SRStoprekt StudioSeedance generation workspace

Checking fal.ai route... Back to dashboard

Image to video

Move one still frame like a real shot.

This rebuild matches the live backend you actually have: a Seedance flow through fal.ai with one start frame, an optional end frame, motion direction, queue polling, and downloadable video output.

Build a shot What works here

5-10sClip lengths exposed by the current server flow.

Start + endOptional destination frame for morphs and guided reveals.

fal.ai liveBuilt around the route you already paid for and can use now.

1. Frame input

Use a strong source image. Add an ending frame only if you want the motion to resolve toward a specific final look.

Visual setup

Primary frame Drop the opening image

PNG, JPG, or WebP up to 10 MB. This becomes the anchor for scene identity, camera language, and subject motion.

PortraitsProductsFashionScenes

Start frame readyReady to render.

Optional ending frameUse this for morphs, reveals, or a clear destination pose.

Ending frame ready

Important: this page does not take uploaded audio input. The audio toggle below only asks the model to generate soundtrack or ambience.

2. Motion direction

Strong prompts describe movement, not just style. Tell it how the subject, camera, atmosphere, and final frame should behave.

Prompting

Motion prompt

Best prompts mention camera move, subject move, wind, fabric, parallax, reveal, drift, morph, or what should happen while reaching the ending frame.

3. Delivery settings

Start balanced, then refine after you like the motion language.

Control deck

Resolution

Duration

Aspect ratio

Seed (optional)

Generate soundtrackModel-generated audio only, not uploaded reference sound.

Lock cameraUse this if you want subject motion but a steadier camera.

Advanced controlsOptional

Enable safety checkerOff by default in the current backend. Turn on if you want stricter screening.

4. Render

The page queues the request, polls your server, and upgrades the result panel as soon as the final video URL lands.

Preparing requestWaiting for the generation request to be accepted.

No request yet

Frame readyInputs are set.

QueuedBackend accepted the request.

RenderingPolling for video.

DeliveredClip is ready.

IdleQueue position and render state will appear here once a request is created.

Result

The output panel stays on-page while the render runs and flips into playback when the video is ready.

Output

No clip yet.Once you launch the render, the finished video will appear here with a direct download action.

Shot setupNo generation yet.

Prompt summaryNothing submitted yet.

Queue resultWaiting for your first run.

Request IDRequest ID will appear here.

What works on this page now

Start image to moving clipThe core flow stays wired to the live fal.ai-backed Seedance route.

Optional end frameThis rebuild exposes the destination-frame option already supported by your backend.

Clearer failure statesFile-size validation, timeout handling, config visibility, and explicit queue/render states are now part of the UI.

Honest capability copyThe page no longer pretends it already does uploaded audio-driven generation when it does not.

Direct Seedance later? Fine if official access opens up cleanly for you. Right now this page is rebuilt around the thing you already own and can actually run: fal.ai-backed generation.