Dolby Vision now possible through MP4 Mux.

[jlpsvk]

I have been blindly converting hundreds of my Dolby Vision UHD BDs to Profile 8 for compatibility, but now I am starting to wonder if there is a point. When I compare the Profile 7 and 8 results in Plex on either the AppleTV or Shield they look the same. Now I can't see them side by side, but am I just wasting my time or is there a benefit maybe I am just seeing? At this point the only advantage I can easily recognize is DV playback with Infuse on the AppleTV with the Profile 8 conversion (albeit it doesn't always work).

I am exactly in that position and didn’t convert my DV UHD BDs bcoz I wanted to convince myself objectively that I wasn’t loosing on PQ even if if can’t see a difference (even just MEL ones). This topic is quiet complex, I think one has to do pretty much the same amount of work and diligence that RESET9999 did to start reaching an understanding.

One approach is to use DV test pattern(s) and compare the performance of a reference DV player (Oppo 203) with ATV4K DV performance (by taking accurate photos of the screen), and that of Infuse or other DV software capable media streamer. I attempted to do this & you can find good reading starting w this https://forum.makemkv.com/forum/viewto ... 63#p135163 and the next few posts and replies from RESET9999.

One conclusion (whether significant or not), and dismissing any errors in the experiment, is that using that P8 DV test pattern, the Oppo shows elements of that test pattern that ATV (internal player) nor Infuse doesn’t. How does this reflect on PQ and what a human perceives during movie playback is a subject beyond me.

Maybe the convenience of having all digital media being available under the control of a streamer software (Infuse, …) trumps whatever PQ difference there may be. I’ll redo what I did w tvOS17 which brings P8 internal playback support.

Profile 7 MEL is the same as Profile 8. I am baking FEL into encodes out of P7 FEL titles, and just creating profile 8 from P7 MEL and injecting it to HDR encode.

[aboulfad]

^^^ Thanks, it is known that P7 MEL conversion to P8 is “lossless”. What I think was being pondered on, whether PQ for viewing a P8 movie is the same in a video processing chain using a reference video player vs the aforementioned software streamers (Infuse, Plex) on an ATV 4K.

Using the above test pattern, ATV DV processing of that P8 test pattern yielded different results than using the same pattern on an Oppo UDO-203 w same cables/TV. Reading S&M explanation of that test pattern kinda implies something funky happening with ATV DV processing for that P8 pattern.

[ArArdin]

Seeing difference between FEL and MEL/Profile 8/recodes and really enjoying it are two different things indeed.
To be able to assess, first off you must be able to see FEL in its full glory, which I think very few of us can do. It takes a native 12 bit screen in the first place and its ability to display full blown 12 bit luminance. To my knowledge there are none for consumers (yet?), unless things have changed. So be aware that you're probably watching down-to-10 bit FEL DV on your screen. Nevertheless, even then FEL can look better. Also, FEL advantage takes players which Dolby Vision Composer actually process FEL. Such as DV compliant UHDBD players and some mediaplayers.
I won't go into personal taste. I do think a lot of recoding is done to save up on harddrive space. For those who do experience FEL as the walhalla, won't do any of it, my guess.

[RESET_9999]

^^^ Thanks, it is known that P7 MEL conversion to P8 is “lossless”. What I think was being pondered on, whether PQ for viewing a P8 movie is the same in a video processing chain using a reference video player vs the aforementioned software streamers (Infuse, Plex) on an ATV 4K.

Using the above test pattern, ATV DV processing of that P8 test pattern yielded different results than using the same pattern on an Oppo UDO-203 w same cables/TV. Reading S&M explanation of that test pattern kinda implies something funky happening with ATV DV processing for that P8 pattern.

that P8 pattern file looks correct only on cmv2.9 devices.
It has issues on all the cmv4.0 devices.

