A simple command-line C++ tool that allows you to apply a Cube format LUT to an image.
It offers high-performance multi-threaded and GPU-accelerated modes, various interpolation methods, and support for 1D and 3D LUTs.
See performance tests for a performance comparison between the offered modes.
- Simple CLI
- 1D LUT support
- 3D LUT support
- Nearest value mode (faster, low quality of tonal transitions)
- Trilinear interpolation (slower, better quality)
- Tetrahedral interpolation (slowest, best quality)
- Multi-threaded implementation (the number of physical CPU threads is used by default)
- CUDA GPU implementation (highly effective with larger images)
| CPU | CUDA | |
|---|---|---|
| 1D Nearest-value | ✔️ | |
| 3D Nearest-value | ✔️ | ✔️ |
| 3D Trilinear | ✔️ | ✔️ |
| 3D Tetrahedral | ✔️ |
-h, --help Display this help menu
-i [input_path],
--input=[input_path] Input image path
-l [lut_path], --lut=[lut_path] LUT path
-o [output_path],
--output=[output_path] Output image path
--intensity=[intensity] Intensity of the applied LUT (0-100
[%]). Defaults to 100%.
--interpolation=[method] Interpolation method (allowed values:
'trilinear', 'tetrahedral',
'nearest-value')
-f, --force Force overwrite the output file
-j [threads], --threads=[threads] Size of a thread pool used to process
the image. Defaults to the number of
logical CPU cores available on the
system.
-p [processor],
--processor=[processor] Processing mode (allowed values: 'cpu',
'gpu', 'metal')
--width=[width] Output image width
--height=[height] Output image height
A suitable NVIDIA GPU is required for the GPU mode to work (preferably with compute capability 6.1 or up). However, if you don't have one, you can still build and use the program in CPU mode!
- GPU acceleration for the remaining modes
- Further performance optimizations
- Support of color depths other than 8-bit
- Batch processing
- Metal acceleration (macOS)