I am a RT newbie and I try to achieve a result similar to Adobe Camera RAW (ACR).
In ACR, I open the RAW image, shot with Canon 400D, I simply press the “Auto” button and save the image as JPG with maximum of quality. I have attached the source, and the resulted the sidecar and jpg image. IMG_2682.CR2 (11.7 MB)
However I try, I can’t tune the same RAW file in RT as close as possible to the simple ACR result by pressing “Auto”. Especially the sharpness, not to say the colors…
How could I do this? I know is a different processing in RT, but still, how could I get this as described?
Hi, I have tried a lot the settings, read articles, so I have come here.
What do you mean by “Your question comes up all of the time”?
As for “do share it so that we could give you advice”, I have attached the source raw and the ACR files. What else?
If you read the two links and the forum, the general answer is that you won’t be able to match the ACR result because it is a different piece of software. You could come close but it might not even be desirable since RT, dt and friends often can do a lot better! That is, once you learn them, which could be a long journey.
If you have RT installed (let us know which version!), try to do an edit and drag and drop the sidecar (*.pp3) into a post so that we can examine what you did. Then we can see where you are struggling and give advice.
Thanks, now, without any hard-feelings or maliciousness, or any negative connotation, ACR was very natural and handy for me, but RT not so much. I try to find out the way now.
Here is a pp3, I’ve tried to do a simple processing. IMG_2682.CR2.pp3 (11.0 KB)
My struggle is related to sharpness, then to find a better color match to the ACR one.
Of course, I don’t expect a 1:1 match, only as close as possible.
Any better color match to your taste is also welcome. But now, the colors look very saturated/hard/vivid I can’t fix even with the saturation sliders.
I use RT 5.6 under Sabayon Linux.
That’s probably because it is overexposed, but recovers acceptably. It is not a commercial photo, I just try to process my holiday photos and this is one of the worst I have, but nice to test one. I have a bunch to make a pano with Hugin, I try to make a profile to feed Hugin and to resynth the result a bit in Gimp.
What @afre was referring to (@afre, correct me if I’m wrong here) is that the LightRoom creative cloud thing has driven a lot of folk to try the open-source alternatives; they try to do the same with RawTherapee, and when they don’t get the same rendition, they come asking for help. It’s okay, the folk here are generally helpful souls.
You’re going to come to two realizations:
What ACR’s ‘auto’ button gives you is one rendition, and it happens to please your sensibilities. The other sofwares can do similar, but they’ll be different renditions, so you can’t expect them to look identical.
The open-source softwares are less about doing the work for you and more about you making intelligent choices in your processing. At a minimum, you’ll eventually have to learn about tone curves in all their manifestations, from level sliders to control-point shaped curves, in order to get control over your raw data. What you’re doing in that learning is freeing yourself from the ACR programmer’s opinion regarding what your image is to look like, and taking control of it yourself.
Your difference endeavor intrigued me, so I attempted to duplicate your ACR JPEG in my hack software, rawproc. I did it there in order to completely eliminate any hidden processing, and then explicitly apply what it took to approximate the JPEG. Here’s the compare screenshot, the rawproc working image on the left, the ACR JPEG on the right:
If you can read the commands pane on the right, it’s the camera color profile, the camera black subtract, the camera white balance, and @heckflosse’s AMAZE demosaic implementation, then a black/whitepoint scale of the raw data to fit the display black/white. That’s about as close to “Neutral” as you can get to obtain a starting raw image. So, the tool you see displayed in the bottom-left is a tone curve, and this is what it took to make the tones approximate the JPEG. I think your perception of sharpness may be differences in contrast.
Now, this isn’t really a fair comparison, because saving to JPEG does its own changes to the image. Here’s a JPEG-JPEG screenshot:
Well, I don’t use LR, I find it so uncomfortable. I use ACR, which uses the same API as LR, but differently. ACR offers me a nice starting point - according to my taste.
Regarding the software… There are technical guys and simple users, who don’t need such refined details. They need simple controls, as a few sliders… This is the reason why powerful tools are not easily adopted these days, and thinking to those simple users needs consistent efforts, with the gain of having an increasing user base. This is the story of Linux, Linux DE’s and lots of free/open software. In time, the first impression remains in their heads and won’t come easily to use software that made them suffer once.
Now, what I have struggle with here, was, in part, the unresponsiveness of the edit boxes and made me think they have no effect. Some timer to decide the user have ended editing some value and start applying it would be nice.
On other hand, to help the migration, also my case, some profile for the initial automatic processing, with a very close result to some common commercial tools used these days, is very useful. You give to the new users something familiar that makes them stay. Just my two cents…
@msdobrescu The fact, that I had to adjust the raw white point, means that the white level for your camera is probably not correct in RT. Can you provide some completely overexposed raw files from your camera using filebin.net and post the link here?
I did something, which usually should not be necessary if the white point for the camera in RT is correct. If the white point from dcraw or camconst.json is too large you get this pinkish tint when applying negative exposure compensation. In this case the white point was too low, means, to recover as much highlights as possible, I increased it by reducing the value of the white point correction adjuster until it was too much (pink) and then increased the adjuster value again by 0.01.