Home of Professional Cinematography since 1996
>This is puzzling me for some time. I do quite some blue screen photography, video and film, and used greenscreen only a few times when the project called for pronounced blue colored foreground elements. In using green backgrounds I personally feel I need to invest more time on the set in getting rid of green spill on the subjects than I have to do with blue. The Green reflects more color and also, if ever a slight blue cast remains sometimes in a composite, I feel it is generally less disturbing than a spill of green color.
>During the BKSTS SFX 96 seminair, Mitch Mitchell held an interesting presentation quite strongly promoting blue screen, not green. Among other arguments he made the point that the human (white) skin tone does not contain any blue, so one can obtain better masks on blue, especially on human skin, thus leaving the skin tones and richness more easily untouched by the process.
>An other presentation related difficulties on the composites for the final sequence of "Goldeneye" having cost a lot of work removing green spills, the presenter expressing regrets the shoot was not done on blue.
>During an SFX 98 presentation, there was a story about white polar bears shot against green in studio (plenty of green spill on them, especially their feet and between their legs) and the presenter once again made us understand that to his opinion he would have had an easier job if blue screen would have been used.
>All this said, what puzzles me, is that on the BIG majority of photo's I see in magazines like AC, "making off" documentaries etc of major feature productions I everytime see the use of green screens rather then blue. Blue screen seem to be rare exeptions.
>How come, everytime I hear someone from the post-production side talk about this, I hear them begging for blue while apparently green seems to be meanstream now.
>Why do so many people use green. What are the advantages?
>Are there any advantages besides the possible use of saturated blue in the foreground elements?
>Any thoughts ? What do you prefer ? Your experiences with post-houses on the subject?
>I also see advertiments of red screens now. Is this a new hype? Or just a solution for those rare cases we have blue and green in the foreground? How about skin tones on red-screen? Anybody used it?
>Kommer, puzzled Kleijn
I think that a lot of the preference of post facilities for green is to do with the fact that they don't understand film.
>They think that because green is the major component of a video signal and also the cleanest component of a video signal then it's the best way to go.
>They don't seem to take into consideration that the format that the image originates in has an influence as well.
>I prefer using blue, I have less problems with blue, I get some very strange requests from post at times.
>>Green reflects more color and also, if ever a slight blue cast remains sometimes in a composite, I feel it is generally less disturbing than a spill of green color.
Actually, it's usually the opposite; with most color screen packages, green spill resolves to brown or grey. Bluescreen spill resolves to a teal or greenish color - very rarely grey. This means that any transparent edges you have - hair, bottles, smoke - will have more believable color with green screens.
Regarding spill, there's no excuse for it. If you ever see the color cast of the screen in the final composite, just fire your post guy. Don't waste time dealing with him. Ultimatte, Primatte, and my software, The Matte Pack, can handle any concievable amount of spill, and my software is under $500, so it's not a cost issue.
>>During the BKSTS SFX 96 seminair, Mitch Mitchell held an interesting presentation quite strongly promoting blue screen, not green. Among other arguments he made the point that the human (white) skin tone does not contain any blue, so one can obtain better masks on blue, especially on human skin, thus leaving the skin tones and richness more easily untouched by the process.
That's a common myth from the optical printer days. Caucasian skin does indeed contain blue - if it didn't, it would be deep orange. It just contains slightly less blue than green or red. Notice I say "slightly."
Skintone has not much to do with pulling mattes; the colorspace of white skin is so far from the colorspace of the screen that it should never be a problem. In terms of leaving skin tones untouched, it's not a good idea.
All color screen subjects will experience a slight overall color cast, whether by reflection or lens flare. If you simply try to composite the foreground "untouched," the subject will often look pale with blue screen, and sick with green screen. ;) All high-end tools remove this cast automatically, so no composite is "untouched," really.
>>An other presentation related difficulties on the composites for the final sequence of "Goldeneye" having cost a lot of work removing green spills, the presenter expressing regrets the shoot was not done on blue.
The different screens have their place - blue screen is obviously better for this type of thing, since a lot of footage is to be composited against the sky. However, what it boils down to is "different color, same problem."
