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 :)

More testing - The bad greenscreen move...

Gunleik Groven

Gunleik Groven

REDuser Sponsor
Joined
Dec 28, 2006
Messages
13,465
Reaction score
5
Points
38
Location
Norway
Website
quine.no
Well this is a greenscreen test, but a very practical one :) (To me, that is)

I run around a bit showcasing the cam and workflow for a few companies, and by now EVERYONE and his aunt and his aunts cousin have heard about "the really bad IR issues with RED", and the unusefullness of the cam in tungsten lit greenscreen work...

(many of my outdoor shots are shot with ND 0.9 + 0.6 and sometimes with a pola on top of that...)

well. true to my methodology: I show them exactly those issues :) Try to emulate the worst case...
(Bare with me, I aknowledge the issues, but this is however how the cam will be used in many situations...)

Off we go to a screen, put some poorly lit subject (and sometimes screen) in front of the cam - all tungsten of course, shoot with wb and everything way off and do a test.

we load the developed files into Shake and do the "one click in keylight" test. You be the judge.

BTW. These are not tests of great greennscreen work....
Just an emulation on how this will be done by a lot of people...

Hope you find it usefull

http://byacserv.com/vulture/fastkeytest.mov

Cheers
 
is this really on click with tungsten? hard to believe!!!

thanks
 
There was some really bad blue screen shot at LART. One or two shots were lit with a combination of fluorescent and tungsten... The noise in the blue channel was ungodly bad, but it still keys great, pretty amazing really. We were mostly doing one-click keylight tests during LART as we tested lighting and different subjects (blondes and brunettes in different outfits). Some of the keys were a little funny on the noise bluescreen, but nothing that a good lumikey or a couple tweaks or multiple key nodes couldn't fix.

I don't recall if any of it made it to the Vuze channel or not, but I'll see if I can post a clip sometime in the next few days. A few stills were posted. At this point, it's kinda bad to do so because it was shot on pre-X cameras.. Actually, worse than that. The bluescreen tests that Gavin (im.thatoneguy) and I did were shot with one of the pre-production cameras that GibCor acquired from RED. I don't know what (if any) hardware differences there were between it and the first 100 cameras before the Gen-X upgrades. I'll double check my notes, but I think the camera was using Build 9 firmware (all cameras at LART were build 9 or 10). So I have to wonder if it would even be appropriate at this point for me to post footage. But it's more of a confirmation that good keys could be pulled from crappy bluescreen even back then on old model cameras.
 
Dude that is amazing.. Very cool tests.

and yes.. Norwegian Women! I want to move there now.

post more!
 
Tend to agree with you Jeff (without testing, that is) One could probably pull a good key on bluescreen as well.

To the extent that there is a "trick", it is in the development of the files.

A well lit screen and subject (at 5600k) wouldn't hurt either, but I think the RED is a quite useable little cam for keying as it is, even though one can maximise that potential with filters & daylight...

With this test, the "one click" approach was desireable, because it's what I have to show customers quickly. For a demo they don't want me to tell them that what they have is useless or:
a) preproduced material or too controlled conditions (that comes later...)
b) me sitting for hours.

It's more like
"and then: Keying"
(running in and doing a quick shoot, dumping the file, process it, into shake, click keylight: OK? Fine!")
"And now: Highlights..." :)

There would still be some work to do, before it was good, but for a quick start... :)

there is an IR issue with black. Of course.
But as black is... well: black :) It's not all that hard to correct for.
 
gunleik said:
we load the developed files into Shake and do the "one click in keylight" test. You be the judge.
Click to expand...
When you say "one click in keylight" what do you mean? You didn't adjust any of the settings or controls? If so, can you list out any adjustments?

Thanks,

-shooter
 
Every time I get worried about noise or colour temp, Gunleik comes to the rescue!

Great job as always.

You're my new hero!
 
Hi Gunleik,

I agree with you that RED in 4K mode and GS is an excellent tool for footage aquisition. For some jobs I had to do GS shots that we lit with daylight (remember REDs sensor is 5000K and on a paid production you want it right...) The result was great. But even if your footage has the quality of that what RED delivers it takes some time to create a good composit.

For a convincing composit I had to use at least two key-nodes (holdout mattes). Matching plate and forground does also take some time. Adding Defocus and rendering the whole thing in 1080p adds another time to wait. Then some flaws has to be ironed out and the whole thing has to be rerendered.

I just want to make clear that achieving a good composit takes some time regardless how great the GS footage is. Nevertheless RED is the best low budget alternative for GS work and par to 35mm IMO when shot in 4K. Keying and composition is time consuming and RED does help a lot if compared to F900 and the likes.

Hans
 
Yeah, but as you say, that's work you have to do no matter what the acquisition. The issue is that film is relatively easy to key, whereas miniDV is a nightmare just to get a reasonable key, let alone composite. And when you talk about "shooting digital", people immediately balk because of the well-known horrors of 4:1:1. Being able to strike the green almost instantly with RED -- as you can -- is a major selling point for any FX-related work.
 
Hi

I'll just state again that this is not a compositing test as such, more like "what I have to do quickly a lot of times to show that this CAN actually be a great compositing tool"..

When I say "develop the files and do a one-click in keylight" I actually mean exactly that.

But the key word here is more "develop" and not "one-click"...

I usually export DPXs for these tests and work with them in RedCine, so that the customer can see that color/noise issues are handleable.

Hans. I agree with all your points. There isn't really ever a "quick fix". This is just my little stab at "mythbusting", and I know it is a quick one :) But as such, it usually is quite convincing...
 
Shooter... as to "settings in keylight"...

For these kind of things, it is better just to drag the mous over the screen until it "sits". well "enough".

When working further on a key, all Hans points are good and valid, and I just basically "feel/see/test" my way through the keylight settings which are neccesarry. from shoot to shoot I think they vary a lot, so I don't use one default "RED keylight setup".

Really looking foreward to test this with the nuke primatte, though...
 
Hey Gunleik,

Two questions:
Are you keying from DPX's or Quicktimes?
In your opinion which seems to be working best for RED footage, Primatte or Keylight? I have Primatte for AE, but am not a huge fan of it, is it worth upgrading AE to get Keylight?
 
Keying from full rez DPX (4k/2k/3k... whatever is shot)

I haven't used Primatte in Shake on RED footage, basically because I have only used it for garbage/holdout matte earlier...

But I've tested the new one in Nuke, and that's a different story alltogether (not on Red footage, though)

Keylight is what I like, but I think you should rather talk with Jeff Kilgroe or digitalfx about those issues.

Cheers!
 
gunleik said:
Keying from full rez DPX (4k/2k/3k... whatever is shot)

I haven't used Primatte in Shake on RED footage, basically because I have only used it for garbage/holdout matte earlier...

But I've tested the new one in Nuke, and that's a different story alltogether (not on Red footage, though)

Keylight is what I like, but I think you should rather talk with Jeff Kilgroe or digitalfx about those issues.

Cheers!
Click to expand...

Wicked.

Thanks for the feedback!
 
Hi.

just a quick follow up.

I had a client who insisted on "daylight ballancing" tungsten lights with filters on the lights during a shoot just now. I was not in a posission to do a test without the blue filtering, so I'm not into the full info on this, but...

I suggested that they rather filtered up the camera, but that was apparently not an option. (A bad advice???)

The result however...?
The worst "one click" key I've seen from RED so far. Still very useable though, but quite a bit harder than the tests I published. The compositing house was pleased with the results, so I guess they're gonna be fine.

I am not too experienced in the math behind this... I share this just as an experience.
 
You must log in or register to reply here.
Back
Top