Multi-image Custom mode
Upload product photos, logos, sketches, screenshots, or references, then describe the finished image you want.
Put the product photo, logo, sketch, and style reference in one request. PatentFig AI uses the full visual context to generate a result that matches the object, mark, material, and scene you have in mind.
Upload up to three image references in Custom mode
Combine product photos, logos, sketches, UI screens, or style references
Describe the final look, placement, material, and style in plain language
These product-shot examples show what you can create after providing product photos, logos, sketches, or style references.



Use multiple references to show the model the subject, the mark, and the visual style before generation.
Use a product photo, logo file, sketch, screenshot, reference render, or style image as source material.
Tell PatentFig AI where the logo should appear, what structure to keep, what material to use, and what kind of scene or output you want.
Generate the image, review it in your project, and keep refining or exporting from the same workspace.
Use it when one image is not enough to explain the desired result.
Upload a carrier photo and logo artwork, then describe where and how the logo should appear.
Use one image for the object and another for the finish, lighting, material, or layout direction.
Reference a screenshot, a target layout, and a supporting asset when asking for custom interface or diagram output.
Use multiple images when the result depends on more than one visual reference.
Because one image often explains the object, another explains the logo, and a third explains the style or material. Together they reduce guesswork.
Include the logo when placement, shape, proportions, or brand mark details matter. Then describe where it should appear in the final image.
The Custom composer accepts common web image inputs such as PNG, JPG, and WebP.
Yes. Upload both images and describe the placement, scale, material, and scene you want.
Open Custom mode, attach up to three reference images, and describe the final image in plain language.