AI Video Upscaler+
Help & Reference
Get it from Microsoft
×

Help & Reference

What every option does, which GPUs are supported, where your files go, and how AI Video Upscaler+ handles your data.

Processing Options Explained

Target Resolution

The output resolution you're upscaling to — 1440p, 2K, 4K or 8K. The app can add real, AI-generated detail up to roughly 4× the source's native resolution; going higher automatically switches to a tiled processing path so very large targets stay stable.

Target Frame Rate

Raises the video's fps — for example 30 fps to 45 or 60 fps — by generating new, genuinely interpolated in-between frames with AI, rather than simply duplicating existing ones. Leave it unchanged to only upscale resolution.

Face Restoration (GFPGAN)

Applies an AI face-restoration pass to every frame for noticeably sharper, cleaner faces. Adds processing time, and is most worthwhile on videos where faces are a focal point.

GPU Device

Choose which graphics card handles the processing, if your PC has more than one — for example an integrated GPU and a dedicated one.

Video Codec

Auto (default) uses H.264 for most videos, but switches to HEVC (H.265) automatically for very large outputs (8K and above), since H.264 decoding often isn't reliably supported at those resolutions. You can also force Always H.264 for maximum player compatibility, or Always HEVC for smaller files at high resolution.

GPU Support

AI Video Upscaler+ runs its AI models through Microsoft's DirectML and ONNX Runtime — the same official Windows AI acceleration path used across GPU vendors. In practice this means:

  • Any GPU that supports DirectX 12 (D3D12) can run the app — including NVIDIA, AMD, and Intel graphics.
  • No vendor-specific SDK (like CUDA) is required — you don't need an NVIDIA card specifically.
  • The app also runs on lower-end PCs, including ones with only integrated graphics — processing simply takes longer on less powerful hardware.

Accessing Your Video Library

Every video you upscale is saved automatically and listed in the Projects tab inside the app, along with the settings used to create it. Click Refresh there to load your history, or use Open Folder after any job to jump straight to the file on disk.

Finished videos are saved locally to C:\Users\[you]\Videos\AI Video Upscaler+\ and never uploaded anywhere.

Privacy & Data

AI Video Upscaler+ processes everything locally on your machine. No video is sent to any cloud or external server.
  • Source videos are read directly from wherever you selected them — never copied to us.
  • Temporary working frames are written to a local working folder and deleted once a job finishes.
  • All AI models ship inside the app itself — nothing further is downloaded to process a video, and the app works fully offline.
  • The only network activity is a periodic check for a local connection certificate — no analytics, no telemetry.

Frequently Asked Questions

Does AI Video Upscaler+ need an internet connection?
No — every AI model is bundled with the app. You can disconnect entirely and processing still works exactly the same.
Do I need an NVIDIA graphics card?
No. Any GPU that supports Microsoft's DirectML — NVIDIA, AMD, or Intel — is supported. The app also runs on integrated graphics, just more slowly.
Why did my upscaled 8K video not play in my video player?
Some players and hardware decoders don't reliably support H.264 at 8K. Set Video codec to Auto or Always HEVC in Settings — HEVC has far broader real-world 8K playback support.
Is AI Video Upscaler+ really free?
Yes — the full app, with no feature restrictions, no watermark, and no subscription.