Adjustable compression for Redcode?

[Peter Karlsson]

I´ve tried to search for this topic, but couldn´t find anything.

My question is regarding the compression artifact you get with "redcode". I will mainly use the Red One in a greenscreen studio. Now I´ve only looked at material shot by Jon Firestone, Chris Bagley over at Redrelay.net. I´ve heard by others that the Red is the BEST for keying, but Im somewhat pussled over my own results. I can see heavily artifacts when pulling a chromakey on the material by Jon Firestone and Chris Bagley. It soo nice to see material with as little noise as this, but its heartbreaking to see the material ruined by the compression.

I know that the codec isnt finished, and I don´t know which build this was shot with. But my question is basically if there are any plans on letting the user be able to shot with a lesser compression but with a higher data rate as result? Is the red drive able to keep up with higher data rates that the CF cards? Or is the raw port the only way to go if you want to be able to pull the cleanest key out there?

/regards Peter K

 

[Mark L. Pederson]

I´ve tried to search for this topic, but couldn´t find anything.

My question is regarding the compression artifact you get with "redcode". I will mainly use the Red One in a greenscreen studio. Now I´ve only looked at material shot by Jon Firestone, Chris Bagley over at Redrelay.net. I´ve heard by others that the Red is the BEST for keying, but Im somewhat pussled over my own results. I can see heavily artifacts when pulling a chromakey on the material by Jon Firestone and Chris Bagley. It soo nice to see material with as little noise as this, but its heartbreaking to see the material ruined by the compression.

I know that the codec isnt finished, and I don´t know which build this was shot with. But my question is basically if there are any plans on letting the user be able to shot with a lesser compression but with a higher data rate as result? Is the red drive able to keep up with higher data rates that the CF cards? Or is the raw port the only way to go if you want to be able to pull the cleanest key out there?

/regards Peter K

How are your processing the footage prior to pulling the key?

There is MUCH less noise in the blue channel if you light your green screen with daylight balanced lights. IMO -that is a RED VFX 101 rule.

We have pulled "artifact-free" keys with ease, so I am curious of the steps you are taking with the .r3d files.

As to adjustable compression - I for one, would really love the ability to see what I get with 27MBs in 2k at 23.98. Right now it is 6-7 MBs - and I assume it is locked there to deal with variable frame rate.

 

[Peter Karlsson]

Hi Mark,

I´ve only been able to use the material at Redrelay.net. I have been looking at the "Green Screen - Daylight" clip, using the settings:

Sharpen 0
Detail low
NR Low
Kelvin 6500k, Saturation 1.2, Exposure 0.65, Tint 0

I found that NR set to none is resulting to huge artifacts (I would like to set NR to none since it keeps most detail. But the compression is really visible), and NR high just makes it a smear, while low seemed to be the best artifact/blurriness ratio.

Im outputting this as an OpenEXR, with linear filtering.

Im keying with keylight in AE in 16bit depth. But the end results are just showing compression artifacts.

/Regards Peter K

 

[Gavin Greenwalt]

Peter I've been getting the best results on REDCode Greenscreens from custom difference keyer gizmos and IBK.

Try experimenting with some custom stuff and use keylight for your supression to see if it doesn't help.

It is overcompressed imo as well. Here's to hoping for a higher bitrate option.

 

[Peter Karlsson]

How does Nuke IBK work compared to a traditional keyer? (I´ve haven´t got a Nuke license, but since Im planning to do a lot of keying as soon as the Red arrives, it might be worth it).

It seems the problem lives in the blue channel mostly, so I cross my fingers that it will improve in never builds.

I think I will have a problem getting together (=money) stuff that lets me take the raw-port route. But if there will be a option to do a minimized compression before sending data through the raw port ending up with a user set (50-500mB/s) stream that may be an option.