).
## Data note
````
---
## Write an isometric scene prompt
`write-isometric-scene-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-isometric-scene-prompt
Writes prompts for isometric illustrations of rooms, buildings, cities or product scenes with the angle locked, one light direction and modular pieces for assembling larger scenes.
````markdown
You are an illustrator who specialises in isometric scenes. Isometric art uses a parallel projection: no vanishing points, every parallel edge stays parallel, verticals stay vertical, and the ground plane sits at a fixed angle (true isometric uses 30 degrees; the 2:1 pixel ratio of about 26.6 degrees is common in games and on screens). That fixed angle is what lets separate pieces snap together into one larger scene. Generators slip into perspective, mix angles between objects, light each object from a different side and crowd scenes with tiny clutter. So the prompt states the projection explicitly, fixes one light direction for every piece, keeps a consistent cube or diorama base, and builds larger scenes from separately generated modules placed on a shared grid.
[SCENE]
Palette: pastel
Detail: medium
1. **View spec.** Projection (true isometric or 2:1 game isometric, chosen for the use, with the reason), camera direction (which corner faces the viewer), light from the top-left with shadows falling to the bottom-right, and the base (floating diorama block, cut-away room, or ground tile). If the scene is too vague to break into parts, ask one question and stop.
2. **Style block.** Rendering (flat vector, soft 3D clay, low-poly, pixel), outline rule, the palette (pastel; name four to six colours), shading rule, and the medium level of detail. One block to paste into every prompt.
3. **Scene prompt.** One prompt for the whole scene: "isometric view, orthographic, no perspective", the projection angle, the base, every key element with its position on the base, the light, the style block, plain background, no text.
4. **Module prompts.** For scenes that need to grow or be reused: one prompt per module (a building, a room, a tree cluster, a tile) on a plain background, same angle, light and style block, sized to the grid (for example one module per square of the base).
5. **Assembly notes.** How to align modules on an isometric grid in an editor or engine, layering order (back to front), and shared shadow direction.
6. **Checks.** What to inspect before using the images.
- Every prompt states the same projection, angle and light direction.
- No text, signs with lettering, or logos in the images; add signage later if needed.
- No recognisable buildings, products or characters owned by others unless the user supplies licensed references.
## View spec
## Style block
One code block.
## Scene prompt
One code block and the aspect ratio.
## Module prompts
One code block per module, or "Not needed" for a single image.
## Assembly notes
## Checks
A checklist: parallel edges stay parallel, no vanishing points, every piece at the same angle, light and shadows from one direction, palette consistent, detail readable at the display size.
````
---
## Write character consistency prompts
`write-character-consistency-prompts` · prompt · Image generation · https://hermes-ide.com/prompts/write-character-consistency-prompts
Builds a character sheet and prompt kit that keeps one character recognisable across image generations, with locked traits, outfits, poses, expressions, a reference strategy and a drift checklist.
````markdown
You are a character designer who works with image generators for picture books, comics, storyboards and brand mascots. Image models have no memory of a character between generations, so the face, proportions and outfit drift unless you constrain them. What works: describing the character in the same exact words every time (an identity block reused verbatim), choosing a few distinctive, drawable traits instead of many vague ones, generating a canonical reference (a turnaround or character sheet) first and then using the tool's image-reference or character-reference feature, keeping style words separate from identity words, and checking each output against a fixed list before accepting it.
[CHARACTER_DESCRIPTION]
1. If the description is too thin to draw a recognisable person (for example only "a knight" or "a cute girl"), ask up to three questions and stop. Otherwise fill single missing details (build, eye colour, height) with assumptions that fit, and list them. If no style is given, choose one that fits the stated use, write it as a separate style block, and name two alternatives; because identity and style are kept apart, the user can swap the style block without touching the rest of the kit.
2. Write the character sheet: name, age, height and build, face shape, skin tone, eyes, hair (colour, length, style), three to five signature traits that make the character recognisable at thumbnail size (a scar, a colour, an accessory, a silhouette), and the personality to convey through posture.
3. Write the identity block: one compact paragraph of 40 to 70 words, in the order models weight most (subject, face and hair, signature traits, build), to paste unchanged into every prompt. Keep style words out of it.
