Covers that read at phone size

AI Dance Thumbnail Maker

A dance thumbnail has one job: stop the scroll. That means one face, one clear motion and three or four words — not a poster shrunk to 320 pixels.

One free 480p · 5s video preview per account. Image renders from 3 credits. No watermark.

A vertical dance video thumbnail with one face and a short headline

The short answer

Describe the frame or upload a photo, choose 9:16 or 16:9, and GPT Image 2 renders a thumbnail at 1K for 3 credits. Keep the headline to three or four words and the subject out of the bottom fifth of the frame.

Short text renders more reliably than long paragraphs. Add the small print in a design tool.

Resolution
Frame
Sign in to generate.

Specs at a glance

Model
GPT Image 2
Frames
9:16, 16:9, 1:1
Resolution
1K or 2K
Cost
3 credits at 1K, 6 credits at 2K
Best headline
Three to four words
Output
Image download, no watermark

How it works

  1. 1

    Pick the one face

    A thumbnail is read at 320 pixels wide. One face fills the frame; three faces become a texture. Crop to the person before you upload.

  2. 2

    Choose the platform frame

    9:16 for TikTok, Reels and Shorts. 16:9 for a YouTube long-form thumbnail. Do not design for one and crop to the other.

  3. 3

    Add three or four words

    “NEW TREND”, “DAY 1”, “POV: YOU”. The words should be readable at a glance. Long captions disappear into noise.

  4. 4

    Keep the UI safe area clear

    Platforms cover the bottom fifth with a caption and the right edge with buttons. Keep the face and the headline in the upper two-thirds.

What a thumbnail is not

It is not a poster. A poster is read at arm’s length and can carry a date, a venue and a lineup. A thumbnail is read at a glance on a phone, often while the viewer is already scrolling. Density is the enemy.

It is also not a substitute for the first frame. The video itself still has to deliver. A thumbnail buys the click; the first two seconds decide whether the viewer stays.

The safe area rule

TikTok and Reels put a caption and a username across the bottom, and action buttons down the right edge. If the face or the headline sits there, the platform covers it. Keep the important content in the upper two-thirds and the centre.

For 16:9 YouTube thumbnails, the bottom-right corner carries the duration badge. Keep the headline on the left and the face on the right, or leave the corner empty.

  • Face in the upper two-thirds
  • Headline on the left or top, three to four words
  • Bottom fifth empty on vertical covers
  • One accent colour, not five

Photo input and consistency

Uploading a photo keeps the creator’s face consistent across a series. Without a photo, the model invents a face, and the next thumbnail will invent a different one. For a channel, photo input is the difference between a brand and a pile of images.

Describe the expression, the light and the background in one sentence each. “Surprised, warm key light, dark studio” is enough. A long paragraph of styling usually produces a busier image, not a clearer one.

What people use it for

Shorts and Reels covers

A custom cover makes a profile grid look intentional instead of a wall of random frames.

YouTube long-form

A 16:9 thumbnail with one face and three words is still the highest-leverage image on the platform.

Series branding

Reuse the same composition and colour across a week of clips so the series is recognisable in a feed.

A/B testing a hook

Generate two covers for the same clip and swap them after a day. The image is cheap enough to test.

Frequently asked questions

Open the generator