creator-product-photo
Skill7 starsChanged 18 days agoMITView on GitHub
Brand-quality product imagery from a product photo or description, using whatever image provider the user has: studio packshots, lifestyle scenes, hands-in-frame close-ups, Pinterest pins, hero banners, social carousels, ad packs, virtual try-on, conceptual/CGI shots, restyles, and marketplace listing sets (Amazon/Etsy/Shopify main image, secondary images, A+ style modules). Use when the user wants product photos, packshots, e-commerce or marketplace images, listing images, product ads, or "put my product in a scene". Not for logos or brand systems (creator-brandkit) or video ads (creator-ad).
In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- creator-adSkill
- creator-audioSkill
- creator-brandkitSkill
- creator-characterSkill
- creator-explainerSkill
- creator-imageSkill
- creator-setupSkill
- creator-thumbnailSkill
- creator-videoSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 30
Check the route supports references: python3 scripts/creator.py plan --task image --needs refs.
- Distinctive: line 31
If nothing supports references, say the product's exact look can't be preserved and offer (a) description-only concept shots, or (b) adding …
- Very specific: line 41
Never let the model rewrite label text or logos.
- Distinctive: line 64
Don't fabricate certifications, awards, claims, or reviews in images or copy.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Creator Product Photo
Step 1 — Inputs
- Product reference — strongly preferred: 1–3 clean photos (front, angle, detail). Without one, the product will be invented; say so and get approval for that.
- Mode — infer from the request using
references/modes.md; ask only if two modes fit equally. - Brand — read config
brand(colors, logo) if present. - Count & ratio — mode defaults in
references/modes.md.
Check the route supports references: python3 scripts/creator.py plan --task image --needs refs.
If nothing supports references, say the product's exact look can't be preserved and offer
(a) description-only concept shots, or (b) adding a key / enabling Flow or the Gemini app.
Step 2 — Prompt
Build from the mode template in references/modes.md:
Use image 1 as the exact product — keep its shape, colors, label, and proportions
unchanged. + scene + lighting + camera + composition + constraints.
Never let the model rewrite label text or logos; if the label must be legible, say
keep the label exactly as in image 1.
Step 3 — Generate
python3 scripts/creator.py image --prompt "<prompt>" --ref product.jpg --ratio <r> --size 2k --out ./out/<mode>-01.pngFor sets (carousel, ad pack, marketplace set) generate the hero first, get approval, then
the rest with the same style sentence and pinned --provider.
Marketplace main image needs a pure white background: generate on white, then if the
background isn't clean, media.py cutout and composite onto white
(ffmpeg -i cut.png -filter_complex "color=white:s=2000x2000[bg];[bg][0]overlay=(W-w)/2:(H-h)/2" -frames:v 1 main.jpg).
Step 4 — QA every image
Compare with the reference: shape, color, label, logo, proportions, and count of parts. Reject and regenerate on any product drift; that's the #1 failure. Then report the set.
Rules
- Don't fabricate certifications, awards, claims, or reviews in images or copy.
- Marketplace rules (white background, no added text/badges on main image) are in
references/modes.md; follow them for listing images.
References
references/modes.md— every mode: ratio, count, prompt template, rulesreferences/image-prompting.md,references/media-ops.md,references/providers.md
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