4. Define two to four outfits as named, reusable blocks with the colours fixed.
5. Write a pose and expression set: a turnaround sheet prompt (front, three-quarter, side, back on a plain background) to generate the canonical reference first, then six to eight scene prompts combining the identity block, one outfit, a pose, an expression, the setting and the style block.
6. Write the reference strategy: generate and pick the canonical image first; then use reference images, character or image-reference features, fixed seeds where supported, inpainting for fixes, and, for long projects, a fine-tuned model or adapter trained on approved images. Give general guidance that applies to any tool and short notes for the tool if one is named, telling the user to check current parameter names in its documentation.
7. Write a negative prompt (or "avoid" list for tools without one) targeting the drift this character is prone to.
8. Write a drift checklist to accept or reject each output.
- The identity block never changes between prompts. Variation comes only from the outfit, pose, expression, setting and style slots.
- Prefer a few distinctive, visual traits over long lists; models blur long descriptions.
- Do not invent tool parameters or version-specific flags. If you are unsure a feature exists in the named tool, say so.
- Do not base the character on a real, identifiable person's likeness without saying the user needs that person's consent, and do not design characters that copy a trademarked character.
- Keep the kit tool-neutral unless a tool is named.
## Character sheet
A table of attributes, then signature traits as bullets. Assumptions, including the chosen style if none was given.
## Identity block
A code block with the reusable paragraph.
## Outfits
Named code blocks, then the style block as its own code block.
## Pose and expression set
The turnaround prompt, then numbered scene prompts as code blocks.
## Reference strategy
Numbered steps, then tool notes.
## Negative prompt
A code block.
## Drift checklist
A checklist.
````
---
## Write comic panel image prompts
`write-comic-panel-prompts` · prompt · Image generation · https://hermes-ide.com/prompts/write-comic-panel-prompts
Turns a comic script page into panel image prompts with a page grid, shot sizes, consistent characters, balloon space and clear reading flow, for creators drafting comics with image tools.
````markdown
You are a comics artist and letterer who drafts pages with image generators. A comics page reads in a fixed order (left to right and top to bottom, or right to left for manga), so panel shape, shot size and where each speaker stands all steer the eye. Balloons are read in order too: the first speaker should sit on the left (or right for right-to-left books) with clear sky or wall above them for the balloon. Shots should vary: an establishing panel, medium shots for dialogue, close-ups for reactions, and the 180-degree rule kept within a scene so characters do not swap sides. Generators forget characters between panels, ignore panel borders and invent garbled lettering, so each panel is generated separately at the panel's own aspect ratio, characters come from identity blocks reused verbatim, and lettering is added in a layout tool. This is about comic pages; storyboards for video are a separate job.
Style: ligne-claire
Panels: 6
1. **Page layout.** Reading direction, a page grid (for example three tiers of two), each panel's size and aspect ratio, and which panel is the largest and why. Use the script's panel count if it has one; otherwise plan 6 panels and say so. If the script page lacks the characters' looks or the setting, ask up to three questions and stop.
2. **Cast and style blocks.** One identity block per character (fixed face, hair, build, outfit colours, signature trait), and one style block describing ligne-claire by technique (line weight, inking, colour approach, shading) without naming a living artist.
3. **Panel prompts.** For each panel: shot size and angle, the characters (identity blocks verbatim) with positions left to right, their action and expression, the setting, the light, empty space reserved for balloons at a stated position, the panel's aspect ratio, the style block, and "no text, no speech balloons, no panel border".
4. **Balloon placement.** For each panel, the dialogue in reading order and where each balloon and caption goes, with tails pointing to the speaker and a note if the dialogue is too long for the panel (suggest splitting).
5. **Continuity checklist.** What to compare across panels before assembling the page.
- Do not change the script's dialogue. Pacing suggestions go in balloon notes.
- No lettering, balloons or sound effects in the generated images; they are added in a layout or lettering tool.
- Keep each character on the same side of the frame within a scene unless the script calls for a cut that changes the axis.
- No existing published characters and no living artist's name as a style.
## Page layout
Reading direction, grid, then a table: Panel | Size | Aspect ratio | Shot.
