Topics

BRAW tests -> and why we should think about distribution codecs too

axel.mertes
 

After having responded to Jeff, one side note:

Whatever we are talking here about is only 1/3 the story.

We can try finding the codecs that hold up great against uncompressed recording in-camera, to allow smoother workflows. I had a long discussion with two ARRI developers (those doing firmware on Alexas...) about how hard to it is in post to handle ARRI RAW vs. e.g. RED R3D or Cineform material. It shocked me that one reason for them not implementing wavelet based compression was that fact they theoretically calculated it might need like 20W more power... I gave them something to chew when I consider how much more power - iteratively and for hundrets of times longer - we need in post to handle that stuff... Thats a big waste on the wrong end, especially when we find camera compression useful - after all. And wavelets even more.

Anyhow, then the second 1/3 we need to find the right codec in post, that holds up several generations in worsed case scenarios and still does not sacrifice what we got on camera. Many tests tell that Cineform is a good choice for several reasons.

After all, then we are facing the the final 1/3, the real nightmare: Distribution codecs.
Not much choices there. I'd say its MPEG4/H264, H265 and VP9 for streaming, a bit of MPEG2 for DVD and some BluRays (if done wrong). And DCP JPEG2000. Thats it.

Depending on your pipeline those mostly fixed data rate codecs will create more or less artefacts.
And vice versa, dynamic data rate compressed files will get bigger or smaller, depending on your pipeline.
I was able to demonstrate that RED ROCKET processed data came out with 10% bigger files on H264 BluRay dynamic data rate (=constant quality) encoding than its RED software only counterpart, ie. the image quality of the RED ROCKET processed files was less good than the RED software only processed counterpart. As most distribution channels use fixed data rate, that means the RED ROCKET processed stream will have significantly more artefacts than the RED software only processed one. Things not known to many users of this stuff out there. And thats just an example. A codec that might seem to be bad in post might turn out with better results on the distribution codecs reencoding. A lot of combinations to think about and test.

Why I am telling this?
Because if we go getting picky on camera codecs and mezzanine post codecs we should also care about what that means to the results. Its ignorant and naiv to think the xx,xxx $ class 1 display with uncompressed content in the high end studio facility is what the consumers will see. And to me ONLY the result that the consumer WILL see is relevant.

And honestly its a mystery tale that future codecs might get really better. I really don't think so. Distribution codecs are designed to hold up to specific PSNR (peak signal to noise ratio) and MSR (mean square error) values. The deal is either to get smaller bandwidth or a little better PSNR/MSR. But the values didn't really change of the last decades. You trade image detail vs. bandwidth. Lately we had the quadrupeling increases in resolution, which yields about a tripleing in data rates required with codec generations halving the datarates with aan acceptable detail loss (as its now only 1/4th the size - individually looked at).

We will unlikely see the day when Youtube or Netflix will release the uncompressed quality channel for the sake of image quality fanatics. Given that video content of these two alone is responsible for like a half* the entire internet data transfers world wide, you can imagine what it means to the world if one can REDUCE that data streams. It means excessively expensive infrastructure, unbelieveable high energy consumption. I think the guys who work for Google/Youtube on VP9 can easily be outweighed in pure gold, each single day, for what the world is saving by using their codecs.

I just wanted to share these thoughts, not necessarily start a dicussion about it, but something for you to consider.

*not fully up to date, but surely GROWING: https://www.forbes.com/sites/quickerbettertech/2014/11/24/netflix-and-youtube-now-consume-50-of-the-internet-as-the-argument-for-net-neutrality-weakens/#701e025b36e6
and also worth reading:
https://www.forbes.com/sites/bernardmarr/2018/05/21/how-much-data-do-we-create-every-day-the-mind-blowing-stats-everyone-should-read/#2d0fd78c60ba

Mit freundlichen Grüßen,
Best regards,

Axel Mertes

Workflow, IT, Research and Development
Geschäftsführer/CTO/Founder
Tel: +49 69 978837-20
eMail: Axel.Mertes@...

Magna Mana Production
Bildbearbeitung GmbH
Jakob-Latscha-Straße 3
60314 Frankfurt am Main
Germany
Tel: +49 69 978837-0
Fax: +49 69 978837-34
eMail: Info@...
Web: http://www.MagnaMana.com/

Diese E-Mail enthält vertrauliche und/oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese
E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese E-Mail. Das unerlaubte Kopieren
sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in
error) please notify the sender immediately and destroy this e-mail. Any unauthorised copying, disclosure or distribution of the material
in this e-mail is strictly forbidden.

Keith Putnam
 

On Tue, Oct 2, 2018 at 01:47 PM, axel.mertes wrote:
I'd say its MPEG4/H264, H265 and VP9 for streaming, a bit of MPEG2 for DVD and some BluRays (if done wrong). And DCP JPEG2000. Thats it.
It's worse than you think. Here in the year 2018, CBS masters their prime time dramas in 50Mb/s XDCam (with MXF wrapper) for broadcast. So we start with 3.2K ProRes 4444 in-camera and wind up with 1920x1080 50Mb/s XDCam. Ugh.

Keith Putnam
Local 600 DIT
New York City

axel.mertes
 



Am 02.10.2018 um 20:04 schrieb Keith Putnam:
On Tue, Oct 2, 2018 at 01:47 PM, axel.mertes wrote:
I'd say its MPEG4/H264, H265 and VP9 for streaming, a bit of MPEG2 for DVD and some BluRays (if done wrong). And DCP JPEG2000. Thats it.
It's worse than you think. Here in the year 2018, CBS masters their prime time dramas in 50Mb/s XDCam (with MXF wrapper) for broadcast. So we start with 3.2K ProRes 4444 in-camera and wind up with 1920x1080 50Mb/s XDCam. Ugh.

Keith Putnam

Well, be fair. 3.2K was never good enough for UHD, right? So it makes a very good HD, yes?
Fine.
50 Mbit masters is, well, acceptable, if you consider that they actually stream with 5-25 MBit maximum in H264 afterwards.
I think it depends on the content. If you were telling me that Hollywood is mastering 4K at 100 MBit/s I'd be more concerned...

Mit freundlichen Grüßen,
Best regards,

Axel Mertes

Workflow, IT, Research and Development
Geschäftsführer/CTO/Founder
Tel: +49 69 978837-20
eMail: Axel.Mertes@...

Magna Mana Production
Bildbearbeitung GmbH
Jakob-Latscha-Straße 3
60314 Frankfurt am Main
Germany
Tel: +49 69 978837-0
Fax: +49 69 978837-34
eMail: Info@...
Web: http://www.MagnaMana.com/

Diese E-Mail enthält vertrauliche und/oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese
E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese E-Mail. Das unerlaubte Kopieren
sowie die unbefugte Weitergabe dieser E-Mail sind nicht gestattet.

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in
error) please notify the sender immediately and destroy this e-mail. Any unauthorised copying, disclosure or distribution of the material
in this e-mail is strictly forbidden.