Larry Baum

Does anyone here think they have any DPX16 or DNG files from a Scanstation 6.5K stored in reference to some wide gamut format (like P3 or REC2020 or more likely native scanner gamut primary locations)? And again, I am NOT referring to the tonal curve of REC2020 or bit depth or LOG or flat or whatnot, simply about the location of the color primaries and whether the color gamut is wide or standard broadcast. Like how the new UHD format blu-rays not just have more bits per channel and have HDR but also store colors in reference to the very wide color gamut of REC2020 even if mostly restricted to the P3 color gamut within that which is still a good deal wider than the old REC709 color gamut of blu-ray and HD broadcast TV. Like how you can look at intensely colored fall foliage and flowers and sunsets and tropical waters and see much more intense colors if you view pics taken with a DSLR and the images processed in a wide gamut color format and then view them on a monitor set to wide gamut format (even if it is still in SDR mode). Oh, also the metadata in the DNG samples I received has the ColorimetricReference tag set to 1 which means display referred and not scene referred. This triggers Adobe ACR to load the DNGs RAWs as pseudo-RAWs and changes the color temp slider from Kelvin scale to a points scale instead like if you force jpgs to be loaded into Adobe ACR. Which also seems strange. I downloaded some video shot on someone's drone and output as DNG and it had that set to 0 and the same for shooting MLR on a DSLR and then AdobeACR loads them as true RAW and gives you the normal Kelvin slider for color temp.

Is there some setting somewhere on the Scanstation 6.5K that sets either native scanner color gamut vs. old broadcast standard REC709 color gamut that maybe the scan operator can't seem to find, that is maybe labelled in some weird, confusing way that makes it hard to find?

images that I know to be in sRGB/REC709 seem to behave (color gamut-wise, again NOT talking about bit depth or fine gradations or LOG tonal curves and whatnot) just like all the sample outputs from the Scanstation that I got. And they don't appear to behave at all like any images that I know to be in wide gamut format (like from my DSLR or stills scanner when used to output in some wide gamut format).

Well as I said: If I look at the DNGs the samples all have in the metadata a ForwardMatrix1 that is very close to the standard sRGB/REC709 color primary standard forward matrix, just a tiny bit different (and that might be just down to a D50 vs D65 white point). And if change the entries in the ForwardMatrix1 in the file and then reload it then Adobe programs and Resolve both load them with crazy colors so that matrix IS being used by those programs to interpret the color primaries otherwise messing around and putting weird values into it wouldn't make any difference. The matrix the files came with is very far from a really wide gamut matrix like ProPhotoRGB would have and far off from even what a P3 matrix would be like. Also if I go into any RAW converter program and simply assign the DNGs a colorspace of sRGB then the colors look just the same as the DNGs do loaded into Adobe or Resolve (minus perhaps a D50 to D65 shift). If I load say one of the DPX16 files into Photoshop it complains that there is no colorspace associated with the file, if I tell it to assign sRGB then it loads into Photoshop looking pretty much how it loads into Resolve or Premiere Pro and looks pretty much like the DNGs did loaded into Adobe ACR (minus the 1500K difference here) or Resolve and looks pretty similar to the ProRes4444 output from the scanner and pretty similar to how the DPX10LOG looks loaded into PremierePro with the cube LUT the scanner made for it. If I tell it to assign P3 then it looks different and if I tell it to assign REC2020 or ProPhotoRGB then it looks way different and very strange. If I have my display set to REC709 and assign a profile of sRGB to the DNG and view it in an image viewer the image looks pretty normal and pretty close to how it loads into Resolve. If I do the same with an image that I know is actually say stored in reference to say ProPhotoRGB wide gamut primary locations then it looks way dull and tinted. If I change my display to native wide gamut color mode and assign a profile of ProPhotoRGB to the DNG and view it it looks radioactive. If I do the same for any sRGB JPH I have the same it then looks radioactive. OTOH if I do the same to a photo I know is in some wide gamut format it looks reasonably normal (a little hot if it is in say P3 which is still a fair bit smaller than PropPhotoRGB). Maybe I am somehow mixing something up, but I can't think what. If the files are not locked into a restricted sRGB/REC709 color gamut then how come the DNG have a more or less sRGB color gamut matrix associated with them for color interpretation and how come if I simply assing them an sRGB/REC709 color gamut the files act nearly the same and how come if I assing that to DPX and load into Photoshop it then shows the same colors as they do loading into Resolve or Premiere Pro? I could understand if the ForwardMatrix was just nonsense meant to be ignore for these DNGs or associated just with an embedded JPG preview but exiftool says the DNGs from the scanner have no embedded JPG preview and if I change the entries in the matrix then Adobe ACR and Resolve load the images all messed up with crazy colors.

Thanks. But actually that is not what I am talking about, sorry for any misunderstanding. I'm NOT talking about the tonal response curve or bit depth or banding/posterization or getting subtle micro-tones of color differentiation or any of the things I see talked about most often. I am talking simply about what color primaries the data are stored in reference to and surprised that so far there seems to be no way for the scanner to find a way to get wide gamut color data out of the scanner. Everything seems to have colors within REC709. Everything about the files implies they are locked into a small sRGB/REC709 gamut rather than something large enough to hold all the colors that motion picture film can potentially show. And I even see like in one scene with a big orange-red orb how on a gamut plot it's all straight line hard edge against border of the G/R line of sRGB/REC709 (and the file itself has the interpretation matrix that of sRGB/REC709 colors instead of something wide like native gamut of the scanner or something or even P3). I'm not having a problem with getting subtle shade differences or color differences, I'm having trouble getting files that contain colors referenced to a gamut wide enough to properly contain motion picture film's full palette. It all seems to be stored in reference to the old broadcast video standard palette and thus can't show any colors beyond that at all.