Why wasn't he expressing regrets that the shoot was not done WELL?
Thousands of green screen composites show up in theaters every year, and I would venture that even professionals only notice one in a hundred. Take a look at Titanic - that film had so much difficult green screen work, but it's invisible on screen.
>>All this said, what puzzles me, is that on the BIG majority of photo's I see in magazines like AC, "making off" documentaries etc of major feature productions I everytime see the use of green screens rather then blue. Blue screen seem to be rare exeptions.
>Remember that the blue-sensitive layer of the emulsion has the highest granularity, which is amplified when you pull the matte. Green is not nearly as grainy, so it produces a much more pure matte. That gives you more flexibility in compositing. This is why red screens are shot for motion control - it produces the best matte possible by using the best emulsion layer.
>>Why do so many people use green. What are the advantages?
>Better matte, spill resolves better, not many subjects include green. The reason bluescreen was so popular was that in the photo-chemical compositing/optical printer days, they had no spill correction, so an uncorrected blue foreground looked better than the uncorrected green.
>>I also see advertiments of red screens now. Is this a new hype? Or just a solution for those rare cases we have blue and green in the foreground? How about skin tones on red-screen? Anybody used it?
I don't believe red screens have been used for shooting people yet - it would certainly not pull a good matte. Redscreens are used for shooting motion control passes, of miniatures primarily. It came to us from TV sci/fi work. The theory is that first you shoot a "beauty pass" - a motion control pass against a black backdrop, with the subject model lit for beauty. Then you shoot the same model with no subject lights against a red screen, so what you get is the black silloutte of the model against a solid red background. You get a great matte from the redscreen, which you use to composite the beauty pass. The mattes and composites this system generates are quite incredible.
This was utilized in Starship Troopers - you can read about it in Cinefex 75. Some people also shoot magenta screen.
>Anyway, those are just my notes about the technical aspects.
>Crtainly in the feature world, the various post houses ask for blue or green based on, among other things, what their custom software "fixes" have been built for. This is the "if what you have in your hands is a hammer, all your problems look like nails" syndrome. I think that some of them feel that their clean-up programs work well enough that the other advantages that they feel they have outweigh the spill issues. I have worked a lot with blue, green, and red, and they all spill.
>Red is great for blue/green spaceships (Independence Day) and fine for that sort of thing in general, but not great for people.
From the standpoint of on-set comfort, blue is rougher on the eyes than green (if lit with narrow band sources) and on at least one job I did that was a consideration -with several months of first unit on a 12,000 sq ft green screen stage there was some effort made to crew comfort. If it's only the effects unit, no one cares.
>>Theoretically, they all are reflecting the same amount of light once lit in the standard fashion, whether blue, green, red or whatever, so why would there be a difference? I'd love to hear some feedback on this, as this was a pilot for a small cable series and I may be fighting this battle again.
>I see 2 differences.
>1. Blue has a lot less "power". It is by nature a "dark" color. It is harder to get a blue cast into something already lit in white that a green cast. In a B/W television signal f.e. the blue channel accounts for only 11% while the green has 59%. A green screen recieving the same amount of light produces more than 5 times more reflected light (theoretically). That means that blue reflextions will less easy influcence lit parts of you subject then green supposing your subject is lit with (close to) white light. The 5 times are not true in reality because we tend to sightly "overexpose" blue screens for better key. But not 2.5 stops, more like 1 stop. That makes the light coming back from a blue screen still 1.5 stop less on a light meter than that from a green screen.
>2. I feel working with blue easier while it seems more like a natural color to me. What I mean is f.e. that if I have to set up a light to kill it, a small light with quarter or half 85 gel on it often will do. There are more cases that a warm backlight on my subject is OK for the picture, while a magenta backlight is not often acceptable.......
>I quite often work for medium budget productions, wich makes that the post-production facilities do not always have all the neatest software, gear, expertise and/or time to get away with the spills easely..... All the time they put in there will be lost elsewhere in the post work. They often have to work on tight shedules and the better the source material I can deliver them the better the final result will be I think.
