USM parameter values in GIMP compared to Rawtherapee

I’ve been using both GIMP and Rawtherapee (RT) to develop photos. Sharpening seems to almost always be useful. However, while both programs support the unsharp mask method (USM) and use the same terminology for describing the controlling parameters there is no similarity in the values specified for said parameters. For example, they both use Radius where GIMP defaults to 3 and RT to .5. Then both have Threshold where GIMP defaults to 0 but RT has a set consisting of 20, 80, 1200 and 2000. Last they both accept a value for Amount where GIMP defaults to .5 and RT to 200. I tend to leave Threshold alone and fiddle with Radius and Amount.

It seems like there ought to be some relationship between the controlling parameters. For example, how do the values in one program relate to the other? It seems like it should be possible to understand a equivalence relationship between the values being used by each program.

Might someone be able to either shed some light on this discrepancy between GIMP and RT? Maybe reference some documentation that explains well enough to figure this out.

For RT 5.8, I recommend to use Capture Sharpening in raw tab (start with default (auto) values). If that’s not good enough, add USM with low values…

Capture Sharpening (Raw Tab) is NOT something I’ve yet tried but I welcome the advice. I’m a bit of a novice so please bare with me. It is my understanding that sharpening is something that should be last or near last in the development process. I normally think of the raw tab as the beginning but I think you may be saying that it can still be used near the end of the process. Would that be true?

However, my question really has to do with how to compare what should be the same technique when performed using different software applications. For this forum I thought it fair to compare GIMP with RT.

As a novice I’m using several different programs to develop photos where I ultimately compare the results I get from each application. While GIMP doesn’t do raw I frequently use it to do some post processing (crop, add borders, text, etc.) on photos produced from raw files developed by other programs. When it comes to sharpening I have also read where it can be better to apply sharpening multiple times rather than simply increasing the amount in a single operation. In addition to GIMP & RT I also use Canon’s Digital Photo Professional (DPP), which also has a USM tool and the same question arises with respect to the values it uses but I didn’t think it fair to raise that aspect of the question in this forum. I find DPP to be easier, primarily because there are fewer tools available. Therefore, a quick development with DPP is always done. I have had experiences where I’m unable to to get a better result with the other software.

Right now I’m throwing darts and pretty wildly. Sharpening definitely seems necessary but I have no idea if my similar intentions are being similarly applied when using the different applications.

Sharpening that removes image softness should occur early in the development process, while the image is still in linear space, so you don’t clip things. This is what Capture Sharpening does.

Sharpening that enhances edges for viewing should be done at the end, but with capture sharpening this is mainly useful if you resize down for web display, for example, or for printing where you are compensating for the printer’s lower perceived sharpness.

1 Like

You’re scratching the “semantics scab” of post processing… :smiley:

The internal mechanics of USM sharpening work like this: 1) make a copy of the image and blur it; 2) make another copy of the image and subtract the blurred image from it, then finally, 3) add the image from #2 to the original, and voila, sharpening!!

So, the relevant parameters to the gonkulator I described are in the blur operation, to which the relevant parameters are sigma, which in simplistic terms is the rapidity of the blurriness applied from the center pixel, and the radius, how far out should this blurriness be applied. Some raw processors present those two parameters for one to mess with, and those parameters can be pretty well carried from software to software as they’re anchored in a commonly understood math mechanism, the gaussian distribution.

But most folk don’t want to think of the math, they just want “more sharpen” or “less sharpen”. If that’s the slider you’re presented with, the mapping of that slider to sigma and radius is rather arbitrary, and probably inconsistent software-to-software. On top of that, and @heckflosse can describe this more succinctly, capture sharpening has other logic to selectively apply it across the image, and that logic can differ software-to-software. “More cowbell” is not universally appreciated… :face_with_raised_eyebrow:

I hope this helps…

Gimp does the USM in linear rgb color space while Rawtherapee use LAB color space

I think that is possible to match the gimp USM default (radius=3, amount=0.5,threshold=0) with this parameters in RT

um

radius=3, amount=0.5,threshold=1
um_t1

radius=3, amount=0.5,threshold=0.5(?)
um_t

Thanks to all for replies. They did include some tidbits that appear to be good to know that I did not know previously.

While I’m still without an answer I am beginning to understand why there may not be one. Or at least one that is simple/straightforward to understand (i.e., probably inconsistent software-to-software). Likely no reason to ask the same question on Canon forums.

Thanks again, I’ll get back to studying.

1 DPP allows you to get the Canon look. Processed raw images will have the same appearance as OOC JPGs. This is useful when you want to change the picture profile set at the time of shooting. Modern iterations of the software also have a few modest tools but I would rather use more capable software.

2 RT shouldn’t be grouped together with GIMP. GIMP is a general raster manipulator. It may have similar tools but a lot of the data and context is lost after exporting from a raw processor. Thus, it is important to be careful of what you are doing in RT from raw file to export if you intend to use GIMP (or in my case G’MIC).

This is spot on when it comes to my own findings. I develop raw with DPP just to have a base of comparison. This usually involves a little bit of tweaking to improve on what could be obtained directly from the camera.

I do develop all raw files with RT (GIMP cannot do that) but have occasionally found it difficult to get better result than from DPP. I think this is attributed to my being somewhat low on the learning curve and that RT offers so much to learn. It normally happens on photos of people where preservation of skin tones limits ones artistic (?) freedom. With landscapes and wildlife looking good and colorful takes precedence over looking natural.

GIMP is then used for other kinds of image manipulation after the picture has been developed. This includes sizing, cropping, sometimes adding borders and text.

What I haven’t settled on is how to handle what I’d call already developed pictures such as jpg files produced by other people’s cameras. I think I’ve learned enough about photo editing to feel like there is often opportunity for improving such pictures.

It’s challenging for most people especially when you have tones that the cameras aren’t calibrated on; e.g., people of colour.

This forum has been a hobby of mine. I started out knowing nothing that I do now. If you enjoy the pursuit of learning and helping other people, you will catch up in no time. :slight_smile:

1 Like