FLUX 3
by Black Forest Labs
Black Forest Labs' video model, built for the move between two frames.
What it is
FLUX 3 is described as "Black Forest Labs' frontier video model", and the endpoint in this studio does one thing precisely: it "generates the video between a defined start and end frame, interpolating a smooth, coherent transition from the first image to the last".
That is a narrower job than general video generation, and narrower is the point. You approve two stills, and the model only has to solve the movement between them. There is far less room for it to invent something you did not ask for.
It runs at 720p and 1080p, and the Smooth transitions tool in this studio is built on models like this one.
What it is good at
Two frames you have already approved. The model fills the middle rather than inventing the whole thing.
Interpolating between fixed points is a constrained problem, which is exactly why the result holds together better than open-ended generation.
The same lab behind FLUX.2, so a still made there and a move made here tend to belong together.
Where it struggles
- Black Forest Labs publishes no list of limitations for it, so none is invented here.
- It needs two frames. This is not the model for turning a prompt into a clip , for that, use Veo, Kling, Seedance or Wan.
- The rate rises noticeably from 720p to 1080p. The Studio shows both before you run.
How to prompt it
The two frames carry most of the meaning, so the prompt only has to describe the movement between them: what travels, how fast, and what the camera does while it happens. Describing the scene again competes with the images you already gave it.
An example that shows what that means
The camera drifts slowly to the right and settles, the subject turning her head to follow, everything else in the frame staying still.
What we offer here
Straight from the catalogue, so this list never goes stale. The price is the provider’s published rate with our markup on top, and it is on the button before you run anything.
Questions
Do I need two images?
Yes , a first and a last frame. That constraint is the reason the result is more predictable than open-ended video generation.
Can it make a video from just a prompt?
Not this endpoint. For text to video, use Veo, Kling, Seedance or Wan.
Is this what the transitions tool uses?
The Smooth transitions tool picks a purpose-built transition model from the catalogue, and this is one of the family it chooses from.