[FFmpeg-user] Requesting colormatrix advice

[markfilipak.imdb at gmail.com]

On 11/6/23 09:16, Devin Heitmueller wrote:
> On Sun, Nov 5, 2023 at 10:42 PM <markfilipak.imdb at gmail.com> wrote:
>>
>> Hello All,
>>
>> I have a 1990s TV show -- called 'SOURCE' here -- made by Paramount. The show was renowned for its
>> production values. It's on Blu-ray from Paramount. The Blu-ray was made in 2012 (may be a clue to a
>> colorspace variant used).
> 
> If it is standard definition and from the 1990's then the original content is almost certainly in BT.601. 

Thanks, Devin,
To add to the fun/mystery, though it was a TV show, Paramount shot it on film, so maybe it was color 
graded on film(?), whatever 'BT' that would be (probably not SMPTE). It looks like too big a color 
shift to simply be BT601 rendered via BT709. It's the saturation that clues me to that. The reds are 
saturated in play and in the ffmpeg HEVC encoding but they can't actually be saturated on disc 
because I can draw detail out of them. I think there's a 'limited' versus 'full' range thing going on.

I'm currently transcoding it with this:
-vf colorcorrect=saturation=0.85:rh=-0.06:bh=-0.1
It looks good, but that's to my eyes, on my laptop, with my laptop's video settings. I'm out of my 
depth with no oars. (Attenuating the blue bothers me, but hey, to my eyes the discs have a blue cast 
in addition to saturated reds.)

Paramount has done this to other TV shows. I'm hoping someone with experience of Paramount's 
film-to-TV-to-disc work will offer advice.

> Now if you're referring to the
> BluRay remaster, then who knows what they did.  If done properly I
> would expect them to convert the colorspace to BT.709 ...

How convert? According to ffprobe, the discs have no colormatrix. So, I think, ffmpeg has no choice 
other than to accept it as BT709 -- 'conversion' not possible. I don't know how to parse M2TS, so I 
can't check on whether ffprobe is being truthful. Any parsing ideas would be welcome.