Live now · Photos and video

The AI generation studio for photos and videos of your own persona.

Build your dream model in 30 minutes, no prompts. Pick every detail (face, body and each body part) by choosing options and following the steps, or add your own photos as reference. Then generate her: photos and videos, safe-for-work and explicit, with the exact same face and body every time.

Every photo is conditioned on her approved reference shots, so it is the same woman in every frame rather than a new one each time.

Video works three ways: bring one of her photos to life, make a short reel from a prompt, or put her into a clip you upload.

Photos and video on every plan, Free included. Try it now with 15 free tokens, no account needed.

The route, end to end

  1. Pick what to makeA photo, Photo to Video, a Reel or a Swap. Safe or explicit. The choice filters the models that can do it.
  2. Pick herHer character build, or reference shots from her own library with face and body kept as separate sets.
  3. Describe the shotA scene from the list, your own words on top, and anything to keep out of frame.
  4. Keep what worksResults land unapproved. Keep the good ones, drop the rest, and only kept media can reach a fan.
Nothing unreviewed ever reaches a fan. That is enforced, not a setting.

Make her first photo and video now

Open the studio and start generating straight away. Try safe-for-work photos and videos on 15 free tokens, then create a free account for explicit work and downloads.

Open the Generation Studio

15 free tokens · No account needed to try · Photos and video

What it does

The same woman, in every photo and every clip

Identity is the hard part of this and it is never left to the prompt. Photos are conditioned on her real reference shots, and a video that claims to be her starts from her approved photos.

🧍

Face and body, separately

Two reference sets per persona. A full-body shot is exactly where a face reference alone starts to drift, so the body gets its own.

🎬

Photo to Video

Pick one of her approved photos and it becomes a short clip. The first frame is already her, so there is nothing to correct. Safe or explicit.

🎞️

Reels from a prompt

Describe a scene and get a short safe-for-work clip, or pick her from the character build so the reel is of her.

🔁

Swap her into a clip

Upload a video and she performs it: her face and body, the clip's own motion. Safe for work.

✅

Review before anything ships

Generations land unapproved and are invisible to every send path until you keep them. Unkept media expires on its own.

💳

Priced per generation

Tokens are pegged to what each photo or clip actually costs to run, and the studio shows the price before you submit.

How it works

Generated in three steps

No prompt engineering, no LoRA training, no model to fine-tune.

Build her once

Build her in the character builder, or upload her face and body shots. Every later photo and video uses them.

Set the shot

Photo or video, safe or explicit, size, how many, the scene and your own description.

Keep the good ones

Review the batch, keep what works, and it becomes available to the rest of the platform.

The models

What runs behind it

Different models are good at different things. The rating and the format pick the shortlist so you do not have to.

Seedream 4.5
Every explicit photo. The one model that serves explicit work and takes reference images, which is why it is pinned there.
Seedream 5.0 Pro
Safe for work only. Its provider refuses explicit prompts outright, so it is never offered for them.
Nano Banana Pro and Nano Banana 2
Safe for work only, for the shots where their look is the better one.
Safe-for-work video
Seedance, Kling, MiniMax and Wan, shortlisted by the job: Photo to Video, Reel or Swap.
Explicit video
Photo to Video on a dedicated explicit video model, always from one of her approved photos.
Sizes
Two bands, roughly 2k and 4k, resolved per model because each one accepts its own dimensions.
FAQ

The most common questions

Is it available yet?
Yes, on every plan, Free included. You can also open the studio signed out and try safe-for-work photos and videos with 15 free tokens.
How does it keep the same face?
Every generation is conditioned on her real reference photos, and the face set and the body set are kept apart. Identity is never left to the wording of a prompt.
Can I generate a video from text?
Yes, as a safe-for-work Reel: a prompt alone makes a short clip. A video that is meant to be her is made from her approved photo or her character build, so the face is hers from the first frame.
Can it generate explicit content?
Yes, photos and Photo to Video, with an account. The safe-for-work models refuse it at the provider, so they are filtered out before you can pick them.
Can a generation reach a fan by accident?
No. A new generation is unapproved and invisible to every send path until you keep it, and anything you do not keep expires on its own.
What does it cost?
Tokens, priced from what each generation actually costs to run, so a photo and a five-second video are very different numbers. Paid plans include monthly tokens, any account can top up, and the studio shows the price before you submit.
Stack the channels

One persona, every platform

Each channel feeds the same funnel and the same paid page. Running two costs you no more work than running one.

Photos and videos of a persona who already exists

Her face, her body, a new scene or a new clip, and a review step between the model and your fans.

Part of Velvet Funneler: build the character, generate the content, run the funnel.