AI Video Upscaler: Current Availability

Image upscaling is available now. Video upscaling is not currently supported.

Upscale an image

What is an AI video upscaler?

A video upscaler increases the resolution of moving footage while keeping frames consistent. That requires a video-processing pipeline. The models linked here accept still images, so they are an image-upscaling alternative, not a video upscaler. You can upload a photo in the workflow editor, choose an available image-upscaling model, review the credit cost, and download the resulting image.

Models behind this tool

Available models and their credit costs are read from the live catalogue. Review the selected model and settings in the tool before generating.

  • Image Upscale (SeedVR)10 credits

What people use it for

Larger still images

Increase the size of an existing photo or generated image.

Portrait and product detail

Recover skin texture, fabric weave and fine edges that a general upscaler smooths away.

Archive restoration

Enlarge an older low-resolution photo, then inspect the result for artifacts.

How it works

  1. Match the model to the subject

    Aura SR for portraits, fur and fine fabric. Clarity for general detail. SeedVR when you need an exact target resolution.

  2. Pick a scale or a target

    Aura SR is fixed at 4x. Clarity is configurable 2–4x. SeedVR takes a target resolution from 720p to 2160p.

  3. Check edges before scaling further

    Upscaling twice compounds artefacts. One well-chosen pass beats two generic ones.

Frequently asked questions

Can I upload a video to this workflow?
No. The current workflow accepts still images and returns images. Video upscaling is not available here.
Which upscaler should I use for portraits?
Aura SR. It is tuned for fine structures — skin texture, hair, fur, fabric patterns — where Clarity tends to smooth detail away. It is fixed at 4x.
Can I upscale to a specific resolution rather than a multiplier?
Yes — SeedVR takes a target resolution from 720p to 2160p, which is easier to reason about when you have a fixed delivery spec.
Is it better to render at high resolution or upscale afterwards?
Render high when the model supports it natively — MiniMax H3 outputs 2K and VEO 3.1 and LTX 2.3 reach 4K. Upscale when the source already exists or when native high-resolution rendering costs more than a render plus an upscale.

Models used here

Upscale an image
AI Video Upscaler: Current Availability | Floniks