All tasks
Task·video

AI Video Upscaling Models & Workflows

Sharpen, denoise, and upscale footage. Choose an enhancement model, wire it into a NodeTool workflow, and lift low-res clips to a cleaner resolution.

Script to ScreenEdit a Still with WordsPhoto Enhancement SuiteProduct Mockup Generator

Models for video upscaling

Topaz Video AI
video
Topaz

Detail-preserving upscaling and frame interpolation.

Real-ESRGAN
image
Local

Open-weight upscaler, applied per frame.

SeedVR2
video
ByteDance

Diffusion-based video restoration and upscaling.

A workflow that does it

Workflow Editor
Note
String Input
A lighthouse keeper discovers the beam of her lamp has started bending toward something beneath the waves — and tonight it refuses to point…
String Input
moody coastal nocturne, sodium-lamp amber against deep blue-black sea, drifting fog, anamorphic lens flares, fine film grain, painterly rea…
String Input
5
Prompt
Brief
Style
Count
You are directing a short film. Write a complete DIRECTION DOCUMENT for this brief. Brief: {{ BRIEF }} Visual style: {{ STYLE }} Shot count…
Agent
Prompt
Text
gpt-5-mini
Output
Structured Output Generator
Context
Style Frame Prompt
Narration
Music Prompt
gpt-5-mini
Text To Image
Prompt
fal-ai/flux/schnell
Prompt
Direction
Count
From the direction document below, output exactly {{ COUNT }} lines, one per shot from the SHOT LIST, in order. No numbering, no labels, no…
List Generator
Prompt
gpt-5-mini
Prompt
Shot
Style
Cinematic film still, 16:9. Shot: {{ SHOT }} Visual style: {{ STYLE }}. Dramatic volumetric lighting, rich cinematic color grading, fine fi…
Image To Image
Image
Prompt
fal-ai/flux/schnell
Collect
Input Item
Output
Prompt
Shot
Animate this exact scene. {{ SHOT }} Follow the MOTION direction precisely: move only what it names, keep everything else stable. Preserve…
Image To Video
Image
Prompt
veo-3.1-generate-preview
Collect
Input Item
Concat
Clips
Text To Speech
Text
Audio
tts-1
Text To Music
Prompt
Audio
meta/musicgen
Add Audio
Video
Audio
Add Audio
Video
Audio
Output

Frequently asked questions

How does AI video upscaling work?

The workflow splits the clip into frames, upscales each with an enhancement model, then reassembles the frames back into video.

Can I upscale and color-correct at once?

Yes — chain an upscaling node with color and sharpening nodes in the same graph so each frame gets the full treatment.

What resolution can I reach?

That's set by the model and your prompt. Most upscalers target 2×–4× the source resolution.

Do video upscaling on your own machine

Free, open source, and yours to run. Download Studio, pick a model, and wire it into a workflow with your own keys.