My inadequate coding skills are sadly far beyond needed for this task…
Thank you for trying, hopefully someone of knowledgeable Competent will chime in to help.
So it turns out there is another option, simply copy the libdovi that was compiled from CE-20.5 into a CE-21 build.
Also found where code that disabled the 4k gui and reversed that, so a CE-21 build should no longer be problematic to do when I next play with things.
1 Like
Update:
Changes:
- update to be built from CE-21.1.1
- revert back to using 8-bit RGB as the tunnel mode
Know issues:
- Gui colors may be wrong after playing a HDR10+ file as DV if the file is at the same resolution/framerate as the gui. Workaround is just to change the gui settings to be different from actual files, i.e., 25 Hz. Also more choice in this build for the gui than in a official CE build as a native 4k gui is allowed.
1 Like
Crazy to see Dolby vision message from my tv on odroid n2+ . I tested as many videos I could find and they all seem to work fine.
Thank you for making this possible.
Is it a know limitation or bug that screen goes briefly black whenever i bring out seek bar or pause video ?
It’s known.
Out of interest, do 4K 60 Hz videos play properly? It’s not directly related to the DV stuff, but my box can’t handle that without flickering.
It’s only certain types of DV still right? When I tried it a lot of mine had the green color issue.
Still no profile 5 or FEL. Profile 8.1, p7 MEL, and p7 FEL as MEL all work.
Any other profile that doesn’t have reshaping or an enhancement layer should actually work as well, if there is actually any content that fits that description
Thanks, is 5 Web-DL?
They can be either p5 or p8.1. Profile 5 is objectively better though if you have another player, or the internal player of a tv, that can handle it. All discs are either p7 MEL or FEL.
1 Like
Thanks man, all the profiles always confuse me.
If it’s WEB-DL and the file reads DV / HDR, that is P8. If it’s just DV then it’s P5.
Ahhh ok. Thanks for the insight.
a very interesting thread, I’ve a couple of questions that the experts here may know the answer to.
I’m curious about the effective colour depth of the source with P7 FEL, i.e. is it native 12-bit 422 with an FEL source or 12-bit 420 that is then upsampled to 422 prior to transport to the display? I’m struggling to see that it can be true 422, but happy to be corrected. As I understand it 12-bit 422 uses two 12-bit bytes per pixel (hence 8-bit RGB has same data rate). From the FEL disc, we have 10-bit 420 base layer plus the EL. 420 is 1.5 bytes per pixel, so this requires 15 bits of data at 10-bit 420 and 18 at 12-bit 420? Given the typical size of the EL on a P7 FEL disc I can see how we might get to 12-bit 420 as the source colour depth - is that what we’re getting, or does the combination actually recover master-level 12-bit 422? Wouldn’t the EL need to be very large to deliver true 12-bit 422, or is this where Dolby magic happens to reconstruct “true” 12-bit 422?
My other question relates to player-led vs TV-led colour space conversion. From what I’ve read above, TV-led always involves a conversion to ICtCp, whereas player-led does not? Prior to reading this thread I was under the impression that TV-led is universally superior to player-led, but now I’m not sure, i.e might this conversion have any degrading effects?
This is incorrect. TV-led does not require ICtCp, at a minimum it can be any of ICtCp - full or limited, YCbCr - full or limited, or RGB. While the particular colorspace is flagged in the metadata sent to the TV - in testing I found this flag to have no effect. Instead it appears that you could really use any colorspace you want - provided that appropriate matrices are used for the rgb_to_lms matrix and the ycc_to_rgb matrix and offset.
That said, the ‘official’ CE implementation of DV does convert all content into ICtCp in tv-led mode. It is unclear why as this conversion is unnecessary.
Not sure if it can really be called “true” 12-bit 422 - but the process is known and implemented in tools like DoViBaker. There are a few documents that say exactly how the base and EL layers are combined - see ETSI GS CCM 001 for example.
1 Like
Most DV that I watch is P7 sourced from disc. I find TV-led playback from my Ugoos AM6B+ to have slightly softer picture quality compared to player led. This could be my imagination, but I wonder if the color space conversion is a factor here, especially given that softening in context of Dolby processing is noted in the “Perceptual Effects” paper cited in an earlier post, but I don’t have the technical knowledge to judge whether that’s relevant to playback via CE on a P7-supported device.
Learning about Dolby Vision and CoreELEC development - #65 by doppingkoala
Do you see the same difference in p7 MEL / p8.1 content?
If so, it might be interesting to compare with the results from using the internal player on the tv as well. Would guess that the tv-led results are closer to the results when using the tv’s internal player?
I suspect it may be difficult to exactly pin-down what is causing the differences though. Would assume that the different modes undergo different processing both on the device and in the tv.
I have FEL remuxes and profile 5/8 stuff. I can’t say that I see any obvious difference in picture quality or colours etc. compared to standard HDR. The FEL test videos all work though so not sure why it doesn’t seem to look any different to me.
The TV in my AV room is a Panasonic HZ980, which I should probably change. With TV-led DV it has a tendency to drop frames, it’s very obvious when it happens and it occurs with all the DV-supported playback devices I have. This doesn’t happen with player-led DV. I’m not aware of the TV having a useable internal player as it doesn’t support HD audio, 24p or forced subs in mkv files, so I’ve not bothered with it.
The content I’m watching is a mix of P7 FEL, P7 MEL that has been converted to P8 for additional device support, and more recently HDR10-only content that has been converted through Dovi_Scripts to generate P8 using cm_analyze (I’ve been very pleased with the results). I could try and chase down whether differences in picture quality are related to DV type, but given that the HZ980 appears to have issues of its own, I wonder if that would be time well spent.
To see any difference you’d have to compare the same material in P5, P8 and P7 FEL. If you don’t have such material, any comparison is completely meaningless.