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 :)

New RED Color Science - Images & Questions

J

JonathanF

Well-known member
Joined
Nov 29, 2008
Messages
286
Reaction score
0
Points
0
Location
Working in LA, NY, London or Vancouver
Website
www.veritefeatures.com
Thanks for posting the new builds of RedAlert with the new Color Science updates. It's made for an interesting day.

I took a handful of shots from a film we're working on right now and the results are pretty dramatic. It's hard to see without looking at the rec709 grading monitor or a projector calibrated to rec709, but the difference is pretty big.

These images are from RedAlert v3.60 and RedAlert 20.0.1b, both using identical settings to output images in rec709. The last two images are stripes showing a comparison between the two and one with just the blue channel. The blue channel is certainly less splotchy in the 20.0.1 files, but the color shift is pretty extreme. Any comments?

The new build seems to have much more blue in it or be generally cooler than the older 3.6. Is this what you guys are seeing on your side as well? I haven't labled the images as the warmer ones are obviously the 20.0.1 files.

What's the general thinking on working with build 17 or 18 footage and grading using build 20 SDK apps (when we can build them)? Were you expecting this much of a warm shift in the color?

Thanks again,

:J
 
I install this up grade and now I need mix files from the different version and the difference is dramatic
 
JonathanF said:
These images are from RedAlert v3.60 and RedAlert 20.0.1b, both using identical settings to output images in rec709. The last two images are stripes showing a comparison between the two and one with just the blue channel. The blue channel is certainly less splotchy in the 20.0.1 files, but the color shift is pretty extreme. Any comments?
Click to expand...

If you are using identical settings in RedAlert v3.60 and 20.0.1 there will be a difference in the output. You have to redo White Balance in v20.0.1 to achieve a similar grading as with v3.6 but with less noise.
 
Alexander Christ said:
If you are using identical settings in RedAlert v3.60 and 20.0.1 there will be a difference in the output. You have to redo White Balance in 20.0.1 to achieve a similar grading but with less noise.
Click to expand...

Alexander,

Have you switched off colour matrix in RA? Although in 3.60 and 20.0 both pictures look identical in RA the blue channel of the build 20.0 rendered picture is different. Even in "right" exposed footage you can see a more smooth blue channel, although the difference is pretty subtle. Which leads to the conclusion that Red improved their debayer algorithm.

Hans
 
Hi Alexander,

I guess I didn't make myself clear:

If you switch off the Colour Matrix both the 3.6 and the 20.0 RA show the same result. Now, if you render a picture with Colour Matrix switched off you can see a subtle improvement in he blue channel even the footage is well exposed to the right. Hence my conclusion that the new "Colour Science" does not only affect the matrix - which it does severely - but also improved the basic debayaering prior to WB and such.

Kind Regards to Swizerland,

Hans
 
Hans,

If you are using identical settings in RedAlert v3.60 and 20.0.1 there will be a difference in the output. You have to redo White Balance in v20.0.1 to achieve a similar grading as with v3.6 but with less noise.
Click to expand...
This was all shot on build 17.

What I was curious about was the density difference in the blue channel. There is clearly a big difference in the exposure there.

I was hoping to get someone from Red's thoughts on this as well.

:J
 
Wow looks great! I am a little weary of how blue became orange? I like the warmer look and the noise is alittle more textured in the 8bit.

I have to see it at 10bit for my own eyes but everyone in TV/Web/PC land will be looking at 8bit.

A histogram would be great to compare the images:)


Thanks for the Samples!
 
Dane Brehm said:
Wow looks great! I am a little weary of how blue became orange? I like the warmer look and the noise is alittle more textured in the 8bit.
Click to expand...

Jason Wingrove said:
Just a side not.. the pics from the film look gorgeous. more details?
Click to expand...

Dane & Jason,

Thanks VERY much. I really appreciate the warm feedback from you guys.

A few months ago when we started the film we weren't sure what we were going to shoot on. Then I shot some stuff 2nd unit on the camera and was hugely impressed with it using Ultra Primes on another project.

Then we spent some time with that footage in post, made up a short list of things to do different in the future and decided to do another test with MP's and see how it went.

Well, long story short, the footage with MP's once the camera was properly set up, backfocus checked, etc. was f&^ing georgous. I was sold.

The film is the first in a series and we are keeping it pretty quiet for now, but these are some of the less compelling shots in my opinion so I hope it will be fantastic. It's essentially a culture series.

What I will say is that unlike a lot of docs we're shooting primes unless we absolutely must have a zoom, and we're going to try to stick with MP's or UP's for the series and maybe just one Arri LW zoom. It's a bit harder work, but having T1.3 up your sleeve is so nice from a creative standpoint. The workflow shooting with primes is essentially a result of my haivng done mostly feature work, so I'm actually more comfortable thinking about by blocking in those terms so... bit I digress.

I'll let you guys know more once we make some announcements. There should be a few shortly.

Again, I'd love some thoughts from the Red guys on the density change in the blue channel. I was considering doing some image analysis on these and see if my coding guy could construct a corrective LUT based on a bunch of images, but I'm not 100% sure that's the best course of action here so I await feedback from the 'really smart guys.'

Ta,

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