I would love to consult with you a bit my workflow how I develop a Nikon raw images … I use primarily RT + GIMP but in past months I’ve also tried DT and ART … I absolutely agree that probably DT is the best and the most powerful OSS software for developing camera raws but for my workflow RT + GIMP suffice …
On high level I do
in RT I use only Exposure tab and Colors tab while 99% of my work is done just with
Basic Exposure tools (curves mostly and black level + exposure compensation, I don’t use lightness, contrast and saturation sliders at all), Tone Equalizer (before TE introduced in RT v4.x I was using Highlights and Shadows)
Lab
Graduated filter - sometimes
then in RT a Color tab where I just use
White Balance tool
Color toning → Color Correction Regions with masks - for kind of split toning
everything else in GIMP
Sharpening and Denoising
all contrast and microcontrast restoration (high-pass / overlay mode)
Additional local edits via masks (typically dodge and burn)
Blending
All kind of scaling and exporting to jpeg
My question is simple - which tools from opensource toolbox (RT, DT, ART and GIMP) are the best for these particular tasks -
I have to acknowledge that I am biased. I am a dedicated DT user. RT and ART are both excellent programs but DT is my go to program. I also have GIMP installed and rarely need to use it.
There was a time when I used much earlier versions of RT which lacked localised adjustments. So when I was editing a landscape I might do two versions in RT. One was optimising the sky and the other the foreground. I might then bring these two exported images into GIMP and use a layer mask to combine the two versions.
However, when I switched to DT this tedious workflow ceased because DT has localised adjustments for nearly every module including denoising, sharpening, exposure and color adjustments. I simply didn’t need to use GIMP anymore except for very rare occasions.
I struggle to see any advantage in using GIMP for denoising and sharpening when DT offers an excellent range of options for this. I will leave it to RT and ART users to make their comments if they can see an advantage to use GIMP in the workflow, but I don’t see it with DT.
I don’t think either are clearly better than the others. It’s more about how well you can use the tools that you’re using. We’ve had long threads comparing sharpening in RT and DT. People really seem to like RT’s capture sharpening, and it is quite good.
For noise reducution, I think there are two kinds of people: (1) finding a result that is acceptable or (2) must eradicate all traces of noise. If you’re #2 then you probably won’t be happy with either, but if you’re #1 then probably either is fine.
Hi Terry … thanks for your input … I am considering quite some time to fully switch to DT … I am just used on to use my workflow that I am a little bit unhappy with it already because I’ve been trying DT out and agree that it’s incomparably more powerful than RT … for example these two images are from last weekend
I am definitely not knocking RT or ART or even GIMP. For me the reason I picked DT over other programs including commercial programs is the ability to create drawn and parametric masks for every module that can use them. That was the deal breaker for me.
I also see GIMP as a graphic arts tool, but DT and RT as raw file editors. So I only go to GIMP when I want to use my images in a graphic design such as a collage or applying artistic filters to create effects. It certainly has no place in my mainstream editing workflow.
GIMP was my go to program in the early days of digital when we all shot JPGs. It would still be a good choice for editing JPG images.
For Sharpening, I prefer ‘RL deconvolution’ in RawTherapee rather than ‘Unsharp Mask’ in the GIMP.
But sometimes the GIMP can be better because it can apply USM again and again … whereas in RT only one instance can be applied (any new settings overwrite the old ones).
DT also can apply multiple instances of nearly all modules including those that sharpen and denoise. I am not aware of an equivalent to RL convolution in DT. I am happy to be corrected on this if I am wrong. I wonder if RL convolution is something that would be useful in DT?
I think AP was working on that, but it was too resource intensive. But I believe there’s been discussion on incorporating the RT capture sharpening into DT… I don’t know if anyone is championing that.
In DT you can run the RL sharpening from GMIC on export if you run the lua script… It adds a target to the export box…it works well but you fly blind and have to experiment…
Is the denoising superior in GIMP?? I can’t say as I rarely use it but I would think it might be nice to denoise your images early in your edit esp if there is a lot of noise so you don’t edit with that as part of the data but maybe its a non issue unless maybe there is a lot of noise and using GIMP for that is more to your liking…
I personally like to apply initial sharpening and denoise (profiled) to my edits at the start of DT editing so I can see what I am really editing. Decades ago I was taught by an experienced graphic artist to do sharpening as the last step in PhotoShop editing, but I feel that is because GIMP and PS are both pixel editors and that logic doesn’t hold for raw file edits.
I love options, so I would like to have the ability to compare this to what I already do with the diffuse or sharpen module in DT.
I was editing a landscape and the very small leaves on the roadway retained their shape on RT, and DT did not reflect any of that form. However, that was at a 100 percent crop and I didn’t think the differences mattered when zoomed to fit. Plus, this was before Diffuse and Sharpen. I’ll see if I can run a comparison and set it up as a raw play
I mainly use Darktable. For moderately noisy images, Darktable’s denoising is fine. For very noisy images, however, I run Neat Image as a post-processor on the JPEG.
I’ve tried all of the AI denoisers. TopaZ Denoise, ON1 NoNoise AI, DxO Photo Lab, Lightroom Enhance Detail. They’re all very impressive, but at the end of the day, they invent detail where there isn’t any. Sometimes that looks convincing enough, sometimes it just looks fake. Faces in particular tend to get overwritten with generic AI faces, deleting any semblance to real people. Ultimately, I prefer a noisy image over an imaginary one.
It’s anyways mostly the shadows that get badly noisy. If they get too noisy, I’ll sacrifice them to blackness, instead of conjuring zombies.
But Neat Image is a good middle ground. It’s algorithmic, not generative. It’s payware, but affordable, and runs on Linux. If I were to spend more, DxO would be my next pick, with all its flaws (can’t denoise JPEGs, nor phone pics).
I am sometimes gripped by the idea that the grass must be greener on the other side of of the OS hill. You have put my mind at rest. (at least for a while )
I too have Neat Image for occasional use. It is low-cost enough for that. In fact the free demo version is mostly good enough for that. I wonder why it is not better known. I suppose they don’t care to match the big names on PR/Advertising spend.
Just downloaded Neat Image and tried it - does a fantastic job. I edit in darktable so what would be the workflow ? Edit raw or jpg in darktable, export as 100% jpg and load into Neat Image to finish off with denoising?
Hi, I use Art for raw development, and for noise if I want to export to Neat Image, I just remove the chromatic noise in Art and save in Tif and finish the denoising and sharpening work in Neat Image.
If I’m feeling really serious about a picture, I’d export it from dt as a tif, and import that into Neat Image. I don’t recall if that needs the paid copy or not.
It’s rare that I’d want to save a really noisy image. Rare… but it does happen
)