>Kommer wrote :
>>2. I feel working with blue easier while it seems more like a natural color to me. What I mean is f.e. that if I have to set up a light to kill it, a small light with quarter or half 85 gel on it often will do. There are more cases that a warm backlight on my subject is OK for the picture, while a magenta backlight is not often acceptable.......
>This is where good communication between the director and the visual effects supervisor is vital. If your visual effects supervisor is using an Ultimatte style system, throwing up a "spill correcting" backlight behind the subject is inappropriate; Ultimatte and similar tools have built in spill correction logic, so doing this will make the composite look wrong. However, that said, some DPs will gel the subject key and fill magenta/85, in order to seperate the subject from the background more, and then the color cast is corrected in the compositing process.
>>I quite often work for medium budget productions, wich makes that the post-production facilities do not always have all the neatest software, gear, expertise and/or time to get away with the spills easely.....
>Even Ultimatte is under $2,000 which is nothing for post houses. Ultimatte removes spill automatically, so it actually saves time. If your post production facility doesn't have an Ultimatte-level tool, they are creating more work for themselves, and if you're paying by the hour, that means they're ripping you off.
>>All the time they put in there will be lost elsewhere in the post work.
>Compositing a color screen subject into the background is a half-hour proposition at the most. Again, if it takes more than a couple billable hours, you need to visit your post house and see what they're doing. If you've shot good footage, don't let them tell you "Oh yeah, uh, it's gonna
require twenty hours of roto work."
>What are the feelings out there on the latitude of exposure on the green/blue screen? I've always been comfortable with a 1/2 stop plus or minus density on the screen. Is there a difference between the blue or green in regards to variation of density.
>On the Blue/Green issue, I have no opinion. The are both "bears" to work with (polar or otherwise) but I'd like to share a little trick I've used with fairly good success:
>With green screen, add some red gel to any backlight on your foreground subject, it will neutralize a good bit of the "spill"
>Same is true with blue screen, only add some amber to neutralize the cyan.
>Joe "I'd rather hang the actor over the cliff" Di Gennaro
Member/ IATSE Local 600/ New York
>Did I say red? I didn't mean red if I said it
>I meant magenta
>OK, OK don't blast me with ridicule!
>Joe "Mistakes make me see red" Di Gennaro
>I know that this is purely anecdotal, but I remember reading about the CG dept. working on "Broken Arrow" saying that they had to use green screen because John Travolta's eyes keyed out in the blue screen composites.
>Now, I'm by no means saying that Mr. Travolta is to blame for all this :) but I know that there are often MANY mitigating factors.
>The one thing I do know about green screen is that it takes more than one coat of paint to get the saturation right, whereas blue can usually get it in one. One commercial I did we had to postpone everything while the 3rd coat was drying because the stage owner had to redo it in the middle of the night after my prelight revealed inconsistencies in the saturation. The production was pissed because it cost them 3x as much in paint expenses- on both ends (getting it green, and then back to white). I've suggested blue screen ever since.
>>With green screen, add some red gel to any backlight on your foreground subject, it will neutralize a good bit of the "spill". Same is true with blue screen, only add some amber to neutralize the cyan.
>Again, if you're using Ultimatte, Primatte or a similar tool, this is inappropriate. Spill compensating backlight should be used only with the most basic compositing systems. It takes quite a bit of time to remove that amber or magenta edge in post.
Hope this sheds some backlight on the subject...
<< With green screen, add some red gel to any backlight on your foreground subject, it will neutralize a good bit of the "spill". Same is true with blue screen, only add some amber to neutralize the cyan.
Again, if you're using Ultimatte, Primatte or a similar tool, this is inappropriate. Spill compensating backlight should be used only with the most basic compositing systems. It takes quite a bit of time to remove that amber or magenta edge in post.
Hope this sheds some backlight on the subject...
>I won't ever do it again!
>This list is great for both learning and un-learning things.
