10% off monthly plans with code NODE10, until August 31
Build the one minute documentaries that loop across social feeds: cut-paper sets, a new shot every two seconds, generated voice-over. The whole workflow is visible here, no account needed.
Real workflow, real results. Pan and zoom freely.
Use this workflowThis workflow is the real template: the art direction feeds every shot, the Video node renders ten seconds of montage, the voice-over lives in its own node. Pan, zoom, open the nodes.
At a glance
These short cut-paper documentaries work because they keep a simple promise: a story told in a minute, with no face on screen, in a material so recognisable that you name the channel from the first shot. The format breaks down into three parts, and this workflow shows all three: a locked art direction, a hard montage, a calm voice-over.
The hard part is not the subject, it is CONSISTENCY. Describe each shot on its own and the paper changes thickness, the cream turns grey, the light jumps between shots and the series falls apart. Hence the Art direction node: it carries the material, the palette and the light once, and feeds every Image node. You change subject without ever touching the style.
On the model side, Seedream 5 Pro holds the paper material with rare fidelity, and Seedance 2.0 can cut inside a single take, which gives five two-second shots in one render. The voice-over runs through OpenAI TTS HD. The guide on writing a prompt that works covers how to describe a material, and the public calculator shows the exact cost before you sign up.
Lock the material
The Art direction node describes the paper, the palette and the light. Only one warm colour allowed: that constraint is what gives the format its identity, and it applies to every shot.
Write the script
The Text node carries the narration. Count about thirty words per ten seconds of video: that is documentary pace, any denser and the voice runs ahead of the images.
Describe the five shots
One video prompt, five numbered shots, a hard cut every two seconds. Without that explicit breakdown the model renders ten seconds of one continuous move and the format loses what sets it apart.
Chain clips to fill the minute
Duplicate the Video node, change the five shots, keep the same art direction. Six ten-second clips end to end make the one minute documentary.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
The original format: a battle, an invention, a lost city. The Art direction node holds the aged paper and a warm lamp, the script follows the chronology, and each shot shows an object rather than a person. Faces in cut paper remain the weak spot of image models, hands and silhouettes hold up very well.
Same mechanics, commercial subject: what the product does, why it exists, who it is for. Keep the voice-over under thirty words for ten seconds and end on a static shot where a logo can sit at edit time. Cut paper avoids the stock-footage look that kills a brand video.
The water cycle, photosynthesis, a revolution: the format explains without a wall of text. Write the script at the right reading level, then check every image, a model happily gets a diagram wrong. See also the coloring page generator for the worksheet that goes with the video.
Animals are what cut paper renders best: crisp silhouettes, no human faces, layered depth. Ask for a marked depth of field in the art direction, that is what makes the set feel built by hand rather than pasted flat.
Switch the Image nodes and the Video node to 9:16 and lift the action inside the frame: the bottom of the screen is eaten by the interface. Nothing else on the canvas moves. A series of six vertical clips is just the Video node duplicated, with the art direction still wired into all of them.
This is the number one cause of a series drifting: the paper changes thickness, the warm beige turns grey. Style belongs to the Art direction node, image prompts carry only subject and framing.
Past thirty words for ten seconds, the voice-over races ahead of the images and the edit feels rushed. Cut the text, not the shots.
Without numbered shots and explicit hard cuts, Seedance 2.0 renders ten seconds of a single camera move. The format loses exactly what makes it recognisable.
That is the pace of the format: it keeps the eye from drifting and makes up for the absence of a face on screen. Past three seconds a shot, retention drops on mobile.
By going through the Art direction node rather than repeating the style in every prompt. It is wired into every Image node: one edit propagates everywhere and nothing drifts.
Yes, a Voice-over node generates it from the script, with a choice of voice and pace. The montage stays silent so it does not fight the narration, exactly as in the original format where there is no music.
Yes, by duplicating the Video node five times on the same canvas, each with its own five shots and the same art direction wired into all of them. Six ten-second clips end to end give the minute, and the whole thing stays under ten euros of credits.
A low voice, a slow pace, no emphasis: the format lives on restraint. OpenAI TTS HD offers six voices, male and female, with a speed setting. Dropping the speed to 0.9 is often enough to turn a read into narration.
Not from the canvas: the edit exports as a silent MP4 and the voice-over as a separate MP3, which leaves subtitling to your usual editor. That split is also what lets you redo the voice without paying for the video again.
A graphic style belongs to nobody, and the images you generate are yours, commercial use included. Just avoid naming an existing channel in the prompt: that is what produces renders too close to identifiable work.
AI documentary video in cut-paper style
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.
Seedream 5 Pro
OpenAI TTS HD