Andy i have seen you commenting on a few forums re Pixel phones…just picked up a 3aXL as they are dirt cheap and now just trying to understand the nuts and bolts of their DNG files…Do they have embedded profiles that will be recognized or can be extracted?? I could use the DCP files with Rawtherapee but I am still trying to understand what RT does automatically re embedded profiles and or the TC settings in exposure and do those conflict with the TC being enabled in the DCP etc etc…also was wondering about extracting to create a ICC to use with Darktable…again acting from a relative point of ignorance on the subject and just trying to understand and correctly establish a workflow…even determine if it is worth it given the processing that goes on in the pixel phones…any insight on your part or links to info would be appreciated…Thanks in advance…
Todd, if the DNG has a D65 ColorMatrix tag, those numbers can be munged into a dcraw-style primary set and, for RT, inserted in camconst.json or, for dt, I think you’d need to put them in the appropriate source code data structure and re-compile.
In rawproc, you can just paste the comma-delimited set in filename box in the colorspace tool…
Whups… the ColorMatrix1’s corresponding CalibrationIlluminant1 is D55. This’d require something on the order of a Bradford transform; @Morgan_Hardwood, how would you handle this?
Actually that one I pulled from the rawspeed collection…mine appear to be a newer version and set up differently
Better… CalibrationIlluminant2 is D65, so you can take the ColorMatrix2, multiply each number by 10000, and use the resulting integers to make a comma-separated list for use as dcraw-style primaries.
Thanks Glen…what is standard light A??
FYI I made this its own thread so we can keep the other one related to collecting raw files. We link to that post quite a bit. Thanks for your understanding!
Good idea
I do have them…they do give a different result than the embedded one…that is for RT and they are for the pixel 3 not 3a so there could be slight differences I suppose. Dark table seems to be seeing the embedded ICC and enhanced matrix for JPG and raw …I
tried to convert the Adobe ones to ICC to use with DT but the files are really small and I messed up the camera name so I need to give it another try
Thanks Glen
It seems to be a name miss match for the camera between the dcp and the dng file information…
Just use the dng as a profile of itself, it should work (I mean select ‘custom’ profile and pick the dng). (Fwiw art will do this by default, but the code can be easily ported to RT)
Thanks Alberto, It does as you say work with Art…if Art is like RT then I think it assigns a camera profile to the image if it has one and there is an embedded profile as it is a DNG. Then if I use the custom I can apply the Adobe one…so good on that front. I have been trying to make an icc from the Adobe DCP to use with DT just to compare that to the internal one which I think is what DT uses when you select embedded matrix in the input profile. Pixel JPG files also have an embedded icc so not sure how all the programs use it but i would suggest they use it unless something else is specified…In the end I may just stick with the embedded one although to me the WB is a bit to the warm side in my initial opinion…also the DCP from Adobe is for the Pixel 3 and I have the 3a so I think they are pretty similar wrt to the main camera …
Somehow I never got the notification for this… I haven’t logged in for a while (Haven’t touched my camera in a while due to COVID and some non-photography projects stemming from buying an electric vehicle…)
In general, I found that the embedded DNG profile of “production release” Google Camera versions on the Pixels seems to be pretty good - enough that I’ve never bothered re-profiling my Pixel.
While you’d expect the DNG profile for a given device to be embedded somewhere in the phone/s rootfs partition, the Google Camera app seems to be where the profile data for shots taken by it is stored - as a result, a change to the app can result in a change to the embedded DNG profiles.
In my case, my Pixel 4 XL arrived 1 day before it was officially supposed to be released - and it turns out that there were a LOT of Day-0 updates planned by Google not available on Day -1 , including a Google Camera app update that completely changed the embedded profile.
The only shots I’ve needed to replace the embedded profile on were shots taken on Day -1 - if you’ve got a scenario like this, I can dig through what I figured out to extract the profile from a newer DNG and use RT to override the profile for older ones. (IIRC, dcpTool can extract the embedded profile from a DNG and save it to a standalone DCP profile.)
Thanks Andy…RT has the most straightforward approach to this and so I used it a bit to trouble shoot color management…but I use Darktale primarily so hence the need to get to an ICC…I am about to establish and compare using Dcamprof to convert an Adobe DCP and extract the one from the pixel DNG also try darktable chart approach and Argyll with color charts. There is a got script color match which I might try…like you co-vid derailed my efforts …wondering if tinkering with raw files on a pixel is worthless given all the math used in camera to create the JPGs…
Hi,
I don’t know about later models, but pixel2 jpgs are ugly to my eyes, whereas the raws are pretty good