None of above but gave it shot anyway:
GIMP scaled 25% then cropped and GIMP NR filter plus G’MIC NR, I forget which. Contrast, brightness and 2 levels of USM applied. I just played.
1 Like
@xpatUSA usually we say a bit about what we have done and what tooling we’ve used.
OK will edit
Hello, in fact I don’t have a precise setting, in general I slide the luminance slider to get a blurred photo without artifacts, and then I gradually bring out the details, until it suits me. In the case of this image I concentrated only on the bird, staying in “conservative mode” knowing that I was going to apply mask-guided smoothing to the rest of the image.
I hope @agriggio maintains his little gem Art for a long time to come.
Translated with DeepL.com (free version)
2 Likes
In response to @Leniwiec , who uploaded a version developed with DXO, I’m posting a version developed in auto mode with Neat Image (evaluation version), which I’ve just resized with Digikam. It’s certainly not open source, but it does very well in auto mode. I even find it better than DXO on this image (less smoothing) and it has the huge advantage of also being able to be installed on Linux.
3 Likes
There are a few out of the box approaches worth combining in cases like this
- Raise the black level just above the noise: Perfectly clean black in areas beyond rescue.
- Push noise in the edges down with a dark vignette when the motive is centered
- Parts that are out of focus anyway can get an aggressive noise reduction. There are no details to destroy.
- Noise isn’t that visible when there is detail
- Complementary masking to treat important/unimportant parts differently
- Crop out parts that make processing difficult, e. g. the ‘hot’ branch at the extreme left
I didn’t waste any time in proper mask drawing and didn’t touch the black level:
2024-08-30_11-02-04.89_DSC0728.nef.xmp (33.9 KB)
5 Likes
Just concentrating on a general noise reduction. Not a partly one, but over the whole picture. This is the best I could achieve, without giving away to much details:
2024-08-30_11-02-04.89_DSC0728.nef.xmp (19,4 KB)
100% view:
2 Likes
Yes 100% luminance correction then bring back the detail seems to achieve the best noise reduction and detail in RT and ART. Seem to remember seeing some advice to do this somewhere and getting better results but this was a long time ago so I can’t remember where.
The image is quite good for that iso. One has to be very careful not to interfere too much with the fine details. I would rather leave some noise.
Here my try with raw and profiled denoise plus hot pixels (!).
2024-08-30_11-02-04.89_DSC0728.nef.xmp (12.4 KB)
Crop with 100% resolution.
3 Likes
Thanks for the opportunity to play with an interesting image!
v5WbOBzo1cDybfZ4H9PGG8UL96X.nef.xmp (14.2 KB)
Key ingredients for my edit:
- restrict most things that emphasize noise (eg local contrast) to lightness (blend mode),
- similarly, do things that decease noise on chromaticity blend mode,
- blur the background a bit,
- add a bit of saturation to the orange and blue plumage (colors otherwise left alone),
- give color smoothing passes in the demosaicing algorithm (improvements above 3 are marginal though)
4 Likes
looks better than mine. Will have a look at your XMP.
Ya it was good to know. I had not explored using it that way………Thx for the feedback….