I just wanted to say thanks for taking the time to resolve this issue.
I’ve noticed there’s an update available for macOS, so I’m updating DCP-O-Matic now. I’ll let you know as soon as possible if the issue is still there.
From my understanding, the process you outlined makes sense:
• Linearising sRGB (approx. gamma 2.2 with the sRGB curve)
• Applying the DCI scaling
• Then applying the 1/2.6 gamma
However, based on what I’m seeing with this test chart, it does look like the source gamma may be being interpreted slightly differently in practice.
For clarity on the attached images:
• Image 1 is the DCP conversion with no changes applied.
• Image 2 is the output from my Mac Studio (2.2 gamma, 4K setup), representing how the image should appear when played back from the Mac via SSD to a 4K cinema projector.
When I let DCP-o-matic handle the conversion using the sRGB preset, the resulting DCP appears slightly lifted in the blacks. The reference from my Mac (2.2 environment) matches what I would expect from the original image.
So to answer your question, it feels like either:
• The sRGB gamma (particularly the toe or linear section) may not be mapping exactly as expected in this case, or
• The PNG gamma or metadata might be influencing how it is being interpreted
It could also simply be that this type of black-level test chart is exposing very small differences that would not normally be noticeable in real-world content.
Thanks again for taking a look at this.
Best regards,
TonyK
You do not have the required permissions to view the files attached to this post.
Just to add to my previous message, I carried out a test today by creating the DCP using DaVinci Resolve with the same source and equivalent settings.
The export completed fine, and the result looks as expected, matching the output from my Mac Studio (2.2 gamma environment) and what I would consider the correct appearance.
Based on this, I do believe the issue may be related to how DCP-o-matic is handling the gamma during the encoding process.
TonyK wrote: Thu Apr 09, 2026 8:02 pm
• Image 2 is the output from my Mac Studio (2.2 gamma, 4K setup), representing how the image should appear when played back from the Mac via SSD to a 4K cinema projector.
Why do you assume this is the case?
You are not clear enough about what you see and how you evaluate right or wrong. How do you display the DCPs you create?
To clarify, I am evaluating the DCP in a professional cinema environment. Playback is through a 4K NEC digital cinema projector using a Dolby IMS3000 server, projected onto a matte white screen calibrated to approximately 14 foot-lamberts on white (DCI standard for 2D presentation).
This setup reflects a true exhibition environment, so the image I am seeing is representative of how the DCP would appear during a standard cinema screening. My evaluation is based on comparing this projected image against the original reference image viewed on a calibrated display, assessing consistency in luminance, contrast, and gamma response.
For reference, the calibration images being used consist of:
• One image with white rectangles on a bright background
• One image with black rectangles on a dark background
The purpose of these images is to verify correct display calibration. Under proper conditions:
• The edges of the darkest and brightest rectangles should not be visible
• The second darkest and second brightest rectangles should be just barely perceptible
If all dark rectangles are visible, the display is too bright; if the second darkest cannot be seen, it is too dark.
Similarly, if all bright rectangles are visible, the display is too dark; if the second brightest cannot be seen, it is too bright.
All rectangles should appear neutral, with no colour cast or tint. If a display cannot achieve this, it indicates that it may not be capable of accurate brightness and colour reproduction, and a professional calibrated monitor should be used for proper assessment.
Based on this methodology, if the source image is correctly calibrated and the conversion to DCP is handled accurately, the projected result in a properly calibrated cinema environment should visually match the expected output.
However, in this case, I am observing a discrepancy between the expected result and the DCP playback, which is why I raised the question.
I would also like to clarify that I am not pointing the finger at anything in particular. I am simply trying to understand what might be happening. As part of this, I am currently remaking the source files in case there was an issue at that stage.
The main purpose of asking these questions and creating these test charts is to establish a reliable sanity check. This will allow me to confidently assess contrast during screenings with editors, as well as colour when clients attend QC sessions, and also provide a consistent reference for file-based checks.
Hello TonyK
What I can say is that DoM does sRGB to DCI X'Y'Z' conversion very well
I tested your pattern with other DCP creation software and I do not see any difference in the values of each black square
You do not have the required permissions to view the files attached to this post.
I am following the dialogue here since the beginning.
I haven't heard anything about what I was thinking from post #1, so I am going to ask myself:
Wouldn't be wise for one to use test patterns used for the specific medium/color space?
Is it by luck, that the specific test patterns were created for sRGB, with a different gamma, a different white point, a smaller number of bits per channel? (That was a rhetorical question.)
Would one expect DCP-o-matic player (for instance) to show the same ramp and contrast on a computer screen that appears on the "silver" one?
I see two photos. They seem to be of a cinema screen, maybe I am wrong, but if it is the same projector that is fed the DCP and the Mac input, I expect visible differences between the (same) test patterns on one and the other. Even more, if the Dolby IMS3000 is doing the conversion (as the previous IMSs did) from the HDMI input and the installer never measured colors and white against such a source. And to be honest, if the installer's measurements were to be used equally on cinematic Flat and alternate source HD/UHD, who would blame them?
The photos inherit the issues of the digital camera, and the grey tones seem to be tinted. Which is no surprise.
Calibration is great, but -contrary to the DCI purposes- not all screens, projectors, final X and Y coordinates are the same, nor all screens keep their image brightness (14FL) throughout the edges. In fact, it is much safer to say that all screens loose light as one goes further from the center of the screen. And if this 4K projector is laser, you have another beast to fight, metamerism. (You fight it either way, but the laser one is bigger and angrier.) I won't start about screens.
My opinion: If you want to make a test pattern for cinema, make it on a proper monitor for that and then test it on a cinema. Or, at least, find a post production system, calibrated accordingly and work on that to make the test pattern.
If you take a convergence PNG test pattern from an ICP and make a DCP out of it, it won't be the same on screen. Even more in laser systems. You will be able to see ghosting.
As far as test patterns go, shortcuts (conversions) have their pitfalls.