The wavelet tool in RawTherapee offers several options to treat high-iso noise, as shown in this article.
For this demonstration I use a 6400 iso raw photo taken during a party. When opened in RawTherapee with the neutral profile applied, it looks like so.
The preview window at 100% show this.
As a reference, here’s a screenshot of the preview window with the standard noise reduction filter with default values applied. This filter is located under the Details tab, so not in the wavelet section.
This is a sophisticated noise filter that can do better than this, but here it serves only as a reference.
Go to the wavelet tab, leave everything at their default values and open the contrast entry. Set the second level to minus 50, you’ll notice no change or hardly in the preview window. The trick we’re going to use lies in the Highlight luminance range slider, in the ‘Apply to’ section under the sliders.
Drag the left control points to the left, like so.
Drag the sliders on the right to the left to increase local contrast on the white paper. But not too much to avoid that noise starts to get amplified again. By holding the Shift key you can change the position of the individual control points.
First the result, then the original 100% view. The third one is the standard NR reference screenshot we showed above.
A second method to deal with noise can be found under the dedicated ‘Denoise and refine‘ entry in the main wavelet window. First we ‘reset’ our photo by applying again the neutral profile at the top of the right panel of RawTherapee.
Open the Denoise entry by clicking on the triangle left of its name, you’ll see this.
I seldom use sharpening filters on high-iso shots (3200 iso and more), because they tend to sharpen the noise as well. So I uncheck the ‘Link with Edge sharpness strength’ because that is already at 0. Noise on this kind of shots (6400 iso) can be treated with the sliders at levels 2 and 3, level 1 has only a slight influence. The latter depends without doubt also on the kind of camera sensor, its physical size, its megapixels and its quality.
I move all the three upper sliders a bit to the right, right above the bottom sliders. The last ones (‘Strength‘) increase the local contrast when moved to the right and decrease local contrast when moved to the left. Be careful not to increase the contrast too much, as that directly affects the noise.
And the result, followed by the 100% before view and the reference screenshot.
As always, fine-tune the result with the main slider called Strength at the top of the Wavelet panel.
Final local contrast
The third possibility to treat noise can be found in the Final touchup entry. I guess this has nothing to do with wavelets anymore as we are ‘under’ the wavelet levels and the residual image, but it works anyway.
Load the neutral profile again. Move the left part of the curve under Final local contrast down, and the right side up. The effect is that we decrease the local contrast in the shadows and the midtones, while increasing local contrast in the highlights.
The curve looks like so.
And the result, followed again by the before view and the standard NR reference.
Here’s a downsized photo (1600px wide) with the following denoise settings using the Denoise and refine entry.
I decreased the amount of denoise on level 3 as that blurred the face of the girl in the middle too much. I increased the strength of level 2 (bottom slider) to make it look sharper. Yes, using a denoise tool to sharpen images is possible, as sharpness is just the other side of noise!
Please note that this photo is processed with the wavelet denoise tool only, to exclude effects introduced by other tools. So this is not the best possible end result.