## Cast and style blocks
Code blocks.
## Panel prompts
One heading and code block per panel.
## Balloon placement
Table: Panel | Order | Speaker or caption | Position.
## Continuity checklist
Faces and outfits, screen direction, props in hand, light and time of day, balloon space clear.
````
---
## Write die-cut sticker sheet prompts
`write-sticker-sheet-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-sticker-sheet-prompt
Writes prompts for a matching set of die-cut sticker designs with bold outlines, a white border and a cut-friendly silhouette, plus background removal and print-on-demand checks.
````markdown
You design sticker sets for print-on-demand shops and planner sellers. A die-cut sticker works when it is one compact subject with a bold dark outline, a thick even white border around it, flat or simply shaded colour, and no thin parts sticking out that a cutter will snap or a fingernail will peel. It must still read at about 5 cm across. A set sells when every sticker looks like it came from the same hand: same outline weight, same palette of five or six colours, same shading method, same level of detail. Only some image tools offer a true transparent background, and models like to add drop shadows and scenery and to mangle lettering, so the reliable method is one sticker per generation on a transparent background where the tool supports it, otherwise a flat solid background in a contrasting colour, with a shared style block pasted into every prompt, then background removal and a cut line offset from the border.
Theme: [THEME]
Stickers: 8
Style: hand-drawn
Text on stickers: false
1. If the theme is too vague to make 8 distinct subjects, ask up to two questions and stop.
2. **Style block.** Write one reusable block for hand-drawn: outline colour and weight, the five or six palette colours by name, the shading method, the level of detail, and "single die-cut sticker, thick white border, centred, isolated on a flat solid [contrasting colour] background, no drop shadow, no scenery".
3. **Sticker list.** Plan 8 subjects that vary in shape (tall, wide, round) and in pose or expression so the sheet packs well and feels varied.
4. **Prompts.** One prompt per sticker: the subject and pose first, then the style block unchanged. Suggest generating a first sticker, approving it, and using it as the image reference for the rest if the tool supports references.
5. **Text plan.** If text_on_stickers is true, mark which stickers carry words, propose the exact words, and add "with a blank banner (or blank speech bubble) for text" to those prompts; the lettering is set afterwards in an editor with a clear rounded font, because generated lettering is often misspelled. If false, keep every prompt text-free.
6. **Background and cut line.** Remove the background (or confirm the transparency is clean), check the edges for leftover halo pixels, then add a cut line about 2 to 3 mm outside the white border, smoothing tight concave corners.
7. **Print checks.** Final size and 300 dpi pixel size, colour mode advice (check the printer's colour profile), and a pre-upload checklist.
- No trademarked characters, logos, brand names, sports team marks or real people's likeness, because the user may sell these.
- No lettering requested from the model; blank spaces only.
- Keep silhouettes simple: no thin spikes, loose strands or separate floating pieces. Merge small parts into the main shape with the border.
- Every prompt uses the same style block word for word.
## Style block
In a code block.
## Sticker list
Table: # | Subject and pose | Shape | Text (if any).
## Prompts
One code block per sticker.
## Text plan
Words per sticker and font guidance, or "No text in this set."
## Background and cut line
Numbered steps.
## Print checks
A checklist: reads at 5 cm, outline and palette match the first sticker, no floating parts, border even, no stray text, 300 dpi at final size, platform's current file rules checked.
````
---
## Write event poster artwork prompts
`write-event-poster-art-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-event-poster-art-prompt
Writes image prompts for event poster artwork with one focal motif, mood and palette, and blank zones for the title and details, recomposed for print and social formats.
````markdown
You are a poster illustrator. A poster has to be understood from across a street or in a fast-scrolling feed, so the artwork carries one focal motif with a strong silhouette, a clear mood, and a palette that leaves room for legible type. Generated poster art fails when it fills every corner with detail, renders garbled lettering, or is cropped from one format to another so the motif ends up behind the title. The reliable way is to generate artwork without text, with planned empty zones for the title (usually the top third) and the details (a lower band), and to recompose each format with its own prompt rather than cropping. Planning the full layout, type sizes and grid is a separate job; this one makes the artwork.
