AI nutrition label image: composite your real panel, don't generate it
A nutrition or ingredient panel is regulated text. Generate the package art, then place your real, proofread panel on it. Never let a model write the numbers.
For an AI food package image that shows a nutrition facts panel, generate the pack without the panel, then composite your real, proofread panel onto it. Calories, allergens and ingredients are regulated, exact text. An image model can render plausible small print that is wrong, so the numbers must come from your own file.
Why small print is risky
Small dense text is the hardest thing for an image model to hold. Rows can merge, digits can change and a line can be invented. For a mockup that nobody will read, that is harmless. For a listing, a deck for a retailer or a label proof, a wrong number is a compliance problem.
Generate the pack with a blank back panel
Ask for the package, the material, the front art and a clean white rectangle where the panel goes. State that the panel area must stay blank, with no text and no lines.
curl -X POST https://api.sume.com/v1/images \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"openai/gpt-image-2.5","prompt":"Flat back view of a matte kraft granola pouch with a plain white rectangle in the lower center left blank for a label, no text on the rectangle, soft studio light","quality":"high","aspect_ratio":"3:4"}'Place your panel
Export your approved panel as a PNG at the right proportions and paste it into the white rectangle with an image library or a design tool. Keep the panel as a separate layer or file so a later change to a value does not need a new generation. If the pack is curved in the picture, use a perspective transform so the panel follows the surface.
If you need an edit, use a reference
To place a flat artwork onto a surface in one call, send the pack as the first reference and the panel as the second and ask that the second image be placed unchanged on the white rectangle. Then read the panel at full size against the source. Models can still alter fine print, so treat the edit as a draft and swap in the real file for any final.
Before it goes to a buyer
- Compare every number with the approved source.
- Check the allergen line character by character.
- Keep the panel file and the pack art file with the job id.
- Label the image as a mockup where the platform requires it.
- Have the regulatory owner sign off on the final file.
Related posts
More in Use cases
- AI picture book page art: text-free art, story text set by you
Make each picture book page as text-free art in one style, then set the story text in your layout. Reference the first page so the style holds.
- AI price tag image: make the blank tag, set the price in code
For a price tag or shelf label image, let the model draw the tag and write the number yourself. Prices must be exact, and generated digits can drift.
- AI product photo with a hand holding it: packshot plus pose
To put a hand in a product shot, send the packshot as the first reference and a pose photo as the second, and say to keep the product identical.
- AI thumbnail readability test: check the text at 168 by 94 pixels
Before you publish an AI thumbnail, shrink it to a small display size and read the text. If it fails there, regenerate with fewer words and larger type.
Written by Sume