- This topic has 9 replies, 2 voices, and was last updated 3 years, 5 months ago by
.
-
Posts
-
Limit the color gamut in anyway, like will it reduce the gamut on the monitor to sRGB, if so how can i make it so sRGB content is shown accurately (not saturated) but at the same time not limit the gamut when either photos with embedded wide color profile is displayed, or when viewing HDR > SDR movies which supposedly brings the wider color gamut but not HDR so basically SDR2020.
Do i need to make some customized ICC profile for this taking into account that all the grayscale, display characterization will be in the HW 3DLUT?
HW cal with (good) “X colorspace simulation” in a memory slot Y:
-display behaves like X as long as monitor is configured to use slot Y
-generated ICC profile looks like X
-when showing an image with a colorspace Z much bigger than X in a color managed app IT WILL CLIP the colors of Z outside X (rel col from image to display), so your statement after “(not saturated)” is false as long as monitor is on memory slot Y.
-Color managed app in the statement above will always need a display ICC: the one made during HW cal, a new custom one or a generic one for X colorspace.HDR modes are NOT actually “colorspace simulations”, they are “colorspace translators”. They work as the 3rd statement above, like a color managed app, where Z is rec2020 and X is native gamut.
-display DOES NOT behave like Z, display accepts signals in Z
-display deforms colors in Z to look good in X the best it can.HW cal with (good) “X colorspace simulation” in a memory slot Y:
-display behaves like X as long as monitor is configured to use slot Y
-generated ICC profile looks like X
-when showing an image with a colorspace Z much bigger than X in a color managed app IT WILL CLIP the colors of Z outside X (rel col from image to display), so your statement after “(not saturated)” is false as long as monitor is on memory slot Y.
-Color managed app in the statement above will always need a display ICC: the one made during HW cal, a new custom one or a generic one for X colorspace.HDR modes are NOT actually “colorspace simulations”, they are “colorspace translators”. They work as the 3rd statement above, like a color managed app, where Z is rec2020 and X is native gamut.
-display DOES NOT behave like Z, display accepts signals in Z
-display deforms colors in Z to look good in X the best it can.I should have clarified that when I said HDR > SDR I meant HDR videos that only use the wide color gamut without displays actually entering HDR, for example when MADVR playback HDR videos on displays which can’t accept HDR signal.
Using the uploaded 3DLUT clips everything to sRGB which is not what I really want :/
So is there really no good way to have 3DLUT sRGB emulation for sRGB content which is most content, but leave WCG photo/video content seamlessly intact without needing to switch settings/3DLUT’s each time?
You need 2 LUT3D in madVR. One for rec709 content (simulation rec709 from widegamut) and one for HDR (simulating the part you cant see from rec2020). You can make them from the sam displayprofile.
This way display is always in widegamut mode, you use a suitable ICC profile for display in OS settings, Firefox, Adobe &etc shows color as intended (you may need some config). When you use a madvr compatible player it detects or guess content colorspace and uses one LUT3D or another. Full auto, configure once then forget. There should be up to 8 LUT3D slots in madVR for each type of content although you usually need 2 (rec709 SDR & rec2020 HDR).If you meant having an sRGB emulation for games and such, you need to use 2 memory slots / presets in your monitor if it has that. One sRGB (factory or user mode “1” with saturation controls), one for native gamut. When you play games you need to swicth to sRGB one and switch back once you end. You may need to change display profile in OS two since these two modes may have very different GPU calibrations in their ICC profiles.
You need 2 LUT3D in madVR. One for rec709 content (simulation rec709 from widegamut) and one for HDR (simulating the part you cant see from rec2020). You can make them from the sam displayprofile.
This way display is always in widegamut mode, you use a suitable ICC profile for display in OS settings, Firefox, Adobe &etc shows color as intended (you may need some config). When you use a madvr compatible player it detects or guess content colorspace and uses one LUT3D or another. Full auto, configure once then forget. There should be up to 8 LUT3D slots in madVR for each type of content although you usually need 2 (rec709 SDR & rec2020 HDR).If you meant having an sRGB emulation for games and such, you need to use 2 memory slots / presets in your monitor if it has that. One sRGB (factory or user mode “1” with saturation controls), one for native gamut. When you play games you need to swicth to sRGB one and switch back once you end. You may need to change display profile in OS two since these two modes may have very different GPU calibrations in their ICC profiles.
Yeah I was afraid of that seems that I really need to change 3DLUT slot/ profile in my monitor each time I want to either view sRGB or wide color content if using the HW 3DLUT functionality, so no way to view sRGB content, and WCG content seamlessly at the same time in all applications.
Or would the procedure described below work?
