Hi everyone,
in trying to choose a reference tool for developing RAW files (Fujifilm RAF), I tried a comparison with three solutions, one of which is RawTherapee which I would prefer for a variety of reasons.
However, I note that compared to the other tools I have examined, RawTherapee always generates rather grainy images with not very homogeneous colors.
I put three examples here; it is the detail of a balcony with a vase containing ivy. I zoomed in to 200% to try to highlight what was said.
Do you think this is a correctable problem or an unavoidable feature of RawTherapee?
Thanks in advance for the answers and suggestions.
This is probably the right explanation for the graining. Use some chroma denoising (default if you active Noise reduction) and youâre much closer to the other examples.
Hello all and thank so much for the quick replies.
In the example above I didnât apply any filter, in any of the used tools.
I wanted to compare the development process of the tools with the minimum process: only white balance (always detected by the photo tool from the camera shot data).
As from your suggestion I tried just few minutes ago to apply in RawTherapee the âImpulse Noise Reductionâ and âNoise Reductionâ with the default values, but I canât appreciate improvements (maybe my mistake or incompetence?). I upload here a screenshot.
Good idea to post an original RAF file and the related pp3 file.
I hope to have a little bit of time today to post a shareable shot (this one is from an house not of mine and I want to avoid any permission trouble).
RawTherapee is really powerful when it comes to noise reduction. Admittedly you do need to find your way around and might want to create a set of profiles based on noise (ISO) levels that are specific for your camera. The latter isnât needed, though, it will just make your starting point much easier.
I noticed that you use the slider for luminance control. You would be better of by using the curve. You have much better control over which area (dark/light) gets how much noise reduction.
You can also use the denoise tool in the Local Adjustments module (see the below New Tools Basics video) to target very specific areas while leaving other parts alone. Areas can be selected by colour, luminance, etc
Andy Astbury made a bunch of videos that deal with noise reduction that will probably get you up and running:
Hello, I always set Luminance to 20 and Detail recovery to 70, for me that works in most of the (less or more) normal cases, so from iso 400 to 1600 or so. I never use Impulse Noise reduction, because the Noise reduction tool is all I need.
Great suggestions, thanks a lot to all of you (and of course especially @Jade_NL).
I absolutely have to spend a little more time studying and testing RawTherapeeâs features.
Hi,
RT seems to be producing much more saturated colors than the other softwares and the darkest areas tend to show much more noise. I believe that you have the âAuto-matched Tone Curveâ active that tries to fit a tone curve to the embedded JPEG. Also be sure to be using the âAuto matched camera profileâ and activate all options Tone, Base and Look. Now your image should be developed in a neutral way and from here you can do your customizations. I tend to avoid noise reduction, and I avoid altogether the luminance noise reduction: Fuji cameras have a pleasant grain in my opinion. Of course hi-ISO images are good candidates for BW development.
Last but not least: check which RAW demosaicing youâre using!!! Noisy images tend to prefer the âFastâ method at the expense of resolution and Moireâ artefacts. When you become more expert you may play with Markenstein+fast with a variable threshold. Instead of using the noise reduction module you could also play with the âFalse colors suppression stepsâ: I usually set it to one for noisy stuff.
Hope that helps!
marco
ps: I forgot maybe the main point: if youâre not planning to see your pictures through a microscope, please avoid 200% magnifications⌠Even 100% is equivalent to look at a huge poster from 30cm⌠usually 50% is comparable to seeing an A3 print at monitor viewing distance.
I checked my settings and did not have âAuto-matched Tone Curveâ active.
Instead I activated, as you suggested, âAuto matched camera profileâ with all the options: Tone, Base and Look.
I tried to improve with Demosaicing but I couldnât take out that damn multicolored grain.
In this case only âNoise Reductionâ seems to work, sorry
I set it like this:
Color space: L * a * b
Mode: Conservative (with Aggressive the rendering is a bit too fake)
Range: 1.70
Luminance: 30.00
Details recovery: 50.00
At this point the result was already very good but remembering that at the beginning you noted that RT could generate more saturated colors than other tools, I retouched:
Contrast: 10
Saturation: -10
For me the result is excellent, I show it here.
Obviously some of these settings work and make sense in this specific case, but they helped me understand a little more how to get around with RawTherapee and where to start in order to have a first neutral profile.