ATV:
strange artifacts in the red and blue ramps
issues in the Y’CbCr clipping

Shield:
banding in the blue ramps (I think its the reason for the red push)
issues in the colorspace conversion (goes away in LLDV)
issues in the Y’CbCr clipping

FireTV:
st2084 tracking different that all the other device.
looks darker

C2 internal player:
issues in the colorspace conversion
issues in the Y’CbCr clipping

[aboulfad]

…
that P8 pattern file looks correct only on cmv2.9 devices.
It has issues on all the cmv4.0 devices.
…

thank you for the summary from previous experiment: viewtopic.php?p=134248#p134248 , I think ATV can only process P8 w cmv2.9, does it mean it’s a cmv2.9 device ? so I am a bit confused by the above two statements.

[RESET_9999]

thank you for the summary from previous experiment: viewtopic.php?p=134248#p134248 , I think ATV can only process P8 w cmv2.9, does it mean it’s a cmv2.9 device ? so I am a bit confused by the above two statements.

it doesn't matter it still uses the most recent CMV4.0 engine to playback DV regardless of the input version and the pattern is cmv2.9 anyway.

[aboulfad]

it doesn't matter it still uses the most recent CMV4.0 engine to playback DV regardless of the input version and the pattern is cmv2.9 anyway.

Hmm, so if we had a cmv4.0 DV test pattern, would we expect better conformance to that test pattern on ATV?

[RESET_9999]

Hmm, so if we had a cmv4.0 DV test pattern, would we expect better conformance to that test pattern on ATV?

No, I tried a 4.0 version of this pattern and it's the same issues. (At least on the C2 and Shield)

But I think I read about Stacey Spears saying that Dolby ordered a bunch of their new benchmark discs so maybe that will be improved in the future.

[aboulfad]

…
But I think I read about Stacey Spears saying that Dolby ordered a bunch of their new benchmark discs so maybe that will be improved in the future.

That’d be funny, Dolby Labs using S&M test patterns. I wouldn’t hold my breath between Dolby & Apple to have something changed anytime soon.

If you take a guess, what do you think would be impacted when viewing a movie based on the difference(s) of ATV’s processing of that one pattern?

[RESET_9999]

It it possible to skip ProRes generation while converting HDR10+ to DV (using 3-1) or is it just better to use the old method if I want to do it quickly?

no
cm analyze requires a prores file.

[RESET_9999]

If you take a guess, what do you think would be impacted when viewing a movie based on the difference(s) of ATV’s processing of that one pattern?

No idea compared to your oppo but I prefer the PQ of the ATV better than the Shield.

[aboulfad]

No idea compared to your oppo but I prefer the PQ of the ATV better than the Shield.

Sorry I wasn’t clear. Knowing that the ATV has some clipping (Y low, Cr Low), does that translate in some perceivable difference for Picture Quality(PQ)?

I recall that Y(Luma) tends to be more important than Cr/Cb , I wonder if we’d expect crushed blacks or whatever ...

Doing back to back comparison with Oppo is difficult bcoz it takes like 5s on my TV to switch from DV to whatever then to DV. so I found it hard to do that comparison.

I copied inline the S&M explanation for that area of the pattern:

Clipping
Y high / low and Cr high / low

These boxes test whether the Y’CbCr channels are getting clipped at the reference high / low levels before being converted to RGB. This is always wrong; clipping should always be done as late as possible in the display pipeline, and certainly only in RGB.

If Y’CbCr is being kept intact until RGB conversion, there should be clearly visible lighter or darker small squares in the center of each of the four larger squares. If one or more of the channels in Y’CbCr is being clipped, one or more of the smaller squares will not be visible.

[RESET_9999]

I still dont know how this reflects in real content but the problem seems to be more than just clipped channels because IIRC, the whole square turns black instead of green/red as on cmv2.9 devices or in the SM docs.
If that was just a clipping problem you just wouldn't be able to see the lighter/small red/green square.

[aboulfad]

^^^ thanks for your replies, indeed the Ylow & CrLow squares are black. I just thought maybe I can send the ATV screenshot to S&M and ask. Buying their new UHD disc would be the official way to get some of their support : )

[RESET_9999]

@Stacey usually responds to users questions here: https://forum.blu-ray.com/showthread.ph ... 27&page=41