MP4 White & Super White Issue

[Andreas_Gumm]

Fact is, that I have pointed to this behaviour a year ago in the German forum base on GH4 footage and this behaviour could be shown with test patterns in clear way. But nothing was done. Other tools like Vegas does not show that behaviour with mp4. So not every tool has this bug.

May you remind that we also have cleared up in the same thread that no specification
exists that defines a biunique flag for these non-spec range flavours!

Btw you can theoretically provide as many values you want in a range of 16-235.
Floating point does the trick, so at the end it's equal if a file uses full range or legal
range of 10bit values. The non-spec conform ranges has been used to extended color
range, just to break limited number of values of 8bit files in an YCbCr color space,
just to get more overhead for color correction.

I think some profiles in the primary color correction for these special files could easily help.

Andreas

[wschmid]

According to rec709 every value between 0..255 has to be transferred. And you are wrong that there cannot be endless figures between 16..235 - even not for 8bit where you are very limited, better for 10bit. But you are limited up to now since Edius supports a 10bit mode only, even if the internal calculationt may talke place with a higher number of bits. There is no 32bit floating point modus applied within Edius as it is done in Resolve, After Effects or even Vegas.

But we are not talking about that but about that the lumarange of mp4 is decoded in a wrong way as one can see in an easy way in the waveform monitor. And that is a bug that has not been addressed since long time. And filters like the primary color correction should not be nessesary to get the mapping of the luminance in the correct way (and even this filter performs wrong in that aspect for v log l or even slog3 footage, but that is another story).

[Andreas_Gumm]

We were debating in our German forum about odd color ranges such as 16-255 and other crude stuff.
I agree that full swing and studio swing should be correctly recognized in Edius.
The other crap is not following an official specification.

Andreas

[wschmid]

Yes 16-255 is odd and there is no way to flag such a thing.
I have no idea why would you even give such an option in cameras?
What is the purpose of it? 16-235 or 0-255 are fine and make sense, but 16-255?

The purpose is that user will have more and more the intention to use full swing or video range, simply because more and more cameras has started to be able to use that increased signal range. That has started some years ago with units like the GH4, and Panasonic is not a small player. And in reallity it has been always there - every consumer camera records in superwhite signals. And it will not go away at all, because with the latest UHD-TVs you are able to use 0..1024 with 10bit and 422 more and more in rec2020 or color spaces that tend to cover that.

It would not be necessary that it is recognized automatically maybe even if that would be great. But it is necessary that the user is able to adjust that in a simple way. For sure not with filters like the primary color corrector but the switch should work where one can adjust how the clip has to be decoded. Simply to stretch the luminance range what may result in clipping is dangerous and undesired behaviour of the software really.