Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Debayering R3D's in the new FCP>Color Workflow

Joe Walker

Joe Walker

Well-known member
Joined
Dec 28, 2006
Messages
218
Reaction score
1
Points
18
Age
46
Okay, so I've been using the new FCP>Color workflow and I'm enjoying it quite a bit because for the most part it makes sense and it sort of speeds things along in the editorial process. However, I gotta ask a simple question. A while back I remember reading somewhere that Graeme Nattress said something along the lines of RED footage needing to be properly de-bayered to maintain peak quality, and I've noticed that in the Log and Transfer and Color settings, I don't see any settings for debayering. So does FCP's L&T or Color at any point in the new workflow ever do a debayer of the R3D's or is this workflow just referencing the quicktimes? I'm a bit confused. Anything you learned scholars can educate me with would be much appreciated. Thanks.
 
Any source from a single chip imager must be debayered to generate a color image. A bayer filter consists of red, green and blue filters, normally in a checkerboard pattern, covering the sensor which just measures light intensity. Each pixel location in the RAW data has just one color channel. One row is RED-GREEN alternating values, the next is GREEN-BLUE. The debayer algorithm makes an educated guess about the missing color values at a particular pixel based on the values of the surrounding pixels.

There are several debayer algorithms in the literature, and since they all just guess, a camera, or image sensor, manufacturer makes a compromise based on a speed vs quality tradeoff for most situations that you may encounter. Normally, the "better" the debayer, the more CPU required for the processing.

Yes, to just see the colors, a debayer algorithm is being performed.
 
I also wish we had some more info on all this. We heard over and over how RED does not add "sharpening" in camera and that we have to sharpen in RED Cine...yet there is no "sharpness" setting in Color.

Anyone have any ideas on this?

kevin
 
Kevin Wild said:
I also wish we had some more info on all this. We heard over and over how RED does not add "sharpening" in camera and that we have to sharpen in RED Cine...yet there is no "sharpness" setting in Color.

Anyone have any ideas on this?

kevin
Click to expand...

Have you looked in the Color Fx Room? There is a sharpen node in there!
 
In Color the proxy control gives you half-standard or half-high accordingly. For sharpening, you're best off doing that with a tool in the ColorFX room. You can do lots of cools stuff there.

Graeme
 
Thanks. Any more information sure would go a long way. I was under the impression that any changes in the non-RED room under "primaries" was NOT using actual Metadata/RAW.

Is this right?
 
Hi Graeme,

I am a little confused...

Can we use the full debayer power in Color ?

Also, What is the best way to manage 3K footage to 2K ?

thanks!


Antoine
 
Graeme was too humble to mention his own plug-ins for Color, which are excellent.

If you are feeding native RED footage to Color and set it to 32 bit float, everything you do in any room is full quality, but you don't have the option for a full debayer.

Native 3K is still a major pita in FCS/Color. If you need it, do a first light in RA and convert it to ProRes or DPX with RR.
 
Antoine Fabi said:
Hi Graeme,

I am a little confused...

Can we use the full debayer power in Color ?

Also, What is the best way to manage 3K footage to 2K ?

thanks!


Antoine
Click to expand...

I second that question...logically since the the qt wrapped files are 4k but are debayered to 2k in fcp and color I believe that they arrive at 2k via half debayering...is that right? If they used a full debayer they would require scaling to be sampled down to 2048x1024 which would take a lot of time...way more time than the L&T module uses...also the qt wrapped files would be much smaller in size - please someone correct me if i'm wrong..

thx
 
With RedRushes on my 8 core 3Ghz, it is about a 25:1 transcode time ratio making 1920x1080 ProRes HQ with a full debayer. With Half High, it is about a 12:1 ratio. On the sharpest detail I can see a difference, but most of the time they look the same at 1920x1080. I would go full debayer if it didn't take so long. 600GB of R3D took two and a half days to render at half high.

STeve
 
2k is done as full debayer. 3k is done as half. So you may want to do 3k externally as 3k and downsample to 2k for a Color project.

Graeme
 
Hm. I've done 3k externally, and I'm not done testing yet by far, but is it the lack of options of scaling algorithms that make the difference?

For lowlight shots, I am leaning quite heavily towards doing an external pass.
For 3k and reframed shots + composites, that is the viable option. (go to DPX)

But I'm curious about the straight-foreward shots, too.
 
4k to 2k via "half high" is probably around 95% (rough guess) as nice as 4k scaled to 2k, so I don't see half high as any problem at all for a 4k to 2k output, especially as it's very much faster.
 
Graeme Nattress said:
2k is done as full debayer. 3k is done as half. So you may want to do 3k externally as 3k and downsample to 2k for a Color project.

Graeme
Click to expand...

Graeme,

RE: 2k presented using a full debayer...Are you referring to the QT wrapped L&T workflow through to Color or the Prores transcode?
 
I'm talking about rendering in Color from a wrapped QT.

Graeme
 
Graeme Nattress said:
4k to 2k via "half high" is probably around 95% (rough guess) as nice as 4k scaled to 2k, so I don't see half high as any problem at all for a 4k to 2k output, especially as it's very much faster.
Click to expand...

But what about 4k to 1080 via "half high" then a small re-scale, compared to 4k direct to 1080?
 
I shouldn't think so.

Graeme
 
You must log in or register to reply here.
Back
Top