Using my apps is also a way to support me:
HiPixel is a native macOS application for AI-powered image super-resolution, built with SwiftUI and leveraging Upscayl's powerful AI models.
- π¨ Four built-in AI models: Standard, Lite, High Fidelity, and Digital Art
- π¦ Custom model support β load your own models from a local folder
- β¨ Double Processing β upscale twice in one pass for higher effective resolution
- π¬ Optional TTA (Test-Time Augmentation) for even better quality
- π· EXIF metadata preservation β camera and orientation info survives upscaling
- π GPU acceleration with manual GPU selection and custom tile size
- π±οΈ Drag-and-drop processing β images are processed automatically upon dropping
- ποΈ Batch processing for multiple images at once
- π Folder monitoring β new images added to designated folders are processed automatically
- π URL Scheme (
hipixel://) with full processing options for scripts and third-party integration - βοΈ Apple Shortcuts integration β automate upscaling with the "Upscale Images" action
- ποΈ Zipic integration for smart compression after upscaling
- π₯οΈ Native macOS application built with SwiftUI
- π Image comparison viewer with zoom (1xβ5x), pan, and one-click reprocessing
- πΎ Manual or automatic saving β review results before saving, or save automatically
- π Menu bar access with optional Dock icon hiding and silent launch
- π Three notification styles: HiPixel banner, notch-style, or system notifications
- πΌοΈ Output as PNG, JPG, WebP, or the original format
- π Light/dark appearance and alternate app icon
While Upscayl already offers an excellent macOS application, HiPixel was developed with specific goals in mind:
-
Native macOS Experience
- Built as a native SwiftUI application while utilizing Upscayl's powerful binary tools and AI models
- Provides a seamless, platform-native experience that feels right at home on macOS
-
Enhanced Workflow Efficiency
- Streamlined interaction with drag-and-drop processing - images are processed automatically upon dropping
- Batch processing support for handling multiple images simultaneously
- URL Scheme support for third-party integration, enabling automation and workflow extensions
- Folder monitoring capability that automatically processes new images added to designated folders
- Simplified interface focusing on the most commonly used features, making the upscaling process more straightforward
HiPixel aims to complement Upscayl by offering an alternative approach focused on workflow efficiency and native macOS integration, while building upon Upscayl's excellent AI upscaling foundation.
HiPixel ships with four built-in models from Upscayl:
| Model | ID | Best For |
|---|---|---|
| Standard | upscayl-standard-4x |
General-purpose upscaling for photos |
| Lite | upscayl-lite-4x |
Faster processing with lower resource usage |
| High Fidelity | high-fidelity-4x |
Maximum detail preservation |
| Digital Art | digital-art-4x |
Illustrations, anime, and digital artwork |
Beyond the built-in models, you can load your own:
- Put your model files in a folder β each model is a pair of
<name>.binand<name>.paramfiles (ncnn format, the same format Upscayl uses) - Select this folder in Settings (or pass it via the
customModelsFolderURL parameter) - Pick the custom model from the model list
When both a built-in and a custom model are specified, the custom model takes precedence.
To keep the app download small, HiPixel fetches upscayl-bin and the AI models over the network on first launch:
- All downloads are verified with SHA256 checksums
- Resources are stored in
~/Library/Application Support/hipixel/ - Updates are checked automatically at launch; a manual check is available in the Advanced settings tab
HiPixel supports URL Scheme for processing images via external applications or scripts. You can specify image processing options via URL query parameters, which will override the default settings in the app.
hipixel://?path=/path/to/image1&path=/path/to/image2
| Parameter | Type | Description | Example Values |
|---|---|---|---|
path |
String | Required. Path to image file(s) or folder(s). Multiple paths can be specified by repeating this parameter. | /Users/username/Pictures/image.jpg |
saveImageAs |
String | Output image format. | PNG, JPG, WEBP, Original |
imageScale |
Number | Upscaling factor (multiplier). | 2.0, 4.0, 8.0 |
imageCompression |
Number | Compression level (0-99). Only applies when not using Zipic compression. | 0, 50, 90 |
enableZipicCompression |
Boolean | Enable Zipic compression (requires Zipic app installed). | true, false, 1, 0 |
enableSaveOutputFolder |
Boolean | Save output to a custom folder instead of the same directory as source. | true, false, 1, 0 |
saveOutputFolder |
String | Custom output folder path (URL encoded). Requires enableSaveOutputFolder=true. |
/Users/username/Output |
overwritePreviousUpscale |
Boolean | Overwrite existing upscaled images if they already exist. | true, false, 1, 0 |
gpuID |
String | GPU ID to use for processing. Empty string uses default GPU. | 0, 1, 2 |
customTileSize |
Number | Custom tile size for processing. 0 uses default. |
0, 128, 256, 512 |
customModelsFolder |
String | Custom folder path for AI models (URL encoded). | /Users/username/Models |
