Thumbnail Maker

You describe the video’s hook — the subject, the moment, and any title text — and Made sends the description to the Nano Banana 2 image model, which renders a 16:9 cover image and returns it as a PNG at 1K resolution. Generate a few takes and pick the one that reads best small.

Not on this page — this workflow generates entirely from text, so the people in the render are generated, not you. To build a thumbnail from your own photo — cut yourself out, boost the colors, add text and a background — upload the photo to the AI Photo Editor instead.

Current models render short text well, but spelling is not guaranteed on every run. Keep titles to a few words, put the exact wording in your prompt, and proofread before publishing — regenerate or rephrase if a letter drifts.

The output is a 16:9 PNG at 1K resolution, which matches the shape video platforms expect. If your platform asks for an exact pixel size such as 1280 × 720, run the result through Image Resizer; PNG uploads are broadly accepted.

Design for the glance: one clear subject with a strong expression or action, a high-contrast background that separates from the platform’s white or dark UI, and no more than about four words of text. Say “readable at small sizes” in the prompt — it pushes the model toward bolder, simpler compositions.

Yes, by treating the style as a reusable template: fix the palette, text treatment, and layout in one description and change only the subject and title per video. Each generation is a new render, so expect sibling-level consistency rather than identical frames.

Yes. Switch the aspect ratio to 9:16 for vertical formats, and keep the focal point centered since short-form UIs overlay buttons and captions along the edges. The same prompt principles apply — one subject, high contrast, minimal text.