The latent_upscale_method parameter was defined but never used.
Adds 'latent' as a third upscale_method option that performs
direct latent-space resizing via torch.nn.functional.interpolate,
skipping VAE decode/encode entirely for faster upscaling.
Add comprehensive documentation for:
- Karma Lens FX (chromatic aberration, vignette, distortion, halation)
- Karma Tone Curves (shadow/midtone/highlight, S-curve, split toning)
- Karma Film Emulation (10 film stock presets with usage tips)
Also add new workflow examples including full camera simulation pipeline.
Professional tone curve adjustments:
- Independent shadow/midtone/highlight control
- S-curve midtone contrast
- Black/white point clipping
- Split toning for shadows and highlights
- Add Karma KSampler Cycle node with multi-cycle sampling and progressive upscaling
- Add Karma Film Grain node for authentic analog film texture effects
- Add Karma Kolors node for professional color grading and correction
- Include comprehensive documentation with usage examples and troubleshooting
- Add support section with donation link for future development
- Include all necessary project files: LICENSE, pyproject.toml, requirements.txt
- Set up proper Python package structure with __init__.py
KarmaSwint