Joe "I Stand Color-Corrected" Di Gennaro
>I just have to jump in on this thread .I think that either blue or green done well in perfect circumstances is going to work for the post people BUT. on set you just dont get the time or resorurces to do things perfectly evey time. I find green screen easier to light because as stated by a previous post it takes less light (read less money)to get the same luminence.This is a major consideration when you are lighting for high speed or massive depth in real speeds(producers get very ugly sometimes).
Next.Ever try underexposing your green screens?
This is the shot .Girl in skintight BLACK LATEX!!!!! on geen screen,shot high speed(read Lots-o-Lite) to be composited onto a torch lit scene(read Very Dark)No chance of faking little magenta backlights or whatever tricks you can think of.OK get a written note from you Vis FX superviser wittnessed by at least ten people and signed by everyone in the post dept and underexpose your green screen by two stops.Spill disappears and the key locks in with the dark BG like magic.
Don't tell anyone ... Its a secret.
Note. Blue screens underexposed go black.
>Next.Ever try underexposing your green screens?
I've consistently underexposed my green screen by 2-3 stops and this thread has been discussed at length regarding the way one determines the underexposure. In other words is the spot meter giving you an accurate reading with such a limited spectrum of color. But regardless it works with
>>Note. Blue screens underexposed go black. Not in my experience. I've found the blue to be able to be underexposed by 2 stops as well.
>However, this is on 35mm, on 16 I will tend to be more conservative.
>My question though is how much latitude does one have within a lit blue/green screen? I've usually been plus or minus 1/2 stop with no problem but again on 16mm I would be more conservative. What are the limits in 35mm in pulling an acceptable key? ie: if my exposure is f5.6 and I put the green screen at a 2.8 (2 under) do I still have the latitude for a part of the green to be 2-1/2 (f2.4) stops under and 1-1/2 (f3.4) stop under for a successful key??
>Or if I'm at f5.6 exposure and I underexpose the green screen one stop at f4 do I have the latitude to again within the screen be 1/2 over and under.
What are the limits? And at what exposure or under exposure of the screen?
Is there a difference?
Walter Graff wrote :
>Of course there is an acceptable limit both above and below the threshold where the green or blue background remains properly keyable.
Yes Walter that's what I'm getting at, what is that acceptable threshold?
Does it vary from machine to machine or how it's set up?
Sometimes we must make compromises on the eveness of the screen (ie:around 3 dimensional objects) and I would be interested to see at what limits we have and by what criteria can we base those limits. There are times when the screen looks good by the meter but the camera gets a reflective angle off the screen which makes a key more difficult which makes it important to judge the screens luminance at the camera angle.
>>... I think it important to understand that in the end you should be lighting the color for what it is and not just to illuminate the background with a consideration of the foreground element first.
>Absolutely true but if you can get a great key off 2 stops under and it helps eliminate spill and reflections on your subject then that's the choice to be made. But then how much latitude within the screen is there before reaching the unkeyable threshold on the underexposed side? I've seen discrepancies with this and was wondering what variables can cause them in post.
<< I used to do one show with a 30'x40' blue screen each week. At first I lit it for illumination only. Then I started to realize how much easier I made the engineers job when I lit it for proper color saturation.
I'm curious. When the lights were dimmed to 33%, was the change in color temperature (lighting shifting warmer/less blue light on the blue screen) a factor in the key? I'm wondering about the key's tolerance of a shift in color temperature of the light source on the screen. How warm can you go?
How blue could you go? Would you have noticed this difference on a vector scope? And what's more valuable for reading the screen on a shoot like that, a waveform monitor or a vector scope?
>>What are the limits? And at what exposure or under exposure of the screen?
>Again, I think our fine colleagues in post will say that they can get it within a stop range (+ or - a 1/2 stop), but the real question is do you want them also to be able to add shadows to the b.g. that are cast on stage? If so, I believe that you have to be more consistent with the exposure for the green/blue you want keyed out in order to make it easier for them to pull your shadow cleanly.
>Anyone else w/experience on shadows you WANT to use doing this kind of work?
>When the lights were dimmed to 33%, was the change in color temperature (lighting shifting warmer/less blue light on the blue screen) a factor in the key?