Make a HW 3DLUT that characterizes the display to its native gamut, and then make a simple ICC profile for the CMM (Adobe, qcms, lcms etc) to use for sRGB emulation as described here:
https://www.lightspace.lightillusion.com/icc_profiles.html
For anyone who really cares about colour accuracy their display of choice should always have in-built 3D LUT capabilities. Without this, any colour accuracy will be limited for all the reasons outlined above, especially when using an ICC based workflow.
And with a 3D LUT capable display there is a much better, more accurate, and inherently simpler workflow that can be adopted.
The display should be accurately calibrated using LightSpace CMS, using a 3D LUT loaded directly into the display. This calibration can be to a given colour space standard, such as Rec709, P3, sRGB, Adobe RGB, etc. Or, the calibration can be targeted to the displays native gamut, for applications that are not for film & TV deliverables (i.e., for photographic workflows).
The associated Display ICC profile can then be a simple curve and matrix profile, or curve and colourant profile, simply defining the display’s calibration.
For example, if the display has been 3D LUT calibrated to Rec709, the ICC profile would define the gamma as being 2.4, and the gamut as being, well, Rec709.
As the display is accurately calibrated there is no need for any additional information to be held within the Display ICC profile, and the graphics program CMM will only be asked to perform a simple colour space conversion, if needed!
Calibrating a Display to itself
In the above section it was suggested that for image workflows that are not tied to a set colour space standard, such as when performing photography work, with paper prints as the required final delivery, it could be advantageous to calibrate the display to its own native gamut.This is exactly the same as calibrating to a given colour space, such as Rec709, except the ‘target’ colour space is a user defined one, based on the peak RGB values of the display.
Using LightSpace, and with the display uncalibrated, use the ‘Calibration Interface to measure 100% Red, Green, Blue, and white. Record the xy values, and use them to generate a new colour space via the ‘Convert Colour Space’ menu, with the addition of the displays target gamma – most likely 2.2 for photographic work.
This new colour space is now the colour space to which you will calibrate the display, using the normal LightSpace process!
The same gamma and xy values can then be used to generate a new simple curve and matrix , or curve and colourant Display ICC profile.
Or would i need Lightspace for this?
Lightspce empty propaganda… nothing new.
What you ask is acomplished using a sRGB mode for games and a widegamut preset (calibrated & profiled) for color managed apps. That’s all. There is no issue or flaw in ICC profiles. There are limitations caused by programs using them like Photoshop and their lack of dithering/high bitdepth in “all screens and all GPUs”… but that issue is the same with monitors and LUT3D HW cal: there are monitors that need lots of grey patches to fix grey and 17 point across grey grange cannot fox them.
Also a good monitor without LUT3D, like an Eizo CS (LUT+matrix+LUT), can be described very accurately by simple single curve+matrix profile depsite empy claims from Steve.
The funny thing is that an user with a CG2420 (same as above) used CN to get Rec709 g2.4 and validated with LS.. and it passed, then Steve started arging that yellow 255 distance towards reference was unaceptable because “CIE xy” distance was too high… but we all know that CIE XYZ euclidean distance is not even close to actual human color distance (and dE00 distance was very low, below q), etc
Do you know the story of Edison and AC current frying dogs in towns across USA? More or less the same.Actual issues with widegamut dispays is uniformity which is not correctable. If contrast is good enough you can make a detailed XYZLUT profile and capture all colorspace volume’s non ideal behaviour. That profile can be used in color managed apps or feed as input to create a “crystalized” transformation form one spcte to another (a LUT3D). That’s actually what LS does and seems to be an good app doing that, just happens that this intermediate step (create a profile) is not one of their outputs, just the final step.
Lightspce empty propaganda… nothing new.
What you ask is acomplished using a sRGB mode for games and a widegamut preset (calibrated & profiled) for color managed apps. That’s all. There is no issue or flaw in ICC profiles. There are limitations caused by programs using them like Photoshop and their lack of dithering/high bitdepth in “all screens and all GPUs”… but that issue is the same with monitors and LUT3D HW cal: there are monitors that need lots of grey patches to fix grey and 17 point across grey grange cannot fox them.
Also a good monitor without LUT3D, like an Eizo CS (LUT+matrix+LUT), can be described very accurately by simple single curve+matrix profile depsite empy claims from Steve.
The funny thing is that an user with a CG2420 (same as above) used CN to get Rec709 g2.4 and validated with LS.. and it passed, then Steve started arging that yellow 255 distance towards reference was unaceptable because “CIE xy” distance was too high… but we all know that CIE XYZ euclidean distance is not even close to actual human color distance (and dE00 distance was very low, below q), etc
Do you know the story of Edison and AC current frying dogs in towns across USA? More or less the same.Actual issues with widegamut dispays is uniformity which is not correctable. If contrast is good enough you can make a detailed XYZLUT profile and capture all colorspace volume’s non ideal behaviour. That profile can be used in color managed apps or feed as input to create a “crystalized” transformation form one spcte to another (a LUT3D). That’s actually what LS does and seems to be an good app doing that, just happens that this intermediate step (create a profile) is not one of their outputs, just the final step.
