Setting a First Frame and Last Frame gives the video model a clear starting point and a clear destination — it fills in everything in between. This is the most reliable way to get a cinematic camera move, a controlled transition, or continuity between characters, props, and environments across a sequence.
How it works
Alongside your video prompt — which still matters for describing what happens — First Frame and Last Frame lock in the visual tone, setting, and pacing. Think of them as guideposts, not final stills: the model interprets them to shape the motion, it doesn't just fade between them.
Setting your frames
From the video generation form (click Generate video on the bottom toolbar):
Pick a video model. First and last frame support varies by model — Layer surfaces this on the form itself.
Add your First Frame and, optionally, your Last Frame. You can upload an image, pick one from your Library, or use an asset already on the board.
Write your prompt describing the motion and camera work between the two frames.
Generate.
💡 You can also do this conversationally: attach a starting image and an ending image in chat and ask the Agent to "animate a transition from the first image to the second."
Building longer, connected sequences
First/Last Frame is especially powerful for chaining clips together:
Generate your first video clip.
Extract the last frame of that clip (see Creating consistent videos with Select Video Frame).
Use that extracted frame as the First Frame of your next generation.
This gives each new clip a strong, visually consistent starting point, so longer sequences read as one continuous scene instead of disconnected cuts.
What's next
The full video generation walkthrough → Generating video on Layer
Extract a frame from any video → Creating consistent videos with Select Video Frame
Drive generation directly → Generate and edit assets directly