Thanks. They were 5k mini test scans from the 6.5K machine under 2 flash. It's not the tonal curves or crunchiness or whether shadows are all there or highlight captured well or there is lots of fine micro-tones and such that are under question in my case. I'm simply wondering about the color gamut, as in the location of the three color primaries that the data is stored in reference to in the files. As in can it capture the intensely saturated cyans, yellows, redorange etc. that film can that is beyond what sRGB/REC709 gamut can handle or is the scanner simply, as hard to believe as it would seem, actually locked into some old school broadcast restricted color palette that is smaller than what film or say P3 can handle. My question is not regarding the usual talk about whether shadows have detail or highlights are blown are nicely separated or bit depth, etc. The samples looked fine enough other than for they seemed to have the color palette clipped to sRGB/REC709 as really intense cyan and red-orange in some frames which just straight line clipped along sRGB/REC709, the film itself clearly had had richer colors than the small, standard sRGB/REC709 gamut could hold (again nothing to do with HDR or shadows or highlights or noise or anything like that, talking color palette primary locations) and more than that what metadata was in some of the files seemed to be sRGB color palette metadata (not necessarily for the tone response curve and certainly not REC709's 8bits, just talking the location of the color primaries).

On the face of it the question may seem absurd, since with film having a larger gamut than sRGB/REC709 and with stills film scanners all handling wide gamut color scans for ages, I'd think a mega expensive pro motion picture film scanner would certainly be able to output in wide gamut color but so far, as best as I can tell, each little sample I have received seems to be locked into sRGB/REC709 primary locations. It doesn't seem to matter whether Print or PrintLog option was used or whether it is DPX16, DPX10, DNG or ProRes4444. If I look at the DNG output in an exif viewer and look at the Forward Matrix it seems be one incredibly close to sRGB/REC709 primary locations (perhaps just slightly twisted since it seems to be in reference to D50 values instead of D65). If I force programs to assign color primary locations of sRGB/REC709, all files seem to look about the same as just letting programs load them as default. Also I see that the DNG file has the ColorimetricReference tag set to display referred (like a JPG) rather than scene referred (like most true RAW files would). If I have a display set to REC709 and assign the file to be interpreted as if it is REC709 it looks pretty normal and not all washed out and slightly twisted in color if it was stored in reference to wide gamut color primary locations. Like if you take a ProPhotoRGB image and do that it looks horribly dull. And if switch the display to native wide gamut and view one of the samples assigned to be interpreted as REC709 it look radioactive instead of at least somewhat normal. If I try to load a DPX16 frame into Photoshop it says no color profile found for file, if I assign it to be sRGB it looks pretty normal and about how it looks loaded into Resolve or PremierePro. Now sure the files are stored in more than 8bits per channel and some samples were log like DPX10LOG and so the gradations are finer than 8bit SDR REC709 and the tonal curve may be way different for like the DPX10LOG files and such, but the color primary locations just seem awfully like they are sRGB/REC709. One odd side note is that Adobe ACR loads the DNG output looking about 1500K color temp different than Resolve does (interestingly the AsShotXY is D50 as is the shot light source and sRGB/REC709 are based on D65, so 1500K apart so it seems some programs are maybe accounting for that and some not, with the files being listed as display referred, maybe Adobe is technically correct in that it treats them only as pseudo-RAW and doesn't adjust any color temp stuff at all and just loads everything exactly as is once the ForwardMatrix and AnalogBalance matrices are applied, although I suspect the 1500K different look of the ProRes samples and how they load into Resolve or the DPX16 loads into PremierePro or the DPX10Log loads into PremierePro with the LUT applied). So for all these reasons I feel like the scanner seems to be most likely only giving sRGB/REC709 standard gamut locked color primary referenced output and clipping what the film is capable of (in some frames you even see the gamut plot just straight line up against the edge of sRGB/REC709 gamut which means the original film surely extended past that gamut). Does anyone have any idea what could be going on? Is there some special setting that the scanner is missing setting in order to get native gamut output from the scanner? Again I am not talking about bit depth or fineness of gradations or LOG vs this or that or anything to do with tonal curve I am talking about color palette, where the locations of the three primaries are set. Now maybe the film is way more saturated than I think and it really is somewhat wide gamut, maybe a little shy of P3 or something but still looks kinda normal treated as sRGB/REC709 color primaries, but then how come the ForwardMatrix in the DNG basically appears to reference sRGB gamut? And if it is some custom semi-wide gamut, then how come some of the files have no reference as to where the primaries should be? How could anything know what the DPX16 data or what not is supposed to be in reference to? Maybe they just default to guessing REC709 if nothing else is there to go by but then how does the user know what to do with the files? And how come the cubeLUT the scanner spit out for the DPX10LOG seems to give the same color look as REC709 primaries? Thanks.