I see so for when im playing games, i can use the created HW 3DLUT sRGB emulation which will correct grayscale, gamma, and have accurate sRGB emulation , and then for color managed apps like photoshop,firefox, and maybe chrome? i can use a ICC with only profiling done as long as a secondary HW 3DLUT slot does the grayscale/whitepoint, and gamma right?
Lightspce empty propaganda… nothing new.
What you ask is acomplished using a sRGB mode for games and a widegamut preset (calibrated & profiled) for color managed apps. That’s all. There is no issue or flaw in ICC profiles. There are limitations caused by programs using them like Photoshop and their lack of dithering/high bitdepth in “all screens and all GPUs”… but that issue is the same with monitors and LUT3D HW cal: there are monitors that need lots of grey patches to fix grey and 17 point across grey grange cannot fox them.
Also a good monitor without LUT3D, like an Eizo CS (LUT+matrix+LUT), can be described very accurately by simple single curve+matrix profile depsite empy claims from Steve.
The funny thing is that an user with a CG2420 (same as above) used CN to get Rec709 g2.4 and validated with LS.. and it passed, then Steve started arging that yellow 255 distance towards reference was unaceptable because “CIE xy” distance was too high… but we all know that CIE XYZ euclidean distance is not even close to actual human color distance (and dE00 distance was very low, below q), etc
Do you know the story of Edison and AC current frying dogs in towns across USA? More or less the same.Actual issues with widegamut dispays is uniformity which is not correctable. If contrast is good enough you can make a detailed XYZLUT profile and capture all colorspace volume’s non ideal behaviour. That profile can be used in color managed apps or feed as input to create a “crystalized” transformation form one spcte to another (a LUT3D). That’s actually what LS does and seems to be an good app doing that, just happens that this intermediate step (create a profile) is not one of their outputs, just the final step.
I see so for when im playing games, i can use the created HW 3DLUT sRGB emulation which will correct grayscale, gamma, and have accurate sRGB emulation ,
You’ll need a monitor with HW cal to do that and use vendor software, DisplayCAL cannot do that.
and then for color managed apps like photoshop,firefox, and maybe chrome? i can use a ICC with only profiling done as long as a secondary HW 3DLUT slot does the grayscale/whitepoint, and gamma right?
Not sure what you are saying. For other uses you’ll need to change OSD mode, HW calibrate to native and use resulting ICC as default profile OR change OSD to some user mode with native gamut, calibrate & profile grey with DIsplayCAL then use resulting ICC profile as default profile in OS.
That OSD mode can be used as a basis to make a detailed profile and make LUT3D for Resolve/MadVR… but that LUT3D will only work on that OSD and Resolve/MadVR.Lightspce empty propaganda… nothing new.
What you ask is acomplished using a sRGB mode for games and a widegamut preset (calibrated & profiled) for color managed apps. That’s all. There is no issue or flaw in ICC profiles. There are limitations caused by programs using them like Photoshop and their lack of dithering/high bitdepth in “all screens and all GPUs”… but that issue is the same with monitors and LUT3D HW cal: there are monitors that need lots of grey patches to fix grey and 17 point across grey grange cannot fox them.
Also a good monitor without LUT3D, like an Eizo CS (LUT+matrix+LUT), can be described very accurately by simple single curve+matrix profile depsite empy claims from Steve.
The funny thing is that an user with a CG2420 (same as above) used CN to get Rec709 g2.4 and validated with LS.. and it passed, then Steve started arging that yellow 255 distance towards reference was unaceptable because “CIE xy” distance was too high… but we all know that CIE XYZ euclidean distance is not even close to actual human color distance (and dE00 distance was very low, below q), etc
Do you know the story of Edison and AC current frying dogs in towns across USA? More or less the same.Actual issues with widegamut dispays is uniformity which is not correctable. If contrast is good enough you can make a detailed XYZLUT profile and capture all colorspace volume’s non ideal behaviour. That profile can be used in color managed apps or feed as input to create a “crystalized” transformation form one spcte to another (a LUT3D). That’s actually what LS does and seems to be an good app doing that, just happens that this intermediate step (create a profile) is not one of their outputs, just the final step.
I see so for when im playing games, i can use the created HW 3DLUT sRGB emulation which will correct grayscale, gamma, and have accurate sRGB emulation ,
You’ll need a monitor with HW cal to do that and use vendor software, DisplayCAL cannot do that.
and then for color managed apps like photoshop,firefox, and maybe chrome? i can use a ICC with only profiling done as long as a secondary HW 3DLUT slot does the grayscale/whitepoint, and gamma right?
