Begin with movement
A dance, an animation, a shifting geometric system. Your source gives the piece its movement and composition; diffusion opens up a new visual direction.
Avant-Garde · Video to video
Your visuals.
New realities.
Oscilloscope Diffusion brings an artistic treatment to the movement you already created. Turn reactive lines and shifting forms into sculptural, painterly or material studies—with source geometry as your guide.
Choose your source, shape the treatment, and see your visuals become anything.
Structure meets interpretation
Keep the movement. Shape the treatment.
Choose how far to go
Keep more of the original or push toward abstraction. Adjust the strength of the treatment, and change it across the video when the moment calls for it.
Find your visual language
Describe materials, colors and atmosphere, or start with a shared prompt. Explore curated styles and save a treatment you want to return to.
Made with Oscilloscope Diffusion
One system. Different worlds.
Explore Uisato’s renders. Play a piece with sound to experience the relationship between movement, texture and music.
The workflow
From source to a new visual world.
01
Choose a moment
Upload your source, inspect its frames and select the section you want to transform.
02
Find the treatment
Describe a look or choose a prompt from the library. Start with the defaults, then explore the creative controls at your own pace.
03
Set your budget
Review the estimate and maximum credit reservation before starting. You stay in control of the approved budget.
04
Review the movement
Renders take hours and keep running if you close the tab. Watch and download the result from My Creations, or reopen the settings to create a new run.
Made by Uisato
Start with a creative system.
Make your own geometry in the software you love, or explore Uisato’s ready-to-use TouchDesigner systems as a starting point.
Before you begin
Questions, answered.
What does Oscilloscope Diffusion actually do?
It reinterprets the frames of a video through AI diffusion. You supply the movement and describe a visual treatment: paper, sculpture, paint, fabric, or something harder to name. The result is a new version of your source, guided by its shapes and motion rather than generated from a prompt alone.
What kind of video can I use?
Live-action footage, animation, 3D renders, abstract loops and generative visuals can all be starting points. This version of the system is especially curated for complex audio-reactive geometry: the moving lines and shapes become the structure for a new artistic treatment.
What are the video requirements?
MP4, MOV or WebM files up to 100 MB. Each render transforms up to 1,800 consecutive frames (about 30 seconds at 60 fps, or 60 seconds at 30 fps); at higher resolutions fewer frames fit in the 23-hour render limit, and the workspace tells you before you start. Output goes up to 1,024 pixels on the longest side and follows your video’s aspect ratio. The original audio of the selected section can be kept.
Do I need TouchDesigner?
No. Bring video made with any software, or footage you recorded. To explore audio-reactive geometry, you can build your own systems or try TouchDesigner’s free Non-Commercial edition for personal, non-commercial work, then explore Uisato’s ready-to-use creative systems. Project files and access details are linked there.
Do I need to understand AI models to get started?
No. Choose a short section of video, describe the look you want or select a prompt from the library, and begin with the default settings. The advanced controls are there when you want more precise direction; you do not need to adjust every setting.
Will it keep the movement in my original video?
The source guides the movement and composition, but diffusion is an artistic interpretation. Stronger treatments can change shapes and introduce new details. Start with a short section to find the balance between recognizable structure and transformation.
Can I keep my music or original sound?
Yes. Keep source audio includes the sound from the selected section when your video has an audio track. The source supplies the rhythm: a video whose geometry already reacts to music is a particularly useful starting point.
How long does rendering take, and what does it cost?
Diffusion is frame-by-frame work, so renders take hours rather than minutes. As a reference at current prices, with the default settings at 512 px each frame takes about 35 seconds of processing, roughly 4 credits: 10 seconds of video at 30 fps (300 frames) takes about 3 hours and roughly 1,200 credits. Larger outputs take longer (about 2.3× at 768 px and 4× at 1,024 px), and fewer steps make it faster. A single render can run for up to 23 hours. Before starting, the workspace shows the estimate for your exact settings and you approve a maximum credit budget. Processing time plus a fixed startup charge is taken from that reservation and unused credits are returned. If the budget runs out, the run stops and returns a partial video from completed frames.
Do I need to keep the page open while it renders?
No. Rendering runs on our servers. You can close the tab and come back later: progress, the finished video and your settings stay in the project under My Creations.
Will my video look exactly like these examples?
Each source, prompt and treatment creates a different result. The gallery shows creative possibilities. Try a short section first, then adjust the style and strength to find a direction that works with your footage.
Open Oscilloscope Diffusion
The in-app Quick Start Guide walks through every stage and its controls.