Text to Video Workflow
Use text to video AI on the Astorie canvas to turn a written scene prompt into a short video clip. Write a clear subject and action brief in a Text node, connect it to a Video node to trigger Text to Video mode and inspect the generated shot from first frame to last.
When you start from text alone, the video model has to invent the subject, setting, lighting and motion at the same time. Comparing an overloaded prompt with a focused Text-to-Video node setup shows why one clear action per shot produces more usable clips.
Start on the Astorie canvas with a single visual moment you can describe in two or three concrete sentences.
Add a Text node on the canvas and describe what the viewer should see: the main subject, the environment, the lighting and one continuous action or camera move. Keep the description concrete and visual.
Add a Video node and connect the Text node output to its input port. In the documented Video node workflow, a Text connection sets the node to Text to Video generation mode.
Choose a video model in the Video node that supports text-to-video input, review the available settings and click Generate to run the shot on the canvas.
Play the generated video from start to finish. Check whether the subject, motion and framing hold together across the entire duration, then revise the Text node or lock a reference still if the next iteration needs tighter visual control.
Inspect three details across the full duration of every text-to-video take before adding it to a project sequence.
Watch hands, faces, moving limbs and held objects throughout the motion. A shot that looks sharp on the opening frame can still morph or lose structure halfway through the action.
Check that the camera pan, push-in or static framing behaves smoothly and that background architecture or horizon lines do not warp as the subject moves.
Step through the first and last frames of the clip to confirm you have a clean entry and exit point for your edit, or capture a strong frame with Frame Extraction to anchor follow-up shots.
Use this summary to plan text-driven video generation and know when to add image references.
| Input | A written scene prompt in a Text node connected to a Video node on the Astorie canvas. | |
|---|---|---|
| Output | A generated video clip on the canvas for playback inspection, model comparison or editorial export. | |
| Review | Check prompt adherence, subject stability across motion, background consistency and usable edit handles. | |
| Plans and credits | Video generation consumes workspace credits according to the selected model and run settings; see the pricing page for plan details. | |
| Rights and workflow | When a scene requires strict character or product identity, generate or upload a reference image first and use Image to Video mode. |
Explore documented Video node modes and adjacent tools when moving from early text concepts to locked production shots.
Review how Text to Video, Image to Video, Video Editing and Multi-Image Reference modes are triggered on the canvas.
Explore reference-image video generation when you want to anchor the starting frame before adding motion.
Capture a still frame from a promising text-generated clip to guide downstream shots.
In the documented Video node workflow, connecting a Text node to a Video node triggers Text to Video mode. When you click Generate on the Video node, it reads the connected Text node prompt to generate the clip.
Use text to video AI when you are exploring early scene concepts, mood directions or motion ideas and do not yet have a locked reference image. Once you need a specific character face, product design or set layout to stay consistent, start from an approved still in an Image node instead.
Video models handle one clear action and camera direction much more reliably than a dense paragraph of sequential events. Simplify the prompt to a single beat, or split a multi-step action into separate shots on the canvas.
Yes. Because the prompt lives in a Text node on the canvas, you can connect that Text node to more than one Video node, select supported models and compare the generated takes side by side in the same workspace.
Open the Astorie canvas, connect a Text node to a Video node and test one clear action brief.