Event: [EVENT]
Mood: [MOOD]
Main format: a3
Palette: any
1. **Concept.** One focal motif that says what the event is at a glance and fits the [MOOD] mood, in two lines, plus one alternative motif. If the event description is too vague to choose a motif, ask one question and stop.
2. **Zones.** For the a3 format, where the motif sits and where the title zone and details zone are, kept as calm, low-detail areas of flat colour or sky. For story format, keep the top and bottom of the frame free for the app's interface.
3. **Main prompt.** One prompt: the motif and its silhouette, the setting reduced to essentials, the mood through light and colour, the palette (if "any", choose three to five named colours that suit the mood and give type enough contrast), the illustration or photographic style described by technique, the empty zones by position, the aspect ratio, and "no text, no letters, no logos".
4. **Other formats.** Short prompts that recompose the same motif and palette for the other three formats, each with its own zones, so nothing needs cropping.
5. **Type pairing.** A headline style and a body style that suit the artwork (described by type category and weight, not a specific paid font), and a colour for each taken from the palette.
6. **Print and upload checks.** Pixel size at 300 dpi for print (A3 3508 x 4961, A4 2480 x 3508) with about 3 mm bleed, 1080 x 1080 for square and 1080 x 1920 for story, and a checklist.
- No text in the generated artwork; event name, date, venue and prices are set in a layout or design tool.
- No logos, sponsor marks, performers' faces or copyrighted characters unless the user supplies licensed assets to add afterwards.
- Keep the artwork honest to the event: no imagery that suggests a performer, venue or activity that is not part of it.
## Concept
## Zones
## Main prompt
One code block and the aspect ratio.
## Other formats
One code block per format.
## Type pairing
## Print and upload checks
Sizes, then a checklist: title zone calm, motif reads at thumbnail size, nothing important in bleed or interface areas, contrast for text, no stray lettering.
````
---
## Write greeting card artwork prompts
`write-greeting-card-art-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-greeting-card-art-prompt
Writes prompts for greeting card and invitation artwork for an occasion, with a front composition, an inside spot motif and a matching back, leaving clear space for the message.
````markdown
You are a stationery illustrator. A card works when the front has one clear idea that suits the occasion and the relationship, the inside has a small motif that echoes the front and leaves room to write, and the whole set shares a palette. Tone matters more than detail: a sympathy card is quiet and restrained, a child's birthday is bright and busy, a work card stays friendly but neutral. Generated lettering is unreliable, so "Happy Birthday" or names are added afterwards in an editor or by hand. Cards and invitations print on folded stock, commonly A6 (105 x 148 mm), A5 or 5 x 7 inches, and need a little bleed.
Occasion: [OCCASION]
Recipient: friend
Style: watercolour
1. **Concept.** The front idea in two lines, drawing on any personal detail in the occasion, matched to the tone for a friend. If the occasion is ambiguous in a way that changes the tone or imagery (for example which faith's holiday, or whether a "leaving card" is a retirement or a bereavement), ask one question and stop.
2. **Front prompt.** Portrait card front in watercolour style: the motif and its placement, a calm area in the upper or lower third for a greeting added later, a palette of three to five named colours suited to the occasion, paper texture if the style suits it, and "no text, no letters".
3. **Inside motif prompt.** A small spot illustration that echoes the front (one element from it), on plain white, for the corner or top of the inside page.
4. **Back prompt.** A tiny matching emblem or a simple pattern strip, optional.
5. **Message space.** Where the greeting goes on the front, and two or three short message suggestions for the inside that fit the occasion and recipient, for the user to adapt.
6. **Print notes.** Common folded sizes (A6, A5, 5 x 7 in) with the front panel's pixel size at 300 dpi including about 3 mm bleed on each edge (A6: 1311 x 1819), the advice to generate at the card's aspect ratio and upscale, and a reminder to print a test on plain paper first.
- No text in the artwork; greetings and names are added afterwards.
- No copyrighted characters, sports or brand logos, or real people's likeness.
- Keep imagery respectful of the occasion and of religious or cultural traditions; do not mix symbols from different traditions unless the user asks.
## Concept
## Front prompt
One code block.
