LabScale logoLabScale · by Labstory Studio

Small photos & videos made bigger and sharper, video smoother.

AI upscaling up to 4× and a 60 FPS boost for photos and video — all processed on your own computer. Drop in a file or a folder, pick what you want, press Process.

FreeOpen source · MITmacOSWindowsNo account

Version 2.1.1 · Mac Apple Silicon (M1+) · Windows 10/11 64-bit · check specs · how to install

A 400×250 graphic poster upscaled 4× with LabScale
400×250 design upscaled 4× · drag to compare
FreeNo account, no watermark, no limit on how many files.
PrivateEverything happens on your computer. Nothing is uploaded, and your originals are never changed.
Photos and videoEnlarge up to 4×, and make choppy video smoother at 60 FPS.
Many at onceDrop in a whole folder, then compare before and after with a draggable divider.

See the difference

A small 400×250 pixel image enlarged 4×. Left: plain enlargement (bicubic). Right: untouched LabScale output — like the design example above (Anime & manga model). Drag the divider.

A 400×250 photo of mossy granite texture upscaled 4× with LabScale
Photo · Realistic photo model · 4×

Samples made specifically for this page with LabScale’s bundled models (fine for commercial use).

Features

Upscale 1×–4×Choose by image type: photo, design or illustration, anime, or anime video. There’s also a Quick draft mode and a plain non-AI option.
Smoother 60 FPS videoThe in-between frames are made by the RIFE AI model when your GPU supports it, otherwise by FFmpeg’s built-in interpolation — choppy motion looks smoother.
A whole folder at onceEvery file shows as a thumbnail with its own progress bar. Results are collected in a <folder name>_upscaled folder.
Compare before / afterClick any file to compare the original and the result at full window width, with a draggable divider.
Safe to cancelStop at any time. Disk space is checked before AI video jobs start, not discovered full halfway through.
Indonesian & EnglishSwitch language in one click. Videos are saved as H.264, HEVC or AV1 with hardware encoding when available (NVIDIA NVENC on Windows, VideoToolbox on Mac), and audio is kept.
LabScale’s compare view: original on the left, upscaled result on the right, with Original / Compare / Result options
Compare the original and the result right inside LabScale.

Can your computer run it?

The AI features (AI upscaling and RIFE 60 FPS video) run on the GPU through Vulkan. Without a supported GPU, LabScale still works for non-AI upscaling and FFmpeg 60 FPS — both run on the CPU.

Mac

Minimum for AI
Apple M1, macOS 14, 8 GB RAM
Recommended
Apple M1 or newer, 16 GB RAM for video
Not supported
Intel Macs

Windows 10 / 11 (64-bit)

Minimum for AI
A GPU with Vulkan, 2 GB VRAM, 8 GB RAM
Recommended
GeForce GTX 1650 / Radeon RX 570 class or better, 4 GB VRAM, 16 GB RAM
Without Vulkan
Basic features only: non-AI upscaling, FFmpeg 60 FPS (slow on old CPUs)
Windows GPUExamples
Supports VulkanIntel HD/UHD 6th gen (2015) or newer · NVIDIA GeForce GTX 650 / GT 740M (Kepler) or newer · AMD Radeon HD 7000 / R7 / R9 (GCN) or newer
RecommendedNVIDIA GTX 10 / 16 or RTX · AMD RX 400 or newer
Basic features onlyIntel HD 3rd to 5th gen (e.g. HD 4000) · older Fermi-based GeForce (e.g. GT 720M)
The surest checkOpen LabScale and look at the dots in the bottom bar: green Real-ESRGAN and RIFE mean the AI features work. Grey? Click Setup & Status to see why.
Speed at a glanceMacBook Pro M1: one 1920×1280 output photo takes ±6.5–7 seconds with the photo model, or ±0.7 seconds with “Quick draft”.
Disk space±400 MB for the app. AI video needs temporary space — keep 20 GB+ free for long videos.
What we've testedA MacBook Pro M1 (all features) and a Windows 10 laptop without Vulkan (basic features). The Windows GPU list is based on the manufacturers' official Vulkan support, not on our own testing of every GPU. The status dot in the app remains the surest check.

What's new

Version 2.1.1 · 13 October 2026
  • On PCs whose GPU has no Vulkan support (common on older laptops), the AI features are now shown as unavailable with the reason, and the 60 FPS boost uses FFmpeg instead of failing.
  • On small screens such as 1366×768 laptops, the window now fits the screen.
  • The process log now shows every job through to its last line.
Version 2.1.0 · 13 October 2026
  • First public release: LabScale is now free and open source.
  • Interface in Indonesian and English, switchable from the top of the window.
  • Jobs can be cancelled at any time without leaving half-finished files behind.
  • Before processing video with AI, LabScale checks that there is enough disk space.
  • The UltraMix, UltraSharp and Remacri models can be downloaded from inside the app when selected (non-commercial use only).
  • A notice appears when a new version is available, plus a log folder for reporting problems.

Download & install

LabScale isn’t signed with a paid developer certificate yet, so the system asks for confirmation once on first launch.

macOS

macOS 14 or later, Apple Silicon (M1 or newer). Intel Macs are not supported.

  1. Download LabScale-2.1.1-macos.dmg (111 MB).
  2. Open it and drag LabScale into Applications.
  3. On first launch macOS says the app can’t be verified: click Done, open System Settings → Privacy & Security, scroll down, and click Open Anyway. Only once.

Windows

Windows 10 / 11, 64-bit. The installer needs no administrator rights.

  1. Download LabScale-Setup-2.1.1.exe (129 MB). Prefer no installer? Get the .zip (165 MB) and run labscale.exe.
  2. If “Windows protected your PC” appears, click More info → Run anyway.

All versions & release notes: GitHub Releases · checksums SHA256SUMS · source code: github.com/labby245/lab-scale

Frequently asked questions

Is it really free?

Yes. No account, no watermark, no file limit, no ads. The code is open (MIT license) and you can inspect it yourself on GitHub. LabScale is one of the ways we introduce Labstory as an app maker.

Are my photos/videos uploaded?

No. Everything runs on your own computer, and your originals are never changed — results are always saved as new files. Internet is only used to check for new versions and to download optional models if you choose them.

Will it run on my old laptop?

The AI features need a GPU with Vulkan support and up-to-date drivers — see the list in the Specs section; the status dot in the app is the surest check. On older computers LabScale still opens: the AI features are marked unavailable, while non-AI upscaling and the 60 FPS boost keep working.

Why is video upscaling slow?

Every frame is upscaled one by one, so it suits short clips best. For most AI models 2× is as heavy as 4× (the model only has 4× weights); the exception is the “Anime video” model, which is much faster.

Can I use it for commercial work?

The bundled models, yes. The three optional models downloaded from inside the app (UltraMix Balanced, UltraSharp, Remacri) are licensed CC BY-NC-SA 4.0 — non-commercial use only.

What about updates?

On launch LabScale checks for a new version and shows a small banner. It never downloads or installs anything by itself — you decide.

Also from Labstory Studio

Fotolane — event photos ready to upload, automatically

Photos come in from the camera over a cable, get resized and sorted odd/even while the event runs. You focus on shooting.

Having trouble? In the app click the ⓘ icon → Open log folder, then attach labscale.log to a GitHub issue, or send feedback. LabScale is built with FFmpeg, Real-ESRGAN and RIFE — full credits in the repo.