This is an early morning shot with ISO6400. I was following some youtube strategy to use different denoising settings for background and subject in darktable. Maybe there is something better. Power up your engines 
151A2902.CR3.xmp (32.2 KB)
151A2902.zip (55.3 MB)
vkdt render:
timings for full res:
[perf] sum denoise: 5.601 ms
[perf] total time: 14.330 ms
What did you do in Gimp?
Played with standard tools in DT an tried to strike a balance between noise reduction and detail preservation …
Went heavy handed on the color noise with ‘contrast equalizer’ and more cautiously on luma noise …
I feel a brighter in camera exposure may have reduced the noise problem, but I first brightened the exposure to +2.457 EV. I then used denoised (profiled) which is my goto for noise reduction. With noisy images I increase the strength to 1.5. I then did some adjustments with local contrast, shadow and highlights (two instances) and tone equalizer to brighten the bird. I applied the diffuse or sharpen preset for sharpen demosaicing (AA filter). Upon further experimentation I realsied that the raw denoise module worked well on this image. I rarely use this module. I discovered astrophoto denoise did not work well for me with this image. Some images it works great, but not this one. Then I discovered the contrast equalizer module worked well on this image to get rid of the final noise if I increased the mix to +1.35. I do not often use this module any more for denoising, but it worked well for me on this occassion. As shown in the xmp file, some modules such as shadow and highlights, local contrast and the diffuse or sharpen module shared a mask drawn aound the bird and branch to localise their effect. BTW, how bright my final image looks will vary depending on the screen it is being viewed on. So don’t judge that aspect to harshly. Also, I used filmic V5 with preserve chrominance set to no in order to bring out the beautiful blue on the bird.
Beautiful image, thanks for sharing.
My try in GIMP. Since the question was about noise reduction, I decided to try the G’MIC Denoise. It says that it is slow - it took approximately 16 minutes to run (4GB/s processor with 8 cores). With the default values there was only a very slight noise reduction.
Here is my cropped result - others have already achieved much better!
Yeah I am still not good at choosing the right settings but I will get there. Thanks for sharing, will look into the xmp file soon to learn 
Yeah, the GMIC AI filter I found underwhelming too unfortunately
Hello, have a look at another denoise filter in G’mic, it’s called MS NL-Means C Noise2. That produced the following image on a 50% downscaled version of the original. I applied as well the sharpen filter Inverse diffusion, with iter=1.
Processing time was 100 sec on an old 4-core machine.
Poor result if pixel-peeping, but I think usable at Full-HD on my 22" display. Somewhat sloppy masking around the spiderweb’s threads, that could certainly be improved.
151A2902.CR3.xmp (22.3 KB)
Many thanks, I will give it a try. Iain Ferguson has produced some excellent filters.
Voilà, Rawtherapee + Gimp
Your development looks better to me, I don’t know yet to use wavelet levels and local adjustments for the denoise which I think would be useful in this case.
However here I tried to remove the grain from the background, but I also liked it in some ways, it reminds me of the old photos at 400 iso…
Please let’s keep in mind that we should apply some compression and keep images to 1080P.
If you want to show 1:1, crop it.
Nice shot and tricky with denoising. Here my try in RT 5.9-dev:
151A2902…jpg.out.pp3 (22,0 KB)
Edit: jpg compressed
