Wysyłanie: DSC_0134-GIMP-darktable-Mantiuk.jpg
GIMP Raw darktable edition, further Mantiuk 2006, LCh Brightness Mode
2 Likes
Well, that was interesting…
I found the linear RGB to almost be good enough, all I really wanted from there would have been a very small curve to pull down the shadows a bit, and I demonstrated that to my satisfaction with both parametric filmic and plain-ole control point curves.
Buuuutttt… There are definitely posterized-out extremes of the red in the folds of the subject flower, near the petal convergence, so I set out to improve that. To do so, I downloaded the DPReview studio scene comparison image for the Nikon D3300 and used it to make a LUT profile of the Colorchecker target shot using dcamprof’s gamut compression, srgb mode. Using that profile, here’s what I get:
There’s a bit of hue-shift, but it definitely put definition back in those extreme reds of the sRGB render. Don’t normally do this, but I posted this one full-sized for gamut-peeping…
Here’s the profile I made:
nikon_d3300_matrix-dcamprof-lut-gamutcompressed.icc (212.5 KB)
Edit: Upon re-regarding my effort in rawproc, I saw I hadn’t saved my most recent edits, so I saved the image and reposted it…
6 Likes
@Soupy This is a nice addition to the Play Raw. Thanks for sharing!
1 Like
Hard one for sure, managing the colors was really difficult (had to use the channel mixer to pull this one out) filmic did a brilliant job but I could not keep the result entirely gamut limit safe …
Took some artistic choices with the tone equalizer.
2 Likes
Despite the fact your version doesn’t try to showcase textures (at least not for srgb display), you have managed to avoid weird/ugly blobs and fringing, and the colours are right on. Funnily enough, the textures IRL are quite subtle, and unless looking closely you might not notice them, so this version as displayed is pretty close to a first impression of the scene. Can’t wait to try Filmulator when I go Linux - just three sliders for a nice result!
Fear I may have lead you astray by saying, “IRL the flowers were all red, no white.” This was referring to the petal, not the pistil or stamen. I mentioned it as increasing exposure could quickly turn the reds white. However, it was quite the feat for you to get it all red!
It was nice to have a look in there. Interesting use of raw white points, I usually just leave that at default. Am wondering why you chose to reduce brightness with that instead of exposure?
Yes I see now, directional light with shape masks. Makes sense.
I love this rendition. Unfortunately when I load your xmp into darktable I get the following result.
Is the xmp correct? Or are you using a different version of darktable? I use 3.4.1. It seems to be color calibration module causing the biggest difference.
Thought this would be right up your alley, and you have not disappointed with profile trickery! This version does indeed have very nice gradients and I appreciate the full size for pixel peeping. Thanks also for providing the profile, I have a number of similar photos so it will likely prove quite useful. Am I right the profile gamut compression is from dpreview’s Nikon D3300 down to sRGB? Thus, the working profile chosen should also be (linear) sRGB? I guess in rawproc you work in input profile, and don’t have to consider working profile.
1 Like
Some of the texture went away because I shrank the image to 1500 pixels wide for the forum, plus chroma subsampling and JPEG compression. It’s definitely visible, though subtle, in the full size.
2 Likes
Ah yes I see it better there, and as I said, the texture is subtle, so your rendition is quite true to life. Great job!
I would definitely use this as a “special-case tool”, as it crushes down to sRGB wherever you do the first transform out of camera space. In rawproc I currently don’t use a working profile, so my first color transform is at display/output.
A bit more generally useful version might be one of the adobergb modes; here’s a version that uses -g adobergb-strong, which is the mode Anders recommends as a starting point:
nikon_d3300_matrix-dcamprof-lut-gamutcompressed-adobergbstrong.icc (212.5 KB)
2 Likes
Exposure is the first thing I tried using to get a good starting point but I couldn’t get the details in the petals the way I wanted them. Reset the module and went to the raw white points module instead.
The raw white points module because it can, at times, give nice results when stuff is (partially) blown out and I was curious if it would be able to help me here. It did. I am somewhat surprised to be honest that changing it this much didn’t have any adverse affect on the rest of the image. With blown out images the changes need to be rather small.
1 Like
i’m really liking this image. it’s mean. here are some more images to visualise the data.
the following is using a lut profile constructed from camera cfa curves. note the deep reds in the cie diagram in the top right:
this one here instead uses the adobe matrix. the reds are slightly too magenta:
since both go out of gamut, these versions use a global saturation of 0.9. spectral lut:
3x3 matrix:
the cie plot clearly shows you the out-of-gamut chromaticity. exposure is orthogonal to this (i.e. would go out of the screen in this plot if you wanted to show it). this means regardless of brightness/exposure time, these colours cannot be represented in srgb.
3 Likes
@CarVac: I just used Filmulator (v0.11.1) and it indeed does a great job with this shot!
If I Use the settings you posted somewhere above I notice that setting the White Clipping Point (0.7) doesn’t give the nicest result in my case. 1.15 seems to be much better. Is this a monitor issue or is something else at play?
BTW: Dell U2719DC, calibrated and profiled (a few weeks back now).
hmmm… I have reloaded the *.xmp and everything is fine.
can it be that in the module color callibration, the options “normalize channels” are not activated?
@Suki2019: I’m having the same issue if I load your xmp.
It isn’t the normalize channels, which is turned on (as expected, it is part of your xmp). Turning it off turns the flowers almost black.
It seems that -0.625 is just too much, if I set it to -0.368 I’m getting a result that is very similar to what you posted above.
EDIT: Just tried to see if your posted xmp and the embedded one are the same but there’s no embedded info in the jpeg 
1 Like
3 Likes
Tried again. Despite now knowing that details of this red monster are very faint, I tried to emphasize them and at the same time to avoid the bleeding of red I had in my first version around pistil and stamen. Really great photo to play with!
Unfortunately the blossom is a bit out of focus.
2 Likes
I think it may just be a matter of personal taste. I didn’t want the rest of the image to look very dark, so I pushed the red flower as bright as I could without clipping.
But if you put the red in the midtones using a higher white clipping point you can see more contrast in the petals.
I really like your reply. It is an example of what we should have more of in these Play Raw topics. Not just posted plays, then nothing. The results should also be discussed. This occurs occasionally, but not very frequently.
Thank you.
5 Likes
Yeah, I’ve been playing with Filmulator and this shot and, as always, personal preference does play a roll.
I’m impressed by how easy it is using Filmulator to get a very good base (arguably the end result). You can go and fine tune till you are blue in the face but the ease at which a good result is established compared to both RawTherapee and darktable, which can definitely get good results but take more fiddling, is noteworthy.
2 Likes
5 Likes
@Tim I agree. When I have the energy, I typically try to strike conversations in Play Raws but it also depends on whether people want to talk. Sometimes the opposite is true: people are too willing to talk about one aspect (ahem filmic) and we get lost in a wall of words.
3 Likes