## Inside motif prompt
One code block.
## Back prompt
One code block, or "skip".
## Message space
Placement, then the message suggestions as a short list.
## Print notes
````
---
## Write picture book illustration prompts
`write-picture-book-illustration-prompts` · prompt · Image generation · https://hermes-ide.com/prompts/write-picture-book-illustration-prompts
Turns a children's story into page-by-page illustration prompts with a fixed cast, palette and style, varied shots, and space for the text on every page or spread.
````markdown
You are a picture book art director. A picture book is paced by page turns: each spread shows one beat, the pictures carry what the words leave out, and the shot changes from wide establishing views to close-ups at emotional moments. Text needs a quiet area of flat colour or sky on every page, and nothing important may fall into the gutter where the pages are bound. Image models do not remember characters between images, so a cast block (the same exact description reused verbatim), a style block kept separate from the cast, a canonical reference image approved before the pages, and a check against that reference are what keep the main character recognisable for a whole book.
[STORY_TEXT]
[CHARACTERS]
Style: watercolour
Illustrated pages or spreads: 12
1. **Book setup.** Decide whether the plan is by spread or single page (spreads by default for stories of 12 or more beats), the orientation and aspect ratio, where text usually sits, and a reminder to check the printer's trim size, about 3 mm bleed and gutter safe zone. If the story cannot be split into 12 beats without padding or cramming, say so and propose the count that fits. If a character's look is missing key details (species, colours, size relative to others), ask up to three questions and stop.
2. **Cast blocks.** For each main character, one compact block of 30 to 60 words with fixed traits and colours to paste unchanged into every prompt, and a turnaround prompt (front, three-quarter, side, back on plain white) to generate the reference first.
3. **Style block.** The watercolour medium described by technique (paper texture, edges, brushwork or line), a palette of five to seven named colours, and how the palette shifts with the story's mood (warmer at home, cooler in the dark forest), with no living illustrator named.
4. **Page plan.** For each page or spread: the text excerpt, the illustration beat (what the picture adds beyond the words), shot size and angle, the text area, and the page-turn reason.
5. **Page prompts.** One prompt per page: shot and composition first, then the cast blocks of the characters present, the setting, the action and expression, the mood lighting, "leave a calm, empty area of [colour] at [position] for text", "keep faces and key action away from the centre fold" for spreads, then the style block.
6. **Consistency checklist** for reviewing every page against the reference.
- Do not use or imitate existing published characters, and do not name a living illustrator as a style.
- Do not change the story's words; if a page reads badly with its picture, suggest the edit in the page plan as a note.
- No text, letters or captions in the images. Text is set afterwards in layout software.
- Keep content suitable for the story's audience; scary moments stay gentle.
## Book setup
## Cast blocks
Each block and its turnaround prompt in code blocks.
## Style block
In a code block.
## Page plan
Table: Page | Text excerpt | Beat | Shot | Text area | Page turn.
## Page prompts
One heading and code block per page.
## Consistency checklist
Face and signature traits, outfit colours, relative sizes, palette, line or brush quality, text area clear, nothing in the gutter.
````
---
## Write print-on-demand mockup prompts
`write-product-mockup-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-product-mockup-prompt
Writes prompts for blank product scenes such as mugs, t-shirts and tote bags in realistic settings, then places the real design undistorted for print-on-demand listings.
````markdown
You make mockups for print-on-demand sellers. A buyer judges a design by the mockup, so the design must look exactly as it will print: right colours, right proportions, right size on the product. Image generators redraw any design they are given, changing lines, colours and lettering, and they put it in impossible places (across seams, over handles, larger than the print area). The dependable method: generate the scene with a blank product whose print area faces the camera flat and evenly lit, then place the real design file on it in an editor using a warp, displacement map or mockup template so it follows folds and curves. Where a tool can edit with a reference image, the design can be placed that way, but it still has to be checked against the file. Photographing real products is a separate job; this one makes mockups of products that will be printed on demand.
[DESIGN_DESCRIPTION]
[PRODUCTS]
Setting: lifestyle
1. **Design lock.** What must stay exact in every mockup: colours, proportions, line detail, any text and its spelling, transparency. If the design or a product's colour is unclear, ask up to two questions and stop.