>Be careful not to draw too many generalizations from specific cases without specific parameter information. For instance, if you have a hypothetical monochromatic blue or green screen (reflects nothing back that is not the wavelength you want) then dimming lights (shifting them towards red) will not affect chroma...it's just that as you dim them you will lose proportionatly more green than longer wavelengths so your exposure curve would drop faster than a light meter would imply.
>If you have a screen which is not monochromatic (say, a white wall) and you light it with lights filtered through a "perfect" green filter, the same could be said to happen, since all the light hitting
our theoretical white wall would be filtered to pass only the "green" that we are selecting.
>Most real world situations fall between these two extremes.
>The degree of reflectance of green (and sbsorbtion of "non-green") of the screen will vastly affect its touchiness about light source purity.
>Conversely, if you are using narrow band iluminating sources for your screen, color variations on the screen and even big chunks of dirt become much less of a problem.
>With painted cycs, for instance, a big consideration is the specular kick off of the front surface of the paint. This shen can be a problem if you are using unfiltered or wide-band sources, but is generally not a problem if you are using narrow-band sources.
>By the way, you can use a polarizing filter to knock down the sheen in some situations if it is a bigger problem than the stop loss would be.
>Now I'll shut up and let Walter answer :-)
I prefer green screen for compositing outdoor scenes because it allows me to fill with bluish skylight to help sell the illusion of a sunny day.
Otherwise I would have to gell the "sunlight" extra warm and rely on post-production to cool the scene down to the proper color balance after they get the matte.
>Thom Harp :
>>Again, I think our fine colleagues in post [..]
>A visit to the PRIMATTE site, where they show a chromakey system based on their patented Polyhedron Slicing algorithm, is worth the detour.
(A unique method of calculating key values,
Clean and precise blue-spill removal functions, etc.)
Yes, Primatte is a very intriguing system, although I'm not sure I would use it for everyday blue and green screen composites; all of that complex math slows down rendering times. But if you have really awfull color screen footage to deal with, it's a good choice, because it can handle almost anything.
Every package has its strengths and weaknesses. Ultimatte has special blue screen procedures, since it was developed when blue screens were predominant, so it's really the best tool for that. My package, The Matte Pack, works best with green screens. (The math is the same for blue & green, but the spill remover works best with green.)
>On the note of condensing all of one's thoughts into one email to keep traffic down, here are my notes on color temp and exposure. Ultimatte, Primatte, and The Matte Pack can all handle significant shifts in color temperature with little effect on the matte.
There is an effect on the spill suppression, but you can use that to your advantage.
For example, if you want spill areas to resolve to grey, you need the background as close as possible to 100% b/g, 0% b/g, 0% r. If you want spill areas to resolve brown, shoot with a slightly yellowish/redish greenscreen. In terms of exposure, you want the closest to 100% green or 100% blue. That will give the lowest grain matte.
Green screens are easier for people to get their minds around, in my experience. When you say "green" to most people, they think of a VERY specific color, which is essentially pure green. And if you show them color that's too blue or too yellow, they'll say "no, that's not green." But pure blue doesn't look "pure" to most people - it looks too dark. Actually, people have a very vague conception of blue - especially producers. ;) So if they see something that's practically sky blue, they'll say "hey, that's blue! Shoot against that!"
>> When the lights were dimmed to 33%, was the change in color temperature (lighting shifting warmer/less blue light on the blue screen) a factor in the key?
>I think you also have to consider that 33% is not necessarily 33% of the output of the lamp. Different dimmers are set up with different curves. On the high end of the board, a small reduction will make less difference to output than the same reduction on the bottom end of the slider. Each type of board will be different. This is where the right lamp for the job thing comes in, or the use of wires and grip nets. Fluro sources like the wall of light etc. where colour temp shift is very minimal across the dimming range come into their own.
Director of Photography
Brisbane Qld Australia
P.S. Never used a waveform or a vectorscope on a tape job (including keys) and not about to. Get a monitor you are familiar with and use your eyes. Sometimes experience and aesthetics come into play, even on effects shoots.