I use Darktable (currently v5.2.1 Flatpak) for processing RAW images, but I don’t use it often enough to really know how to use it properly. I fumble my way through and generally end up with something that seems OK, but I often feel that the embedded JPGs are better than what I end up with and I get the impression that probably means I’m doing things quite wrong. An example is this recent picture. The bird in the embedded JPG is a vivid in-your-face red, which is how it actually looks in real life, whereas my efforts made it look more orange. Having back-tracked through the history stack I see that it already looked orange with the default processing steps that Darktable automatically applies, but that’s not much consolation when I can’t figure out how to change it.
So, my first question is how to make the bird in this particular picture look more vividly red like it does in the embedded JPG.
Secondly, I probably ought to go back to basics in an effort to understand the main modules (which have probably changed since I first read about Darktable) instead of just stabbing wildly at it and hoping. I know there are quite a few video tutorials out there, though some are probably out of date especially since the “scene referred” changes. Are any of them considered particularly good for covering the essentials in modern versions of Darktable?
Nice bird Have you been using the Color Balance RGB (there is where chroma and saturation are)? Also the RGB Primaries may help you to make increase the red purity to make the orange more red.
Hey, and welcome. Overall, the SooC jpeg just looks darker. Reduce the exposure in your edit, then use one of the color modules to intensify the colors to get closer to the jpeg. Just reducing the exposure should help a lot though.
I too have a Canon R7. I have made a style, which I am sharing here, to start my edits. The style aims to give a starting point similar to but not identical to the camera’s jpg. The style is based on the new AgX module found in DT 5.3 and about to be released in DT 5.4.
In this style I have played with the AgX primaries to get the color tones to closely match the camera’s colors. I have also lowered global chroma in color balance rgb module.
But this style is only used as my starting point. It is not meant to be the perfect aim point. It has helped to speed up my edits and often is all I need to do with some shots.
Here is my Sigmoid module based starting point for my Canon R7. The style I have applied here would work in DT5.2. I have switched to AgX in DT5.3 as my default tone mapper. But this xmp file is relevant for 5.2.
Just a simple edit without much color adjustments. Normally I would use masks for the diffuse or sharpen and contrast equalizer modules.
It uses the AgX module in 5.3 but most of the work is in the color balance rgb module. The Canon look has more saturation. The brilliance sliders can be adjusted to taste. In the 4 ways tab, the hue and chroma under shadows lift is just a basic way to shift some of the hues, in particular making the greens a bit more rich looking. Using AgX or sigmoid primaries or the rgb primaries module would give a lot more control.
I’m in agreement with @paperdigits in that for the most part, the embedded jpg is darker (zoom in and look around the eye of each image) and simply reducing the exposure as suggested should get you close.
It’s personal preference, but I like your editing results better than the embedded jpg since you can actually see the details around the eye on yours. The colors also look more “natural” to me in your edit (granted, I wasn’t there, so I don’t actually know).
Yep, I’ll pile on to the exposure thing. If you look closely at the colors in the camera and darktable JPEGs, the hues are all pretty much the same, just that the darktable one is a bit brighter.
Something I’ve known for a while, but I haven’t tried to articulate it 'till now: Learning to recognize the difference between hue and luminance changes is pretty important to figuring out what tool to tweak.
I actually just downloaded the edited JPG and the RAW, and it looks like the embedded JPG has a lot of softness (2500 ISO, but dt can take care of that noise and keep the detail a lot better than the embedded image), some added contrast, and a move toward cooler rather than warmer to get the look. So simply dropping the exposure down doesn’t quite do it—I tried that.
I’m still sticking with OP’s edits as better than the embedded JPG.
The green background just looked too saturated to be true, so I downloaded the .CR3 and opened it in rawproc to see what the “real” no-kidding linear rendition looked like. Yep, greens are more “yellow” with a standard matrix profile. Sooo, I have some spectral data, including a pretty comprehensive set from Image Engineering courtesy of @hanatos, so I went looking for Canon R7 data. No joy, but the dataset includes the R5 so I genned a SSF icc file and gave it a go with this .CR3:
hue-wise, the SSF profile handles the extreme ends of the bird-oranges better than the matrix profile, but that’s not novel, just what a better-trained LUT camera profile will give you.
@mino, he did (and I think most of us didn’t quite catch that, including me ):
Blackmyre, this requires isolating the reds in just the bird and not messing with the entire image. That will most likely require you to learn some basics with masks, so some more video tutorials would be good for that (I’ve come across some but can’t think of them off the top of my head).
Wow - I didn’t expect responses to be so many and so helpful. I’ll have to take a bit of time to work through them all and do some reading around to try to understand the points they raise, but I wanted to post a quick acknowledgement so you don’t all think I’ve melted away. Yes, it’s all a bit overwhelming but another factor is that I’m not at home, I’m currently on a holiday (hence the exotic bird picture) to celebrate my wife’s 70th birthday. It wouldn’t be politically smart to spend loads of time on a laptop in the hotel, especially as today’s the special day. The edits have been really impressive (humbling actually) and useful. I’ll try to respond more fully soe time in the next several days. Thanks very much to all who have comented, I really appreciate the help.
Have you been using the Color Balance RGB (there is where chroma and saturation are)? Also the RGB Primaries may help you to make increase the red purity to make the orange more red.
):