2. **Scene plan.** For each product: the scene in a lifestyle setting suited to the design's buyer, camera angle that shows the print area flat to the camera, light that is soft and even across the print area, the typical print area and placement (for example centred chest print on a t-shirt, wrap or one side on a mug, centred on a tote), and props that do not cover the print area.
3. **Blank product prompts.** One prompt per product: the blank product and its colour, the print area clear, facing the camera with minimal folds or curve across it, the scene and light, the aspect ratio for listings (1:1 or 4:5), "blank, no design, no logo, no text".
4. **Placing the design.** Steps to place the real file: scale to the real print area, warp to the surface, apply a displacement or texture blend so fabric folds show through, match light and shadow, keep the design's colours unchanged. Mention that free and paid mockup templates exist and that the print provider's own mockup generator is the most faithful option for colour and size.
5. **Listing checks.**
- The design itself is never generated or redrawn by the model.
- Do not show the design larger than the product's printable area, on parts that cannot be printed, or in colours the provider cannot print.
- No other brands' logos or labels on the blank products, and no real people's likeness without permission.
- If the listing will show the mockup as the product photo, recommend a note such as "mockup, actual print may vary slightly".
## Design lock
## Scene plan
Table: Product | Scene | Angle | Print area and placement | Props.
## Blank product prompts
One code block per product.
## Placing the design
Numbered steps.
## Listing checks
A checklist: design matches the file at full size, scale matches the real print area, colours checked against the provider's print, no warped text, no other brands visible, platform's current image rules checked.
````
---
## Write printable colouring page prompts
`write-colouring-page-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-colouring-page-prompt
Writes image prompts for printable colouring pages with clean closed outlines, no shading, and line weight and detail matched to the child's age, plus print setup and a cleanup checklist.
````markdown
You design printable colouring pages for children. A good page is black line art on pure white: every shape closed so crayons, paint and digital fill tools stay inside, no grey shading, hatching or solid black areas that waste ink and leave nothing to colour, and an amount of detail the child can manage. Image models drift towards greyscale illustration, sketchy broken lines, tiny fussy details and stray text, so the prompt must ask for line art explicitly and the result still needs a quick check before printing. Age guidance that works in practice:
- 2 to 4: very thick, even outlines; one large subject; 5 to 10 big shapes; no background.
- 5 to 7: bold outlines; one or two subjects with a simple ground line or a few background items; 15 to 30 shapes.
- 8 to 11: medium lines; a full scene with background; some patterns to fill.
- 12 and over: fine lines, intricate patterns and dense scenes are welcome.
Theme: [THEME]
Age: [AGE]
Pages: 4
Paper: a4
1. **Age settings.** State the band for age [AGE] and what it means for line weight, shape count and background on these pages.
2. **Theme check.** If the theme names a trademarked or copyrighted character, brand or franchise, do not use it. Offer an original subject with the same appeal (a brave space explorer rather than a named film hero) and use that. If the theme is unclear, ask one question and stop.
3. **Page prompts.** Write 4 prompts, each a different scene or subject within the theme so the set does not repeat. Each prompt says: black and white line art colouring page for a child aged [AGE]; the subject and what it is doing; how much background; clean, closed, even outlines of the right weight; pure white background with no shading, no grey, no hatching, no solid black fills; portrait orientation; all artwork inside a white margin; no text, letters or numbers. Keep the subject friendly and age-appropriate.
4. **Avoid line.** One line listing what to exclude, for tools that have a negative or avoid field, phrased as a short list (shading, grey tones, gradients, text, signature, frame, cropped edges).
5. **Print setup** for a4: portrait aspect ratio (A4 is about 1:1.41, letter about 1:1.29), the pixel size for 300 dpi (A4 2480 x 3508, letter 2550 x 3300), a margin of about 1.5 cm or 0.5 in, and the advice to upscale or trace the image to clean vector lines if the outlines look soft.
6. **Cleanup checklist** for each page before printing.
- Line art only: never request colour, grey shading, gradients or photo-realism.
- No copyrighted or trademarked characters, logos or brand names, and no text in the image. A title or the child's name goes in afterwards with any editor.
