All use cases

Consistent character AI video

Make a character once and reuse it. Blooper keeps a pinned look for each character, and every frame they appear in is made from it, so the same person comes back shot after shot.

One woman, two shots of NØIR, a perfume film made in Blooper. Every frame on this site was made in Blooper. See the whole project.

What Blooper does for it

The usual failure of AI video is the second shot. The first prompt makes a woman in a red dress; the second makes a different woman in a different red dress. Prompting harder does not fix it, because each generation starts from nothing. Blooper gives every character a home in the project: a set of looks, one of them pinned, used as the reference for every frame that character appears in.

Clips are made from approved frames, so the face you approved in the still is the face that starts the clip.

How to make it in Blooper

  1. Describe the character or upload a reference

    A description, a sketch, or photos of a person you have the rights to use. Say what must never change: the face, the hair, the scar, the jacket.

  2. Make a few looks and pin one

    Generate a few looks of the character and pin the one that is right. That pinned look is the character from now on.

  3. Cast them in the shots

    On a storyboard, the characters each shot needs are bound when the board is written, so every frame is made with its own cast and not with everyone in the project.

  4. Approve the frames, then animate

    Check the face in the still. Approved frames become clips, and a clip can start from the last frame of the one before it, so a move continues across the cut.

  5. Keep them for the next cut

    The character stays in the project with its looks and versions, ready for the next scene, the next ad in the campaign or the next variation.

Which models suit it

Consistency is decided in the frame, so the image models matter most here: Nano Banana 2.1 and Seedream 5.0 make the frames from the character's references. The video models, FLUX 3 Video, Kling 3.0, Kling 4, Seedance 2.5, Veo 3.1 and Wan 3.0, then animate frames you already approved.

Video models6

  • Wan 3.0 Alibaba up to 1080p · 30s from 14 credits Longest clipsLowest priceEdits video
  • Veo 3.1 Google up to 4K · 8s from 28 credits
  • Seedance 2.5 ByteDance up to 4K · 30s from 32 credits Longest clipsEdits video
  • Kling 3.0 Kuaishou up to 4K · 15s from 35 credits
  • FLUX 3 Video Black Forest Labs up to 4K · 20s from 42 credits
  • Kling 4 Kuaishou up to 1080p · 30s from 66 credits Longest clipsEdits video

Image models2

  • Seedream 5.0 ByteDance up to 4096px from 5 credits Lowest price
  • Nano Banana 2.1 Google up to 4K from 9 credits

“From” is the model’s cheapest setting: lowest resolution, shortest clip. Read from our price list when this page was built.

What it costs

A frame starts at 5 credits and a clip at 14, on the cheapest models at their cheapest settings. Six frames, for the character's looks and four shots, and four clips start at 86 credits. Every price is shown before it is spent.

Example: a character and 4 shots

Example: a character and 4 shots, at each model’s cheapest setting
6 frames× from 530 credits
4 clips× from 1456 credits
From86 credits

Pay as you go: credits never expire · no subscription. Packs are $25 for 2,500 credits, $100 for 10,000 credits and $250 for 27,500 credits, or any amount from $10. A shot that fails before the model runs costs nothing. See every price.

What keeps a character consistent

  • Pin the character before the first shot, not after the tenth.
  • Keep the outfit in the character, and change it on purpose. A wardrobe change you did not plan is the first thing a viewer sees.
  • Use close and medium shots for the moments that need the face. In a wide shot the face is a few pixels, and a few pixels can drift.
  • Check the last frame of every clip. That is where a face is most likely to change.

Questions

How do I make a consistent character in AI video?

Give the character a pinned reference and make every frame from it. In Blooper you pin a look once, the agent uses it for every frame the character appears in, and clips start from frames you approved.

Does it work across different AI video models?

Yes. The clips start from approved frames, so the character carries across FLUX 3 Video, Kling 3.0, Kling 4, Seedance 2.5, Veo 3.1 and Wan 3.0, whichever one makes each shot.

Can I reuse a character in more videos?

Yes. Make the next ad, cut or variation in the same project and the character is there with its looks and versions, so it starts from the same person.

Can I use a real person's photo as the reference?

Only with their permission and the rights to the image. Blooper never uses your uploads to train AI models.