As you know I'm always looking to make better use of the signal evolution Tracking data that is being harvested. I just uploaded 1.6.380 which has a completely rewritten wavelet sharpening module.
While the previous module was already signal quality aware, this awareness was only used to suppress the development of noise grain.
This new version of the module is now able to determine per-pixel the minimum scale at which the signal is still good enough to bring out latent detail. It then proceeds to target that scale automatically if/when needed.
In practice, the result is pretty effective in bringing out latent detail and structures that - themselves - are mired in higher scale noise, while greatly reducing the introduction of noise due to oversharpening.
Some examples with otherwise identical (exaggerated!) settings in 1.5 (left) and 1.6 (right);
-
- snrsharp2.jpg (374.14 KiB) Viewed 3293 times
-
- snrsharp1.jpg (377.41 KiB) Viewed 3293 times