upscaylModel |
String | Built-in AI model to use. | upscayl-standard-4x, upscayl-lite-4x, high-fidelity-4x, digital-art-4x |
selectedCustomModel |
String | Custom model name to use. Conflicts with upscaylModel (custom model takes precedence). |
my-custom-model |
doubleUpscayl |
Boolean | Enable double upscaling (upscale twice for higher resolution). | true, false, 1, 0 |
enableTTA |
Boolean | Enable Test Time Augmentation for better quality (slower processing). | true, false, 1, 0 |
Terminal:
# Process a single image with default settings
open "hipixel://?path=/Users/username/Pictures/image.jpg"
# Process multiple images
open "hipixel://?path=/Users/username/Pictures/image1.jpg&path=/Users/username/Pictures/image2.jpg"
# Process with custom options: 4x scale, PNG format, double upscaling enabled
open "hipixel://?path=/Users/username/Pictures/image.jpg&imageScale=4.0&saveImageAs=PNG&doubleUpscayl=true"
# Process with custom output folder and Zipic compression
open "hipixel://?path=/Users/username/Pictures/image.jpg&enableSaveOutputFolder=true&saveOutputFolder=/Users/username/Output&enableZipicCompression=true"
# Process with specific AI model and TTA enabled
open "hipixel://?path=/Users/username/Pictures/image.jpg&upscaylModel=high-fidelity-4x&enableTTA=true"AppleScript:
tell application "Finder"
set selectedFiles to selection as alias list
set urlString to "hipixel://"
set firstFile to true
repeat with theFile in selectedFiles
if firstFile then
set urlString to urlString & "?path=" & POSIX path of theFile
set firstFile to false
else
set urlString to urlString & "&path=" & POSIX path of theFile
end if
end repeat
-- Add processing options
set urlString to urlString & "&imageScale=4.0&saveImageAs=PNG"
open location urlString
end tellShell Script:
#!/bin/bash
# Process all images in a folder with custom settings
IMAGE_PATH="/Users/username/Pictures"
OUTPUT_FOLDER="/Users/username/Upscaled"
for image in "$IMAGE_PATH"/*.{jpg,jpeg,png}; do
if [ -f "$image" ]; then
open "hipixel://?path=$image&imageScale=4.0&enableSaveOutputFolder=true&saveOutputFolder=$OUTPUT_FOLDER&doubleUpscayl=true"
fi
done- All parameters except
pathare optional. If not specified, the app will use the default settings configured in the app preferences. - Multiple
pathparameters can be specified to process multiple images or folders in a single call. - Boolean values accept:
true,false,1,0,yes,no,on,off(case-insensitive). - File paths and folder paths containing special characters should be URL encoded.
- When both
upscaylModelandselectedCustomModelare specified,selectedCustomModeltakes precedence. - If
enableSaveOutputFolder=truebutsaveOutputFolderis not provided, the output will be saved in the same directory as the source image.
HiPixel supports Apple Shortcuts via the App Intents framework. Add the "Upscale Images" action in the Shortcuts app to build your own automation β every processing option is configurable per shortcut, including AI model (built-in or custom), scale factor, output format, output folder, compression, double processing, and TTA.
brew install okooo5km/tap/hipixel- Visit hipixel.5km.tech to download the latest version
- Move HiPixel.app to your Applications folder
- Launch HiPixel
Note: HiPixel requires macOS 13.0 (Ventura) or later.
- Clone the repository:
git clone https://github.com/okooo5km/hipixel
cd hipixel- Open HiPixel.xcodeproj in Xcode
- Build and run the project
HiPixel is licensed under the GNU Affero General Public License v3.0 (AGPLv3). This means:
- β You can use, modify, and distribute this software
- β
If you modify the software, you must:
- Make your modifications available under the same license
- Provide access to the complete source code
- Preserve all copyright notices and attributions
This software uses Upscayl's binaries and AI models, which are also licensed under AGPLv3.
If you find HiPixel helpful, please consider supporting its development:
- βοΈ Star the project on GitHub
- π Report bugs or suggest features
- π Support via:
![]() |
![]() |
![]() |
Your support helps maintain and improve HiPixel!
- Zipic - Smart image compression tool with AI optimization
- π Perfect Pairing: After upscaling images with HiPixel, use Zipic for intelligent compression to reduce file size while maintaining clarity
- π― Workflow Suggestion: HiPixel upscaling β Zipic compression β Optimized output image
- β¨ Enhanced Results: Compared to using either tool alone, combined use provides the optimal balance of quality and file size
Explore more 5KM Tech products that bring simplicity to complex tasks.
HiPixel uses the following components from Upscayl:
- upscayl-bin - The binary tool for AI upscaling (AGPLv3)
- AI Models - The AI models for image super-resolution (AGPLv3)
Special thanks to zaotang.xyz for designing the new application icon and main window interaction interface for HiPixel v0.2.
HiPixel also uses:
- FSWatcher - A high-performance, Swift-native file system watcher for macOS and iOS with intelligent monitoring (MIT License)
- Sparkle - A software update framework for macOS applications (MIT License)
- NotchNotification - A custom notch-style notification banner for macOS (MIT License)
- GeneralNotification - A custom notification banner for macOS (MIT License)


