Connected Video Workflow
Build a connected AI video workflow on the Astorie canvas where reference images, shot generation, video edits and voiceover stay in one visual project. Wire compatible nodes step by step, inspect each take before moving downstream and prepare approved clips for your final edit.
Video production rarely succeeds as a single isolated prompt. Comparing disconnected one-off clip generations with a stage-by-stage AI video workflow shows how keeping references, motion takes and audio cues on one canvas makes revisions predictable.
Start with a clear target sequence and add only the nodes needed for your next verified production step.
Create or upload a reference frame in an Image node when you want to anchor character and setting details, or draft a scene prompt in a Text node for a text-driven shot. Inspect and approve that starting input before adding motion.
Add a Video node and connect your approved upstream node. In the documented Video node workflow, a Text connection triggers Text to Video, a single Image connection triggers Image to Video, multiple Image connections support Multi-Image Reference and a Video connection triggers Video Editing.
Run the Video node and review the full clip for subject stability, camera movement and clean start and end frames. Branch a second take from the same reference node when you want to compare motion directions without losing your original setup.
Use an Audio node to generate voiceover or sound effects from text entered directly inside the Audio node, or connect a Video node when generating video-driven audio. Once your video and audio selects are approved, export them for timeline assembly in your NLE.
Three practical rules keep a multi-node video project readable and prevent wasted generation runs.
Every connection on the canvas follows the input types accepted by the destination node. Check what the target Video or Audio node supports before wiring a port rather than assuming any node can feed any other node.
Video nodes take visual or text prompts for motion, while speech and sound-effect prompts are entered inside the Audio node's own text input. Keeping picture generation and narration in dedicated nodes lets you revise a line of script without regenerating the camera shot.
Downstream nodes retrieve upstream content when you click Generate on that downstream stage. Changing an early prompt or reference image does not automatically regenerate every connected clip, so you stay in control of which shots are rerun.
Check how inputs, node connections and finishing steps divide across the canvas and your editing software.
| Input | Text prompts, permitted reference images or source video clips arranged in compatible Text, Image, Video and Audio nodes. | |
|---|---|---|
| Output | Reviewed video takes and optional audio cues organized by stage on the canvas for downstream use or editorial handoff. | |
| Review | Inspect node connection types, subject continuity, motion stability and audio timing at each stage before running the next node. | |
| Plans and credits | Each node generation uses workspace credits based on the selected model and run settings; see the pricing page for plan details. | |
| Rights and finishing | Final timeline cutting, multi-track mixing, color grading and delivery mastering take place in your external NLE. |
Explore the documented node guides and adjacent tools that support a multi-stage video project.
Learn how nodes, ports and upstream-to-downstream execution work on the Astorie canvas.
Check the trigger conditions for Text to Video, Image to Video, Video Editing and Multi-Image Reference.
See how Text to Speech, Sound Effects, Music and Video to Audio work inside an Audio node.
Prepare approved canvas takes for final assembly in DaVinci Resolve, Premiere Pro or another editor.
A single prompt asks one model to invent the look, subject and camera move all at once. A canvas video workflow separates those decisions into inspectable stages so you can approve a reference image first, generate motion in a connected Video node, add narration in an Audio node and compare takes side by side.
According to the Audio node documentation, text for Text to Speech, Sound Effects, Music and Voice Design is entered directly inside the Audio node rather than through an upstream Text node connection. An Audio node accepts an upstream connection from a Video node when using Video to Audio mode.
No. When you update an upstream Text or Image node, downstream nodes do not automatically re-execute on their own. You decide which downstream Video node to click Generate on so you only spend credits on shots you actually want to replace.
Use the canvas to develop, compare and organize the individual video shots, reference frames and audio elements. Bring your approved files into a dedicated nonlinear editor for frame-accurate cutting, dialogue mixing, color grading and final delivery.
Start with one reference still or scene prompt on the canvas, connect a Video node and review the first shot before adding the next stage.
