Correlating Noise reduction curves with image luminosity regions

Using RT dev appimage under Linux.

In the Noise reduction filter the Luminance curve option allows selective noise reduction of dark or light areas. This is quite useful. But I have sometimes struggled to identify the right areas on the image that correspond to
the darkness (or lightness) indicated on the x-axis of the NR luminance curve. Unlike some other curves in RT (like Color Correction Regions CTRL-Click)
there doesn’t seem to be a facility to ‘pick’ a point on the image for marking its luminosity region on the curve. So what I do sometimes is to put a Luminosity filter (shortcut v) on the image and visually identify the approximatly corresponding luminosity region on the Noise reduction curve by looking at the darkness or lightness of the image. A bit of trial-error usually yields a good result.

Is there any other tip or trick to more precisely correlate the luminosity region between the image and the NR curve?

1 Like

You can use the Lockable Colour Picker set to Lab mode in combination with the curve’s in/out value.

The picker’s L value shown is between 0% and 100% and the in/out curve value range is from 0.0000 to 1.0000, which translates nicely.

This might show a bit better what I’m talking about:

The above is just a very rough one-point example, but it does show that you can, fairly accurately, target the area you are interested in.


Another way to approach this, although not exactly what you asked, is using Local Adjustments with the Blur/Grain & Denoise tool. You can accurately set the area that needs to be targeted and use the sliders to dial in the amount of de-noising that is needed/wanted.

Using the Shape detection part in Settings you can further fine-tune your dialled in settings (not shown in the below video snippet).

1 Like

Excellent tips @Jade_NL, gratefully noted, and tried too. Thank you…

A thought - Such key tips would be a great addition to the help text/manual for various filters. Some method of merely indexing/linking/tagging such posts to the index of the RT manual could be a good shortcut to retain and deliver knowledge of the finer aspects of RT.

1 Like

You’re welcome!

That would indeed be a nice-to-have. Not sure how easy that would be to set up and maintain, though.

I tend to use the knowledge that’s right here on pixls. Its search engine isn’t all that bad and most people are willing to share their knowledge if it isn’t already available.