Noise reduction and loss of saturation

I am using the stable 5.8 version of RawTherapee. While editing files from my Panasonic FZ330 camera I have noticed a loss of saturation when I enable the default noise reduction, especially in areas where colourful details are small.
Here is an example. Please look at the leaves and the moss covered branches and rocks. Left is the camera jpeg, middle is RT5.8 with noise reduction enabled and on the right is RT5.8 with noise reduction off. Is RT mistaking small areas of colour as chrominance noise? Is there anything I can do about this without having to manually adjust the chrominance noise settings for each image I process?

Here is the raw and camera jpg.
These files are licensed Creative Commons, By-Attribution, Share-Alike

P1490442.RW2 (14.2 MB)

I just had a look at this image and, when leaving the Noise Reduction → Chrominance setting to Automatic global you will indeed see a reduction in saturation (mainly visible in the greens).

This one needs to be dialled in manually: Set Method to Manual and, in this case, reduce the setting of the Chrominance -Red-Green slider.

I only did a quick check, so do have a good look if you need to manipulate the other options as well.

Tried your image in DT …same happens … moss desaturates with default settings in denoise profiled.

Yeah, auto denoise really does not like the greens in this image.

The image is really soft and the sky is blown out but I didn’t notice a lot of obvious color noise say in the water or other areas…so its not an answer but maybe this image can get away without it…

My experience is that you loose saturation in a lot of photos when enabling noise reduction. As someone who photographs a lot of architecture I know that brick and noise reduction don’t match . You loose a lot of colour .

Going back to a video rawfiner made for version 2.6…he runs through all the denoise options in DT…back then it was bilateral denoise and now called surface blur…he shows how to do it channel by channel…much better than global color noise for situations like this but also not something you can easily automate but you could pull it out for tough images…