New RED Color Science - Test Pilot Feedback Thread

[Patrick Tresch]

I have done some extensive testing and I would say that this is more that a "monitoring build" as you are putting it.

David

If this is a fact... I'm very happy to hear that!:hurray:

But can someone show me a comparison test between build 17 and build 20 using the same Redalert! build?

Thanks.

Patrick

 

[Jannard]

I guess the question is Jim and gang: If I am going to do the final debayer in post for the DI (Pablo) will there be a difference between what I capture RAW in Build 17 & what I would get in shooting with Build 20.

Peter

Peter... the answer is yes. The difference in camera is less than in the post tools, but there is a difference, especially in tungsten. The other reason to use Build 20 instead of Build 17 (camera side) is that the monitoring path is much better.

Jim

 

[Rob van Gelder]

too bad that a lot of the uploaded pictures and examples have disappeared since the Red user Heart attack / spambot killing.

Is it possible to link them back or do we have to re-upload those pictures?

Rob

 

[David Battistella]

If this is a fact... I'm very happy to hear that!:hurray:

But can someone show me a comparison test between build 17 and build 20 using the same Redalert! build?

Thanks.

Patrick

Patrick,

I have not been able to post anything yet, but what I am finding is that 17 looks better with the new tools, but the 20/20 combo looks better in all lighting situations. It's beta and a bit buggy, but the color reproduction in REC 709 is MUCH better. The sensor still performs much better under daylight lighting but the test I was most impressed with was when I did some classic Mixed lighting between straight tungsten and daylight.

BEFORE: you'd have to use at least a 1/4 blue on all tungsten light because it would just be too warm and noisy.

NOW: With decent light level's (f2.8) tungsten foreground and daylight back, the camera renders it MUCH more like what we are used to without adding the 1/4 blue to the tungsten.

In my opinion this is the best news of all. We can pump in the tungsten because the noise floor has been reduced. I think it's a combo of debayer and squeezing more out of the sensor, but that's just Graeme's special sauce.

I am not afraid of pure tungsten as I once was. That's the best news.

I'll try to set up a link to download my tests.

David

 

[Raymond Adams]

The release notes for B20beta state that image noise was reduced in the blue channel and noise response was improved under tungsten lighting.

Thanks Mike,
This is good info...if it really works that way.
Has anyone tested B20 with blue chromakeys?

 

[Paul Leeming]

Raymond, I haven't tested with blue chromakey but in general you should use green chromakeys for bayer sensors since you are getting twice the colour precision thanks to having twice the amount of green sensor sub-pixels to blue or red sub-pixels. Load some R3D footage into Redcine and examine the sharpness in the individual channels. Notice how the green channel gives you more detail than the red or blue.

So avoid bluescreen unless you have no choice, at least when it comes to Red....

HTH

Paul

 

[Raymond Adams]

Raymond, I haven't tested with blue chromakey but in general you should use green chromakeys for bayer sensors since you are getting twice the colour precision thanks to having twice the amount of green sensor sub-pixels to blue or red sub-pixels. Load some R3D footage into Redcine and examine the sharpness in the individual channels. Notice how the green channel gives you more detail than the red or blue.

So avoid bluescreen unless you have no choice, at least when it comes to Red....

HTH

Paul

Thanks Paul,

I do understand this, and usually would use green...but there are many times for multiple reasons that Blue needs to be used...thus my questioning.

also that in general the issues of noise on the blue channel...along with the rolling shutter issues.

all impact RED's functionality for FX shots, etc...

Thanks for your comments,

 

[Marco Graziaplena]

May be I'm a little late for this thread... anyway:
New Color Science will not affect histograms in RAW wiev, but it will in REC709 or REDSPACE view.
I'm I right?

I couldn't see any underexposure test in daylight... I already updated to v.20 so I can't do tests anymore with old version of RA. Maybe someone got one?
Thanks