NodeCanvas

When to use NodeCanvas

Open NodeCanvas when the video is still taking shape. This is where you decide what a shot should look like, which model should generate it, and how one asset should feed the next step.

The main value is context. Prompts, references, generated results, and publish destinations stay on one board, so you can see what created each clip and what should happen next.

How the canvas works

Each node owns one job. Each connection shows which result becomes the next input.

• Text nodes hold prompts and assist-generated drafts.
• Image nodes turn text or references into key visuals you can review and reuse.
• Video nodes turn prompts and/or start frames into short clips.
• Edit nodes send selected clips into the timeline editor when pacing and overlays matter.
• Publish nodes keep the upload step attached to the exact clip you plan to release.

The nodes you use most

Most projects start with a small set of building blocks, even when the graph becomes more complex later:

• Text — define the shot, angle, tone, offer, or motion cue in words.
• Image — lock the visual direction before you spend credits on motion.
• Video — test one or more animation directions from the same prompt or start frame.
• Edit — combine clips, captions, and overlays once you know which shots should stay.
• TTS and publish nodes — add narration or send the final clip into YouTube or Instagram without breaking context.

How work moves downstream

A common path is text to image to video to edit or publish, but the graph stays flexible. Connect text directly to video for a prompt-only run, or connect an image to video when the first frame should guide the motion.

One output can feed several downstream nodes. One key visual can drive multiple video attempts, and one generated clip can feed both an edit branch and a direct-publish branch.

Model and shot settings

Configure each video node per shot. Choose the model, duration, aspect ratio, resolution, and whether the clip starts from text alone or from a provided image.

• Veo is suited to crisp short clips when you want reliable default motion and clear output tiers.
• Kling is useful when a start frame is doing most of the creative work and motion should preserve it.
• Seedance gives the widest aspect-ratio range when delivery format matters as much as style.
• Kling Omni is the best fit when several reference images need to steer one visual direction.

After generation

Run only the node you need next. Review an image before animating it, then review the clip before spending time on editing or publishing. Credit cost appears at the node level, so each action is explicit before you run it.

When a clip is ready for editing, move it to VideoEditor for pacing, captions, overlays, and audio. When it is ready to publish, connect it to YouTube or Instagram nodes and track performance from the same canvas.

Typical workflow

1. 1Start a projectOpen a blank NodeCanvas project or resume an existing one from the canvas home. This is where you decide the structure of the job before you work on the final edit.
2. 2Define the first inputAdd a text node, a reference image, or both. Use assist mode when the idea needs a clearer first draft before generation.
3. 3Generate a key visualGenerate one or more image options and use them to lock style, framing, and subject direction before moving into motion.
4. 4Generate a video clipFeed the chosen prompt or image into a video node, pick the model that fits the shot, and generate the clip you want to test next.
5. 5Branch, compare, or rerunSend one result into several downstream nodes if you want alternatives, or rerun only the node that needs another take without rebuilding the whole flow.
6. 6Send the result forwardMove the selected clip into VideoEditor when it needs timeline work, or connect it to a publish node when it is ready to publish from the canvas.

Frequently asked questions