As I may have indicated, I donât feel comfortable using dcrawâs wavelet denoising at all. Interesting how you denoised again, and after the resize. Still figuring out how best to resize, sharpen and smooth, but I think I am getting better at it, or at least have an eye for it now.
Oops, âtwo applications of denoiseâ was meant to end with a comma. One wavelet=1000, one nlmeans=29.
The NLMeans algorithm works really well, but my implementation is not optimized (one day Iâll get to it), and Iâve found the dcraw wavelet denoise to work well.
@afre, just looked and couldnât find, whatâs your concern with the dcraw algorithm?
Just saying that when I tried using it the results werenât satisfactory. In my workflow for alt2, I simply split the image into several scales and removed a scale that contained almost all of the noise. Now, that doesnât usually apply but what I am trying to say is that I prefer the good old fashioned way. As an aside, nlmeans is too much for my laptop to handle, so I havenât been using it.
This image was too much for my Surface 3, had to move to the desktop computer with 8GB RAM. On the Surface, had all the processing stacked up, went to save, and the program aborted, no messages. May have to rethink my full-frame ambitionsâŠ
But [why is there always a but?}]: why on earth did you convert into Rec2020 linear?
The only thing it did (?) was to shift the histogram to the left. Or is there something that I have misunderstood?
Initially, I did it to reduce the amount of out-of-gamut values. Now, I do it to confound you. In truth, it has a different aesthetic, like medium format, and has become my signature.
Hmmm⊠if the histogram shifts left when the image is converted with a linear Rec2020 profile, thatâd indicate to me the image already has a non-linear profile assigned. For my cameraâs images, I start by assigning a linear calibrated profile, then Iâll convert to a gamma 1.8 Rec2020 profile for work, and the histogram shifts right. If I use the linear Rec2020 profile instead, the histogram doesnât change. Thanks @elle, for all the neat profiles; they are the foundation of my âprofile zooâ.
I wasnât making any strange beverage⊠after washing the machine, I made a couple of waste coffees to refill the pipes and try.
So, in the end it was just for art. While I was washing the machine Iâve seen the chance for some nice shots.
And wow guys⊠Iâm so glad to see youâre all making these great works. Itâs cool to see youâre having fun and Iâm actually learning quite a lot.
@billznn, sorry for the off-topic, you seem to have changed some lens-related metadata. (I used to do this too, for the same camera/lens combination, but donât any more.) Which editing program did you have to please?
What do you mean by changing lens metadata? I canât fully understand.
What I can say is that for some reason the lens recorded in the metadata is a Canon 17-55, but actually, as I wrote in the initial post, the lens is a Sigma 17-50, so I used the correction for the Sigma.
I used Darktable.
Sorry for the confusion, your lens causes different metadata to be written than mine, perhaps it is a different version. Mine is recognised as Sigma 17-50mm f/2.8 OS HSM or Tamron SP 17-50mm f/2.8 XR Di II VC by ExifTool, which canât distinguish the two similar lenses. (While the programs, using Exiv2, can.) Darktable recognises it as Sigma 17-50mm f/2.8 EX DC OS HSM. Since âfor some reason the lens recorded in the metadata is a Canon 17-55â in your case, I was wondering, if you have changed anything. Well, you have not. For darktable, your lens is Canonâs EF-S 17-55, which triggers the âwrongâ lens correction data.
This is about third-party lenses and their ways, it might not be worth the time. Thank you for your patience, have a nice evening!