Thank you all! I have a lot to learn
ej
P.S. I know the 200% zoom is a little bit exagerated but I wanted to stress the problem of the multicolored grain that in some cases is very boring. Consider this is a shot at ISO 800, S 1/18, A 4 and FL 55.
OK, very good that youâre finding the right values for the plethora of sliders
Could you please post here a screenshot of the RT controls (the first tab, the one with exposure, saturations etc?) Itâs very strange that you should decrease saturation to mimic the other softwares.
This one is for ISO 4000 with an X-T30. With ISO800 itâs very rare to see color grain in most situtions.
Anyway, my point about noise and 200% zooms was just to emphasize that the color noise you see when zoomed is totally gone at 50%, so if youâre not cropping very hard, Iâd prefer not to denoise since it makes the picture lifeless, plastic, fake.
Not sure if I agree with your remarks about (not) zooming in and tackling noiseâŚ
When working on noise reduction zooming in is rather essential. At the beginning of that process I often times zoom to +/- 400% (or higher). This is the only way to get the correct demosaicing method dialled in and, if the noise allows it, set the correct capture sharpening.
If you (also) use the wavelets module to denoise you need to zoom in to 200% or 300% to be able to set the best results.
The other options, impulse noise reduction and noise reduction should be done at 100% view. RawTherapee does not always show accurate results when you do not do this. The modules that are sensitive to this have the 1:1 token next to it.
About things turning out plastic, waxy ,lifeless or fake; Hereâs an example of a rather high ISO image (20K) that was posted here on pixels that I tackled: Hockey game under lights: 20k ISO on m4/3
Only using the noise reduction module to tackle noise is often times not enough to get the job done. Most (all?) of the above is covered in the videos by Andy Astbury that I linked to in my previous reply.
Ok Marco, as absurd as this coincidence may seem, I also have the XT-30 and I shot with XF18-55 F2.8-4
These are two enlargements to 50%, one is from RawTherapee, the other from my current reference tool (which I would like to abandon in favor of RawTherapee).
I also add the screenshot you ask for.
Consider that I have not made any changes in the reference tool, everything is set to default.
Also, I tried to set âClip-out of gamutâŚâ and âHighlight reconstructionâ but without appreciable results.
Ibdont know what the other tools are âŚ
But yes , ACR / Lightroom always does a bit which you canât turn off . With the denoising sliders all the way down , there always is some. With the sharpening sliders all the way down, there always is some .
It seems they try to normalize all the sensors , and then after that give you some controls .
Itâs incredible subtle though, but if youâre comparing like this it might just be the thing.
And yes, every photo tool has a different look , also for color rendition , saturation and color contrast. So out of the box , you expect to see differences.
If you like what you get with little fiddling, thatâs up to you to decide.
I totally agree, it is clear that commercial image processing tools use âtricksâ to get you to the result you expect as quickly as possible.
I prefer not to declare the name of the tool I use as a reference because I donât want anyone to think that I am doing hidden advertising here.
I like RawTherapee for its origin and development philosophy, it also has some out-of-the-box features that I really appreciate: for example it doesnât claim to be the âownerâ of your albums, it provides you with a file with the parameters used for each photo, and so on.
At the same time itâs less intuitive and faster than my current reference, it produces that multicolored grain that I find in pretty much all photos and makes it a little difficult for me to produce smooth (in colors) and sharp (in details) photos. Iâm trying to find a solution.
I donât know or your RAF files are from a Bayer sensor or xtrans. But playing around with the demosaicing methods and the number of âcolor smooth passesâ can help, without touching the denoising parameters yet.
Also, resetting all the exposure parameters and curves on the first tab + first tool, can help . It might now be trying to get the look of the preview jpg. And in doing so might be upping the (color) contrast and highlighting the differences.
In my - short - experience rawtherapee does not show a lot or extra color grain out of the box . Or I never noticed it / never bothered by it.
But I donât have Fuji cameras.
Hi Jacques,
of course with VERY high ISO stuff noise reduction is almost a must.
Your result with that 20k ISO image was very good.
I just tend to shoot at ISO 3200-6400 and then almost always convert those images to BW that I really prefer.
Update
To be honest, the multicolored grain appears to be limited to a bunch of shots.
I donât know why⌠fog? wet film on the lens? critical pink color?
Anyway, with other photos I am getting processings that seem to me of excellent quality and without having to apply any noise reduction.
I am satisfied and, of course, ready for the next challenge