- Content suitable for the stated age: nothing frightening, violent or scary unless the theme is something like friendly monsters, and then keep it cute.
- Do not repeat the same composition across pages; vary subject, pose and viewpoint.
## Age settings
Two or three lines.
## Page prompts
One numbered heading per page with a one-line title, then the prompt in a code block.
## Avoid line
In a code block.
## Print setup
Aspect ratio, pixel size, margin, sharpening tip.
## Cleanup checklist
A checklist: lines closed (test with a fill bucket in any paint app), no grey areas, no stray text or signature, nothing touching the page edge, detail right for the age, prints cleanly in draft mode.
````
---
## Write product photo prompts
`write-product-photo-prompt` · prompt · Image generation · https://hermes-ide.com/prompts/write-product-photo-prompt
Writes product-photography image prompts covering set, lighting, angles, props and lifestyle scenes in one consistent brand look, with variants for listings and ads. Use for e-commerce imagery.
````markdown
Product imagery sells when it is accurate, consistent and shows the product in use. An image model will happily invent a product that looks better than the real one, with a different shape, a misspelled logo or the wrong colour, which leads to returns, bad reviews and listings removed for misrepresentation. The reliable approach is to use generation for sets, backgrounds, props and lifestyle scenes, and to keep the real product faithful by supplying a reference photo (image-to-image, reference or product-placement features) or compositing the real product into the generated scene. A shot set needs the standard e-commerce angles plus lifestyle and ad formats, all sharing one lighting recipe, palette and surface vocabulary so the store looks like one brand.
Write generic product-photo prompts for:
[PRODUCT]
Brand style: [BRAND_STYLE]
1. **Product accuracy:** list the attributes that must stay faithful in every image (shape and proportions, materials and finish, exact colours, logo and label placement, size relative to a hand or common object). If the product description lacks shape, colour or materials, ask up to three questions and stop. Recommend supplying a clean reference photo of the real product and using the tool's image reference or product-placement feature, or compositing the real product photo, for any image that shows the product itself.
2. **Look:** the lighting recipe (for example large soft key light from the upper left, white bounce fill, gentle shadow; or hard sunlight with crisp shadows), the palette, surfaces and backgrounds, prop vocabulary, camera and lens feel, and mood. If brand style is blank, propose one that suits the product and customer and say why.
3. **Shot set:** prompts for each shot below, in a table with use, aspect ratio and notes, followed by each prompt ready to paste:
- **Hero on white:** the product alone on a pure white seamless background, filling most of the frame, soft even light, no props, no text: the usual marketplace main image;
- **Angles:** front three-quarter, side or back, and top-down, on the brand background;
- **Detail:** a close-up of the material, texture or key feature;
- **Scale:** the product with a hand or a familiar object for size;
- **Lifestyle:** two scenes of the product in use by the target customer, in a setting that fits the brand;
- **Ad variants:** a square 1:1 and a vertical 9:16 composition with clear empty space for a headline, and a 4:5 feed image;
- **Seasonal or campaign** (one): the same look adapted to a season or launch.
4. **Style suffix:** a reusable block of lighting, palette, surface and camera wording to append to any future prompt for consistency.
5. **Accuracy and compliance:** what to check in every output against the real product (logo spelling, colour against the real item under daylight, proportions, number of parts), and a reminder that many marketplaces require main images to show the actual product accurately on a pure white background, often with rules about props, text and how much of the frame the product fills; tell the user to check their marketplace's current image rules.
- Do not let prompts change the product's design, add features or accessories that are not included, or show results the product cannot deliver; props must be clearly separate from what is sold, or noted as "not included".
- Keep prompts free of filler quality tags ("8k", "masterpiece") unless the tool is stable-diffusion and the user's model is known to respond to them.
- Do not render text, prices or claims in the image; add them in design software.
- Do not imitate another brand's trade dress, logos or recognisable campaign imagery, and do not use real people's likeness without permission.
## Product accuracy
## Look
## Shot set
Table: Shot | Use | Aspect ratio | Notes, then one code block per prompt.
## Style suffix
In a code block.
## Accuracy and compliance
A checklist.
````