Not sure what you are saying. For other uses you’ll need to change OSD mode, HW calibrate to native and use resulting ICC as default profile OR change OSD to some user mode with native gamut, calibrate & profile grey with DIsplayCAL then use resulting ICC profile as default profile in OS.
That OSD mode can be used as a basis to make a detailed profile and make LUT3D for Resolve/MadVR… but that LUT3D will only work on that OSD and Resolve/MadVR.Ohh the display im using has slots for multiple 3DLUT’s it’s using the eecolor 3DLUT format that i can quickly switch by changing OSD profile, so i thought i can do the grayscale/gamma in HW 3DLUT for native color gamut with the ICC profile only keeping a profile characterization with stuff like grayscale, and gamma being doing internally so i don’t lose unique tone values,that way i basically just need to press one button, and the display can switch between the sRGB HW 3DLUT, and native gamut 3DLUT with gamma, grayscale already pre-calibrated.
Lightspce empty propaganda… nothing new.
What you ask is acomplished using a sRGB mode for games and a widegamut preset (calibrated & profiled) for color managed apps. That’s all. There is no issue or flaw in ICC profiles. There are limitations caused by programs using them like Photoshop and their lack of dithering/high bitdepth in “all screens and all GPUs”… but that issue is the same with monitors and LUT3D HW cal: there are monitors that need lots of grey patches to fix grey and 17 point across grey grange cannot fox them.
Also a good monitor without LUT3D, like an Eizo CS (LUT+matrix+LUT), can be described very accurately by simple single curve+matrix profile depsite empy claims from Steve.
The funny thing is that an user with a CG2420 (same as above) used CN to get Rec709 g2.4 and validated with LS.. and it passed, then Steve started arging that yellow 255 distance towards reference was unaceptable because “CIE xy” distance was too high… but we all know that CIE XYZ euclidean distance is not even close to actual human color distance (and dE00 distance was very low, below q), etc
Do you know the story of Edison and AC current frying dogs in towns across USA? More or less the same.Actual issues with widegamut dispays is uniformity which is not correctable. If contrast is good enough you can make a detailed XYZLUT profile and capture all colorspace volume’s non ideal behaviour. That profile can be used in color managed apps or feed as input to create a “crystalized” transformation form one spcte to another (a LUT3D). That’s actually what LS does and seems to be an good app doing that, just happens that this intermediate step (create a profile) is not one of their outputs, just the final step.
I see so for when im playing games, i can use the created HW 3DLUT sRGB emulation which will correct grayscale, gamma, and have accurate sRGB emulation ,
You’ll need a monitor with HW cal to do that and use vendor software, DisplayCAL cannot do that.
and then for color managed apps like photoshop,firefox, and maybe chrome? i can use a ICC with only profiling done as long as a secondary HW 3DLUT slot does the grayscale/whitepoint, and gamma right?
Not sure what you are saying. For other uses you’ll need to change OSD mode, HW calibrate to native and use resulting ICC as default profile OR change OSD to some user mode with native gamut, calibrate & profile grey with DIsplayCAL then use resulting ICC profile as default profile in OS.
That OSD mode can be used as a basis to make a detailed profile and make LUT3D for Resolve/MadVR… but that LUT3D will only work on that OSD and Resolve/MadVR.Ohh the display im using has slots for multiple 3DLUT’s it’s using the eecolor 3DLUT format that i can quickly switch by changing OSD profile, so i thought i can do the grayscale/gamma in HW 3DLUT for native color gamut with the ICC profile only keeping a profile characterization with stuff like grayscale, and gamma being doing internally so i don’t lose unique tone values,that way i basically just need to press one button, and the display can switch between the sRGB HW 3DLUT, and native gamut 3DLUT with gamma, grayscale already pre-calibrated.
Oh i see, if display has such features I’m afraid that you’ll need to do 2 full calibrations, one for native, one for sRGB/Rec709.
If both slots has the same uncalibrated state (expected) and you have made a LUT3D D65 native gamut and store it in slot1, if you want to do the same for slot 2 but sRGB/rec709 using lut3d from slot1 just using DisplayCAL to compute that LUT3D from rec709 to slot1 then you’ll need to compose slot2 LUT3D with data from the slot1 lut3D and the one you made with DisplayCAL. AFAIK I’m afraid that ArgyllCMS do not make LUT3D composition from another set of n LUT3D
Maybe you can code it, a node per node composition of a new LUT3D: slot2(i,j,k)=displayCAL_LUT3D_rec709_to_slot1(i,j,k)*slot1(i,j,k), going from a RGB sRGB color to a uncalibrated slot1(=uncalibrated slot2) RGB color.
