AI video generation on an infinite canvas
Generate video from text, from an image, from a first and last frame, or from a reference clip. Every run is a node on the canvas you can rewire, adjust and run again. Upscaling, trimming, subtitle removal, translation and multi-track editing live on the same canvas, so a shot never has to leave for another tool.
What you can do
Four ways to generate
Text to video by default. Connect one image for image-to-video, two for first/last-frame control, or feed a reference clip. Every mode is the same node, so you can switch without rebuilding the graph.
Post-processing on the node
Upscale to a higher resolution, trim in and out points, split video and audio, mix or replace an audio track, separate vocals, remove subtitles, and translate with either subtitles or AI dubbing.
Lens combo
Pick camera body, lens, focal length and aperture from four wheels in the generator bar, like setting up an old camera. They ride along as prompt fragments for the next run only, and are not billed separately.
Multi-track timeline editor
Open a video node in the timeline editor for stitching, overlays, subtitles and transitions across several tracks. Export is free and lands back on the canvas or on your machine.
Shot breakdown
Run video analysis to get a structured breakdown — shot size, camera movement and more — so an existing clip can drive what you generate next.
How it works
Open a canvas
Sign in and create a canvas. Everything below happens on that one surface, and nothing gets lost between tools.
Add a video node
From the Add node menu pick Video. Or use the + beside an image node and expand the Video category to get an image-to-video node with the connection already made.
Describe the shot and generate
Write the prompt and set the options the chosen model supports. The generator bar quotes the credit cost before you submit; progress then shows on the node itself.
Refine, edit and export
Run the node again, apply the editing tools, or open the timeline editor. Results become assets automatically, ready to reuse on another canvas or export.
Common questions
What resolutions and durations can I generate?
Available resolutions and durations come from the runtime model catalog and differ per model. The generator bar only offers what the selected model actually supports, and the quote updates as you change them. Video upscaling separately supports 720p, 1080p, 2K and 4K.
How is video work billed?
Every generation is quoted before you submit, for the exact settings you picked. Editing tools differ: trimming, splitting video and audio, mixing audio, capturing the first and last frame, and vocal separation are free; upscaling, video analysis, subtitle removal and translation are paid. Timeline export is free.
What happens if a generation fails?
Credits deducted for a failed task are returned automatically, entry by entry, to the batches they came from. Where an original batch has already expired, an equal amount is granted as compensation credits valid for 24 hours.
Can I work with my own footage?
Yes. Upload, drag or paste video and images onto the canvas as asset nodes, then wire them into generation, reference or editing nodes. Results are stored as assets you can reuse across canvases and export.