XYZ prores problem

[TonyK]

Hi Support,

I believe I may have come across a problem with the software.

My source file is a ProRes 4444 master already in XYZ with 2.6 gamma. On my monitor it looks green/dark, which is expected. However, after making a DCP in DCP-o-matic and screening it on a professional DCI projector, it looks exactly the same as the monitor preview — indicating that the XYZ-to-XYZ handling is not working correctly.

In previous versions there was a “None” option under Source Colour Space, which allowed us to bypass colour conversion. In the current version this option seems to have been removed or renamed, and selecting “XYZ” appears to apply an additional conversion, resulting in an incorrect image.
It appears there is a bug with the way XYZ 2.6 source files are being handled? And doesn’t pass through an XYZ file without it being re-processed.

Thank you for looking into this.

[carl]

I believe there's been another report of this today - is it possible to send me a short example Prores file that demonstrates the problem?

[TonyK]

Let me see what I can do

[TonyK]

Unfortunately the director won’t let me share any imagery of the problem from his project but any ideas or help will be much appreciated

[TonyK]

Hi,
I found a build for this issue on a another thread that seems to revert the source colour settings back to include the “None” option, which allows you to bypass XYZ. However, in the latest 24 update this appears to have changed — “None” has been replaced by “XYZ,” which doesn’t let you bypass conversion if your file is already natively XYZ.

I hope this helps.

[TonyK]

Sorry to keep asking, any news on a fix for this ProRes XZY bypass issue

[carl]

Hi, I think I understand what's going on and I'm working on a fix.

In the mean time, you could downgrade to 2.18.18 which should return you to the old behaviour:

https://dcpomatic.com/download/2.18.18

[StephW999]

Hello Carl,
I just ran a test and indeed the DCI X'Y'Z' function does not do what is required.
It performs a conversion like rec709 but without the luminance normalization function (48/52.37).
There is definitely a bug. versions 2.16.xx and also version 2.18.16 worked perfectly.

[TonyK]

Thanks for the update, it’s much appreciated you getting back to me. I’ll revert back to the older version in the meantime until this bug is resolved. Thanks again.

[TonyK]

Hi do you have a link I can use to go back to 2.18.16 so I can bypass the XYZ please.