RESET_9999 wrote: ↑Wed Sep 23, 2020 11:29 pm
the untouched FEL is brighter & has more punch, no doubt. But I think this is normal because I'm seeing the same kind(if not more) of difference when I compare the Spears and Munsil FEL & MEL
Great work as always.
RESET_9999 wrote: ↑Wed Sep 23, 2020 11:29 pm
also from the bluray discs I inspected, FEL DV hevc stream is the actual movie with a white/grey tint on it while MEL DV is just a static white/grey image. We still can see the video in the MEL conversion so that could be the reason why it doesn't play on the x700.
There are multiple reasons why the FEL structure is intact in my MEL files.
This is not the important thing.
What matters, in my opinion, are the NALu RPU. Imagine the latter as an orchestra conductor. With the right changes I can explain to the Decoder that it does not have to start a second decoding session (as happens in FEL files) but to use "a static image" as substrate (as happens with MEL files) regardless of the presence of FEL data.
This is the goal.
Guys, please note that I don't work for Dolby and I don't have access to any documentation.
All this is the result of a logical analysis shared between me, myself and I.
There is no guarantee that the process will be correct but if it were completely wrong the film would not start on devices like the FireStick 4K (able to read only MEL) and a BD player would not play it (with the help of the right container) without color defects (
and reproducing the same behavior that exists between an original FEL and MEL file)!
Also, I think FireStick owners are okay with anything compared to a full BLACK screen
!
RESET_9999 wrote: ↑Wed Sep 23, 2020 11:29 pm
...I can add some random MEL track to any HDR10 hevc stream and it will still play in DV (single or dual layer). with some flickering but it triggers DV...
True!
The tricky part is to generate glitch-free files on devices that cannot read the original (FEL) files.
Result that cannot be achieved with random data
!
shawnc22 wrote: ↑Thu Sep 24, 2020 5:35 am
I also believe that this is probably the expected behavior. When FEL is processed correctly, it's produces a 12bit picture whereas MEL is only capable of 10bit. Even though we don't have 12bit displays, a 12bit stream downsampled to 10bit will still look different than a native 10bit(MEL) stream. A picture and picture comparison of the original FEL vs converted MEL may not then a be a good way of determining whether the method is working.
One thing you can potentially try is since the S&M disc contains a demo that's provided in both FEL and MEl, convert the FEL to MEL and then compare it to the provided untouched MEL.
I agree with every single word especially the blue part.
I don't want to take advantage of your kindness @RESET_9999!
Also keep in mind that my RPUs carry more information than those normally found in MEL streams.
Maybe the result could be in favor of the converted MEL file over the original MEL.
Or it could be a complete disaster...
EDIT
Pure personal judgment.
I like the MEL frame more than the FEL frame
.
.
. 