Skip to content

Features

Everything between a video idea and a finished thumbnail

Six generation modes, identity profiles, references, AI editing, a layer editor and clean exports — built around how YouTube thumbnails are actually made.

01/Generation modes

Six ways to get to the right image

Start from nothing, from a photo, from a thumbnail you admire, or from the thumbnail you already have.

Text to thumbnail

Describe the video and the idea; Thumbora writes the detailed prompt and generates a 16:9 image with your headline.

Input: title, idea, thumbnail text

Photo to thumbnail

Use photos from your identity profile so the person on the thumbnail looks like you, with the pose and emotion you describe.

Input: identity profile + idea

Reference to thumbnail

Take the composition, colors, lighting, style or background from thumbnails you like and get a new, original image.

Input: one or more references

Redesign

Upload your current thumbnail and get a stronger version of the same idea: a clearer focal point, readable text, better contrast.

Input: your existing thumbnail

AI edit

Change a finished image with an instruction — swap the background, recolor the headline, move the text. The result is a new version.

Input: a version + an instruction

Variations

Get alternatives of a version you like to explore color, layout and framing without starting over.

Input: a version

02/Faces and references

Your likeness and your inspiration, handled carefully

Face photos and reference thumbnails make results better. They also come with responsibilities, so the rules are built into the product.

Identity profiles

  • Upload a few clear, well-lit photos of yourself to a profile and reuse it in any project.
  • Photos are sent to the image model only as references for your own generations.
  • The model keeps the likeness as closely as it can; results can still differ, so review each one.
  • Every face upload requires your confirmation that the photos show you, or that you have the person’s permission.
  • Delete a photo or the whole profile at any time — the files are then removed from storage.

References

  • Choose exactly what to take from a reference: composition, colors, lighting, style or background.
  • Thumbora builds a new composition; it is not a tool for copying someone else’s thumbnail.
  • Logos, characters and other people’s photos remain their owners’ property — use them only with permission.
  • Uploads are checked (JPG, PNG or WebP) and metadata such as EXIF, including location, is removed.

03/Styles

12 styles tuned for YouTube genres

A style adjusts composition, lighting, color and typography in the prompt — you can still describe anything on top of it.

Cinematic
Film-like lighting, depth and color grading for stories and big topics.
Minimal
One subject, plenty of space and a short, sharp headline.
Gaming
High energy, saturated color, dynamic angles and bold outlines.
Business
Clean layouts, charts and objects that signal money, growth or strategy.
Technology
Devices and hardware in dramatic light with crisp, modern type.
Documentary
Natural light, real places and calm, confident titles.
Education
Friendly colors, clear diagrams and a promise of what you’ll learn.
Reaction
Expressive faces, arrows and circles pointing at the moment that matters.
Luxury
Dark, rich palettes, gold accents and refined typography.
Challenge
Stakes, timers and contrasts that make the outcome feel uncertain.
Before / after
A split composition that shows the transformation at a glance.
High contrast
Maximum separation between subject, text and background for small screens.

04/Models

Google’s Nano Banana image models

Thumbora works with Nano Banana models through reAPI. Each has its own strengths, resolutions and credit cost.

Nano Banana 2.1default

1K · 2K · 4K

Google’s newest image model. Best text rendering and subject consistency. Same price at 1K, 2K and 4K.

1K: 10 cr · 2K: 10 cr · 4K: 10 cr

Available on all plans

Nano Banana Pro

1K · 2K · 4K

Gemini 3 Pro Image. Strong photorealism and lighting; slower at 4K.

1K: 10 cr · 2K: 10 cr · 4K: 12 cr

Plans: Starter, Pro, Business

Nano Banana 2 Lite

1K

Fast, low-cost drafts at 1K. Great for exploring ideas before a final render.

1K: 5 cr

Available on all plans

Credits per output image for text-to-thumbnail generation. The models available to you depend on your plan.

05/Editor, versions, export

Finish the details yourself

AI gets you most of the way. The editor, the version history and exports take care of the rest.

Layer editor

  • Add and edit text layers on top of a generated image.
  • Place images and logos and arrange the layers.
  • Save your changes as a new version — the original stays untouched.

Projects and versions

  • Each project keeps every generation, edit and upload as a numbered version.
  • See which version an edit or a variation came from.
  • Star favorites, add notes and choose the project cover.

Export

  • PNG, JPG or WebP.
  • Original size, 1280×720 or 2560×1440 — always 16:9.
  • If an export is larger than the source image, it is upscaled and clearly labeled as upscaled, never presented as native.

06/Credits and privacy

Predictable costs, private files

You should always know what a generation costs and who can see your images.

Credits

  • The exact cost is shown before you generate.
  • Credits are reserved at start and charged only for delivered images.
  • Failed, timed-out or cancelled images are refunded automatically.
  • A notification — and an email, if you enable it — tells you when a generation is done.

Privacy

  • Files are kept in private object storage and opened only through short-lived signed links.
  • Images are shared with the AI provider only to process your request.
  • Thumbora doesn’t use your photos to train AI models.
  • Delete files, projects or your whole account at any time.

See it on your own video

Sign up, describe your next upload and compare the results.