I’m wondering why you couldn’t just manually select the DCP using the Input Profile → Custom option in the Color Management tab? A bit of a PITA, but IMHO preferrable to making a separate DNG.
Oh, and could you make the selection above a part of a .xmp profile that you could make the default processing profile? I’m not a RT user, so this suggestion might not be valid…
Based on discussion herein I have decided to stop using AMCP for Input Profile selection and change to Custom.
The DNG conversion is about reformatting the new Canon (.CR3) raw format to Adobe Digital Negative (.dng) format. The reason being that, apparently, RT cannot handle the metadata in .CR3 format files. I’m hoping (don’t yet know) that the DNG Converter can include the metadata in a manner that it can be handled normally. The metadata problem directly affects my ability to use another suggestion offered herein regarding Dynamic Processing Profiles. I’m NOT smart enough to know what else might be effected by the metadata problem but I wouldn’t bet any money on the idea that Dynamic Profiles are the only other thing.
For sure if you create a profile that uses the custom profile in CMS and then an Adobe or custom profile you can create a Dynamic Profile that will work. I tested it and it does work for a CR3 converted to DNG…you also need to change the default for raw files to use Dynamic profiles otherwise the preferred is the automatch. My RT version is not new enough to open CR3 files so I could not see what metadata you get…if you do see the camera name then you should be able to use a dynamic profile with the CR3…but as I said my version was built back in Oct and it won’t recognize CR3 files at least the sample R5 file I used to try this out…
@ajax I applaud your patience: not every forum member has this virtue. I have been under the weather, so have not given any assistance. 
My advice would be to provide a sample raw file for us to demonstrate how to go about getting your preferred profiles to work. How well, we won’t know until we get our hands on something to show you the process involved. I think a hands on approach would help further your inquiry.
I just built the most recent RT…it will open the Canon EOS R…CR3 files but shows no exif so Dynamic Profiles and AutoMatched are a no go based on camera name or exif data for now it would seem. If those files are converted to DNG it seems like the exif data is available to RT and so dynamic profiles work and Automatched so based on the preference setting in RT one or the other will be preferentially applied. So the options are for now…use CR3 and have no exif data and create a profile based on a selected DCP and manually apply that to those files or you can use dynamic profiles/automatch if you convert them to DNG… From some quick tests using a downloaded Eos R image and the latest RT that is what I found…
Under the assumption that the issue is a lack of a DCP profile - How to create DCP color profiles - RawPedia
Although I think the original EOS R may have a DCP profile in RT dev?
Another possible issue may be that CR3 metadata support is limited so you may have to manually select the profille even if it exists. RawTherapee/rtdata/dcpprofiles/Canon EOS R.dcp at dev · Beep6581/RawTherapee · GitHub does exist in dev builds at least. The file is only 63KB so it might not be a full blown dual-illuminant DCP. Unfortunately it looks like at one point that profile was accidentally deleted and then re-added so it’s hard to see the history of it.
Yes, it looks like I now have it working and I can attest that .dng format files allow Dynamic Profiles to be used whereas NOT so with .CR3 files.
The version of RT in this case should be what is called the latest stable version (5.8). As it turns out my setup involves running RT in portable fashion and that does NOT require me to replace an old version when I update to a newer one. Therefore, right now I have the past 5 stable versions available for use. On those occasions when you want to fiddle with a previously developed photo, my thinking is that it is a bad idea to open older ones with a newer version of RT than what was used to create the resulting image file. You never know what might go wrong.
Thanks for all the help. At present I think the Dynamic Processing Profiles that allow use of the various custom Input Profiles provides an opportunity for making a nice improvement in my workflow. Also, discovering that the newer (.CR3) files for my newer camera are NOT properly supported is something I’m glad I learned about before investing much effort in developing them. These improvements should also be helpful for future development of photos from my older cameras and I have a pretty good size backlog.
I’m now thinking that with the help of contributors to this post I’ve solved my problem/s. However, to the extent that providing sample files would helpful for ongoing development I’d be glad to do that. Please be a bit more specific about what you’d like.
I think Canon (.CR3) raw files have been around for a while now and have pretty darn good market penetration. Getting RT to support them properly ought to be a pretty high priority. I’d be really disappointed if it weren’t for Adobe DNG Converter and RT’s support for such files. Kudos for that!
Per other replies I’ve now given up on using AMCP for Input Profile specification. However, consensus herein seems to be that such support will reappear once RT properly supports metadata for Canon (.CR3) raw files.
Sounds like progress. I actually ran it on WSL g in Win11 …it was just faster to compile on Linux with the auto script.
Sounds like you have a good workflow and if you want to modify any of the dcp files you can use the profile editor to tweak them to your liking…good luck going forward…depending on what you use in RT you should check out ART…very similar…simplified and some clever tools and tweaks and it might support the CR3 metadata…will check tonight…I’m a bit slow to sort things as I am still setting up a new PC from a crash on my old system a few weeks back. Just go the new one so trying to set everything up again…having said that it might not be of interest to you to look in to that but I thought I would just suggest it…
Well now, it didn’t take long to find another discrepancy. My method involves setting the camera to only record raw files. However, once the files I want to develop are selected I like to extract the preview image (i.e., small jpg that is camera developed). My understanding is that this is included (i.e., already developed) in the raw file by the camera. It is, I think, what you see on the camera’s display.
