Graininess of RAW versus PDF

Dear all, I’m learning RAWTherapee step by step and “already” at the stage of having fun, despite still having some challenges :-). Pls find enclosed a zoomed-in art-shot (a museum painting, now-a-days unfortunately having a glass-plate in front). The RAW editted file continues to display some sort of graininess that I can’t get rid off (versus the JPEG file constructed by my camera). Any suggestions have I can reduce this graininess? (I have used haze removal, yet I don’t think that has any effect on this graininess). Thank you for advice / reflections.

Editted PDF.jpg.out.pp3 (14.9 KB)

Editted RAW file.jpg.out.pp3 (14.9 KB)

Welcome to the forum!

I think the In camera processing applied strong noise reduction. You can do something similar in RAWTherapee. If you give your image a creative commons license, you could also upload your RAW file as a play raw. Then other users will show how they would handle the image.

P.S. Should the title not be “RAW versus JPEG”?


Hi Thomas, of course happy to share the original file. It is far from perfect (abberations a.o. caused by the cover glass) but good enough to start practicing my RAWTherapee skills. RAW File is enclosed and happy to receive any tips & tricks. I enclosed the .pp3 file as well.
DSC09116.ARW (20.0 MB)
DSC09116.ARW.pp3 (14.9 KB)

Please add a license to your RAW file, so that users may work on it. Something like this:

This file is licensed Creative Commons, By-Attribution, Share-Alike.

DSC09116.jpg.out.pp3 (17.1 KB)

Stronger noise reduction than I would like but an example.

  1. Adjusted curve
  2. Local edit noise reduction
  3. Capture sharpening off

Stronger NR will blur the grainy details in the table as well.

Hi Oscar,
welcome to the forum. My experience is that when you first start developing RAW files you initially struggle to match the quality of the out of camera Jpgs so they set a good bench mark. But then when you become competent and experienced at using the editing tools you soon realize that you can do a much better job than the camera’s jpgs 99.9% of the time. The camera tends to apply a lot of denoising to high ISO images and this gives a smoother look, but when you start editing the raw files yourself then you come to realise that a huge amount of detail is thrown away in the process. The images may look as smooth as a baby’s bum but is that what we really want or need. This is particularly the case with my Olympus TG6. I can’t comment on your camera in detail.

I would also avoid pixel peaking like you are doing here. Judge the image globally so you can see the effect on the whole image. A person is not going to inspect the final image with a magnifying glass which is what you are doing here. I am a darktable user myself and I opened this image in RawTherapee and took a look. RT is very foreign to me so I couldn’t edit the image well, but even without playing around much I could see that noise was not a problem with this image. I wonder if you introduced noise by aggressive sharpening?


DSC09116.ARW.pp3 (14.7 KB)

i agree with terry, to avoid pixel-peeping. The pic in total isn’t such noisy. As long as youre not pushing the modules and functions (by the way, haze removal can make noise more obvious) the result will be very acceptable low noise with good details.

DSC09116.ARW.jpg.out.pp3 (15,0 KB)


Thank you nosle, nice edit.
Can you elaborate a little bit on the steps you took?

  1. Adjusted curve (color curve? / what edit did you specifically made?)
  2. Did you simply switch on local edit noise reduction? or did you adjust the sliders ? (so far, I have not used this tool yet)
  3. Capture sharpening off (I will look for this option, but pointers where I should look are most welcome!).
    Thank you for your suggestions & help.

Thank you for your reflections Terry.
Fully agree with your remarks and indeed I

Pixel peaking is not my favorite as well :-), but a step in my learning journey to analyse what gouing on in this picture. The effect in this mirrir is seen directly when opening the RAW file, so pretty much pre-processing (although RT uses standard processing for converting this RAW in a picture). I guess it does not apply aggressive sharpening in that process, right? I also like your

[quote=“Terry, post:6, topic:42203”]
as smooth as a baby’s bum but is that what we really want or need.
[/quote]. I’m still at the state Ithat I like the smoothness. Actually I’m editting the JPG with RT instead of the RAW files (not sure whether that makes a lot of sense, yet it gives reasonable results). Hopefully I will transition on the short term to editting RAW’s. Thanks for your reflections & support.

1 Like

Wow, that looks pretty impressive Todd. How can I " learn" what you have applied to this RAW? I will download the pp3 and see whether I can extract what edits you have made, but I’m not sure whether it works that way.

Thanks Martin!

Hello Oscar,

I saw, by loading your pp3, that you activated the Noise Reduction module, but the sliders for Luminance and Luminance Detail were set to 0. That means that no noise reduction is applied.

Set the first slider to eg. 20, the second to eg. 70 and you’ll get this.

I don’t use RT that much …more DT and ART but I thought I would mess around and see how the new release behaved. It seemed very fast and I was getting some really nice results from a few simple changes in the few photos that I experimented with…

Your image I was sort of playing around and that is what I landed on… if its not obvious from the sidecar I will go back review it myself and try to answer why I did what for you if you have any unresolved queries…

It really was a sort of push and pull edit just testing a few things…

My attempt in GIMP 2.10 and GMIC

I pulled the blacks down a bit to create more contrast . I did adjust the noise sliders, just pull them around and check the results. Capture sharpening is at the bottom of the raw tab (mosaic icon). It’s a great feature but for noisy images its better off.

DSC09116.ARW.xmp (12.2 KB)

What noise? :thinking: :slightly_smiling_face:

Rather than process your image, I examined the raw data with RawDigger, voila:

Wasn’t able to notice significant “graininess” (RawDigger uses ‘LibRaw’ to convert to RGB with no adjustments allowed). Looks like there’s quite a bit of over-exposure - and cognoscenti here can comment more about the raw selection histogram and channel parameters.

Participants upload their processing in sidecar files (pp3 for RawTherapee), which you can use in the software.

I’m not an active RT user, so I cannot help you further. Hopefully provides info, and other RT users certainly will, if needed.