Upscale, de-noise, and sharpen videos & photos with AI

Right Image Left Image

Resize, sharpen, and de-noise videos and photos

  • Batch process video and image files with drag and drop
  • Increase video resolution 2x, e.g., 1080p to 4K, 4K to 8K, or 8K to 16K
  • Handles very high resolution videos, e.g., 16K output in ProRes
  • Input .mp4, .mov, .mkv, .png, .jpg, .gif
  • Input any size of video
  • Encode output video as h264, h265, ProRes, or 16-bit PNG sequence
  • Output 10-bit ProRes (Windows/Mac) or 10-bit h265 (Mac only)
  • Enhance videos or photos
  • Content-adaptive de-noising and de-blurring / sharpening
  • Reduce aliasing and compression artifacts
  • Scale AI generated videos to 4K
  • De-noise with bilateral filter for photos with high grain

Comparison: UpscaleVideo.ai vs. Topaz Video AI

Topaz Video AI 6.0.3 Artemis model, UpscaleVideo.ai 1.15 UNet (de-noise, de-blur).
2x upscale (super resolution), default settings for both.

Comparison: UpscaleVideo.ai vs. Adobe Premiere Pro

Upscale 2x (bicubic) in Adobe Premiere Pro vs. super resolution, de-noise and deblur in UpscaleVideo.ai.

Tools for 3D, VR, and stereoscopic video

  • Adjust stereo baseline / interpupilary distance (IPD) for VR180 or rectilinear 3D video and photos
  • Correct IPD contributes to comfort and perception of scale, but is not provided by even the most expensive VR cameras such as the Canon R5 with dual fisheye lens
  • Amplify or compress perceived depth for artistic effect or comfort
  • View interpolation with RAFT optical flow and forward splatting

Illustration of varying IPD from 0.5x to 1.5x in a VR180 photo captured with a Canon R5 with dual fisheye lens. Photo by Thomas Hübner.

Download and Install UpscaleVideo.ai for Windows

  • Download UpscaleVideo.ai 1.15 Installer for Windows 11
  • Requires Windows 11, NVidia GPU (3090+ recommended)
  • You may get a warning while installing that says "Windows protect your PC ... prevented unrecognized app from starting". To get past this, just click "More Info", then "Run Anyway".

Download and Install UpscaleVideo.ai for Mac OS X

Pricing

Free Trial
No purchase required		 Non-Commercial
$129 one time purchase		 Pro License - 1 Seat
$29.99/month
Upscale video or photos using AI on Windows and Mac ✔ ✔ ✔
Export with no 'upscalevideo.ai' watermark ✔ ✔
Non-commercial use only ✔ ✔
Full commercial license ✔
Updates for current version (1.x) ✔ ✔
Updates for as long as you subscribe ✔
Refund within 30 days of purchase for any reason ✔ ✔
Download Buy Buy

Updates

  • 1.15: Text in the user interface is more clear and readable.
  • 1.14: Improved user interface layout.
  • 1.13: Added bilateral filter for denoising, added jpg export.
  • 1.12: Fixed a bug where an error was shown when processing images.
  • 1.11: Added checkbox to control whether audio and other data streams are copied.
  • 1.10: Fixed a crash when transcoding with incompatible audio streams.
  • 1.09: Added capability to adjust stereoscopic baseline with optical flow. Fixed a bug where it crashed when loading videos of certain sizes.
  • 1.08: Batch process videos and images with drag and drop. Audio and other additional streams are now transcoded when video containers allow. Fixed several bugs.
  • 1.07: There is now an option to select between EDSR (the original super resolution model) and a new UNet based model which is roughly 2x-5x faster with only a small reduction in PSNR.
  • 1.06: The app and installer are now "hardened, signed, notarized, and stapled" on Mac, so there should not be a Gatekeeper warning when installing. Fixed a bug where video processing stalled (both Mac and Windows).
  • 1.05: Added options for encoding ProRes 422LT, 422HQ, or 4444. Fixed a bug where processing a video would not finish. Fixed a bug where crf was not used when encoding h265.
  • 1.04: Video transcoding and super resolution engine is now pipelined and multi-threaded to improve GPU utilization. Added setting for number of super resolution threads. Fixed bug where quitting in Mac OS X gave an error.
  • 1.03: Installing ffmpeg is no longer required on Mac. Videos are now decoded in 10-bit depth whenever possible. Videos are now encoded in 10-bit depth with h265 (Mac only) or ProRes (Windows or Mac).
  • 1.02: Added de-noising / de-blurring model. Super resolution model is faster. Fixed bug where h265 exports didn't open.
  • 1.01: Fixed a bug where output frame rate did not match input frame rate. Added support for Intel Macs.

Frequently Asked Questions

  • Q: Do you require any personal information or an email to download the free trial?
    A: No, you just click the link, like back in the good old days.
  • Q: I have questions, or need to modify my subscription. How do I contact the developers?
    A: Email support@lifecastvr.com
  • Q: What super resolution model is this?
    A: We developed and trained our super resolution models from scratch. This means the results have a unique appearance which is not the same as any other software, much like different lenses create different image styles. One of our models is a deep convolutional neural net with 30+ layers, inspired by EDSR (Lim, Bee, et al. "Enhanced deep residual networks for single image super-resolution." Proceedings of the IEEE conference on computer vision and pattern recognition workshops. 2017), but trained on our data. Another is a UNet architecture that we selected for efficiency.
  • Q: On Apple devices, does this run on the CPU or GPU?
    A: It runs on the GPU (Metal) on Apple Silicon (M1, M2, M3, M4).
  • Q: Who makes UpscaleVideo.ai? Do you make any other software?
    A: UpscaleVideo.ai is made by lifecast.ai, creators of Volumetric Video Editor, and Volurama. Lifecast Incorporated was founded in 2021, and is funded by Y Combinator.