Image to OG image
Upload your logo or any card you love. AI paints an on-brand Open Graph image around it — tag images as Logo or Reference, then ship.
Image to OG Image generator
How people use it
Six recipes that combine an upload, a style, and an optional note. Upload above and try one.
- Logo → Swiss minimalDrop your logo, pick Swiss minimal, ship a clean B2B card with brand colors honored.
- Logo → Brutalist launchDrop your logo + add 'v2 launch announcement' as a note. Pick Brutalist.
- Magazine cover you love → newspaperDrop a cover you admire as Reference. Pick Newspaper for a serif editorial spin.
- Product photo → editorial frameDrop a product shot as Reference, pick Editorial photo, no note needed.
- Pixel-art screenshot → splashDrop a game screenshot, pick Pixel retro, get an 8-bit splash variant.
- Logo + accent shot → mark on colorDrop your logo and a swatch you love (Reference). Pick Brand mark.
Image to OG Image FAQ
What gets uploaded, how the model uses it, and what counts as a generation.
What does the Logo tag do?
Image #1 tagged Logo tells the model to preserve your brand mark exactly — same shape, glyphs, proportions, and colors — and to place it at a safe ~18-24% canvas width. Reference images influence mood, palette, and composition only; the model is told NOT to copy specific text, faces, or trademarks from them.
Where do my uploads go?
Files are downscaled in your browser to 1024px on the longest edge, then sent as base64 directly to the AI gateway with your generation request. We do NOT store the original file on disk and we do NOT log it. After the response is returned, the upload is discarded.
What formats and sizes work best?
PNG (transparent logos work great), JPG, and WebP up to 4 MB raw. SVG and animated GIF are not supported — convert to PNG first. Square or 1:1 logos give the model the most room to compose around.
What counts as a generation?
Each Generate press counts as one AI image, regardless of whether you uploaded one image or three. Anonymous users get 3 AI generations per day (shared with the Text to OG tool). Enhance is metered separately at 30/day and does not eat into your image quota.
Is the AI allowed to use copyrighted reference images?
Reference images are meant as your own brand assets or images you have rights to. The model is instructed NOT to copy specific text, faces, or trademarks from references, but you are responsible for what you upload. For famous brands or characters the model will often refuse and return an error — that's expected.
One image in.
A shareable card out.
No URL. No template. No design tool. Just upload what you want the card to be built around — and ship.