I don’t see any reason why it should be different when a .CR3 raw file is converted to .dng format but it quite conspicuously is different. What might be the explanation for that?
I believe the preview image may not be retained and/or Adobe inserts its own version. You could write/use a script to transfer the image from the raw file to the DNG.
I have not used it too much but there are option you can set when creating the dng upto and including embedding a copy of the original file…not sure what sort of preview control there is I would have to look
PS - On second thought, you may not need a script: DNG Converter may have the option to include the original preview files. If RT doesn’t find embedded images, it will create its own preview. Perhaps, that is why the image is different.
Correct #1. However, the issue is a little more complicated. The camera name is identified correctly by RT, but not for automatic DCP loading. You can check that RT knows the camera by either turning on verbose mode and look at the output in the CLI, or make a crazy edit in the EOS R color matrix in camconst.json. This will change the colors of your image.
This indicates that a ‘simple’ fix could be possible to enable DCP loading as well, if somebody (you?) wants to look into that.
Correct #2
Correct #3, but with a little more digging I found that the EOS R profile was imported from ART in this commit.
Hmm, yeah something for the TODO list now that it’s getting warmer and less COVID-y and I’m using the camera more. There’s also some oddball lens metadata issue I’ve noticed that I want to poke at. Haven’t even filed an issue on it yet to track it, and there might be an existing metadata-related issue. I admit I’ve been partly waiting until after Adopt Exiv2 for metadata handling by Beep6581 · Pull Request #5889 · Beep6581/RawTherapee · GitHub is finished (and TBH I should try that PR locally to see if it resolves the issue…)
In fact I should test that PR to see if it resolves the EOS R AMCP issue too…
Edit: Getting slightly off topic, but can you clarify Canon EF 70-300mm f/4-5.6 IS II USM reported as "Unknown 70-300" · Issue #6224 · Beep6581/RawTherapee · GitHub - I think I need to do the same for the Sony 24-105G - dunno if Canon EF 70-300mm f/4-5.6 IS II USM reported as "Unknown 70-300" · Issue #6224 · Beep6581/RawTherapee · GitHub was actually done, since there’s no commit or PR that references that issue.
Decided to undertake some experiments which did take some time.
First, based on feedback obtained herein I did notice that there are options set using preferences that pertain to the preview image when using Adobe DNG Converter (ADC). Since documentation seems to be in short supply, at least as far as I’ve been able to tell, it was NOT very clear, to me, what purpose was served by these options. Therefore, I decided it was necessary to try each one and judge the results.
Therefore, starting with a raw file from my camera I made pair of .dng files. One specified “Medium Size” preview image the other specified “Full Size” preview image. The option “None” for preview image was NOT tested. Also, an option to “Embed Fast Load Data” seemed to be contrary to what I’m after so it was turned off. Then I used ExifTool to extract the preview image from both .dng files as well as the original raw file. Also extracted metadata from all of the preview images.
Bottom line is that both preview images from the .dng files were significantly different than the one from the raw file. This was true with respect to both size and appearance which was judged using a program called XnViewMP. I’m pretty sure that the preview image in the raw file was developed by the camera which is what I’m looking for. If there is a camera developed preview image in the .dng file I haven’t figured out how to get it. However, I’m starting to think it just isn’t there. Given that I now know about this, it isn’t a serious problem since I can get the preview image I want from the camera produced raw file. However, given that my expectation was that the .dng format contained the same data as the original raw file, just formatted differently, I have to conclude that simply is NOT true. It also makes me very skeptical about the authenticity of the rest of the data in .dng file. As a result I’m not too keen on investing much time and effort into learning how to use Adobe Digital Negatives.
Also, I have downloaded what looks to be a pretty recent release of ART. Maybe that provides a better solution than Adobe Digital Negatives.
The data resulting from my experiments can be found here. Of particular interest might be a comparison chart that includes what I’d call the factual data as opposed the more subjective idea of comparing what is seen.
Okay, so my penultimate comment stands. Could you not copy the tags and preview over from the raw file into the DNG?
In general, I keep the raw file no matter what, so I don’t need a proper preview personally. To me, it just takes up storage.
I think that’s what I meant when saying, “Given that I now know about this, it isn’t a serious problem since I can get the preview image I want from the camera produced raw file.”.
While I’ve been doing this for a while I still consider that I’m learning how to develop raw images. Knowing what the camera would have produced helps to understand whether or NOT I really achieved some improvement for what ends up being a fair amount of work. Obviously, another way to get this would be to let the camera produce both jpg and raw. Since I develop only a small portion of the shots I take I prefer getting the jpg only for the ones I choose to develop. Also, in addition to the raw file I typically develop at least 2 16bit uncompressed .tiff files (one with the Canon software and one with RT). Then I do things like like size to print, add borders and text (i.e., transformations if you will) using GIMP which has it’s own proprietary file format which is similar in size to raw files. By the time I get done with that the jpg Preview Image consumes a negligible amount of storage.
Did get ART installed and low and behold it supports my camera, out of the box, to include Auto Matched Camera Profile. Therefore, I will be giving it a try. With a little luck it might carry me, at least until, until RT does support my new camera. If so there may be no need for digital negative files. Even though Auto Matched Camera Profile is supported, based on discussion herein, I’ll be trying Dynamic Profiles as an alternative.
Give Art a shot…it has nice masking as well and lots of clever little add-ons to RT. you might find it is all you need…I love the tone mapping module and the tone map it displays is really nice to target tonal ranges