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 :)

Ask David Mullen ANYTHING

Most often we use greenscreen behind the foreground "Tara" on location / in the set, then shoot the background "Tara" with the real background.

We had one tricky shot though where Tara comes into her bedroom from the hallway, and then sits down in the foreground, followed by Buck, who ends standing over her in the background. We could not greenscreen the set because of the door and narrow hallway that she had to pass through as Tara, would have created a green tunnel or something unlightable, plus all that green would be too close to her.

So we lined up the shot on bedroom set and shot the background element with Buck coming through the real door, from the hallway, then we measured everything, camera distance, height, tilt degree, distance to the door, etc. and went outside the bedroom set to a clear area on stage so I could basically use a very large (20x20) greenscreen farther back so Tara could start as if she were deep in the hallway. We put tape marks for the door area and her path to walk and I used lighting and flagging to simulate the effect of passing from the hallway lighting, through a "door", and then into the bedroom lighting.

For a few over-the-shoulder shots where you never see the face of the foreground person, we have gotten away with a double, though that's tough because Toni Collette acts with her whole body and a stand-in often doesn't have the same expressiveness.

When they don't overlap, like in profile angles, we just do moving splitscreens in post.

We had one tricky shot that director Craig Gillespie designed where we dollied back from Toni as Shoshana behind a desk, dollying out the room and into a hallway... and just as she disappears from frame as a doorjam comes into the foreground, we see Toni as Tara cross into the shot and sit down in the hallway, the camera still pulling back. So we basically did a poor man's motion control shot and let the doorjam wiping through the frame create a place for a moving split screen.

Luckily most of our greenscreen shots have been on stage where we can pull a wall and get the greenscreen further away than the set wall would normally allow. But we had one tricky sequence outside after Tara accidentally drives her car off the road. We thought we'd have to put a greenscreen inside a Volkswagon Beetle between the driver's seat and the passenger's seat to get Tara to sit next to Shoshana, but I think we decided that the foreground person could be shot outside of the car against green, or maybe we ended up realizing that they didn't cross each other and we could just do it with a splitscreen.

We also had two more shots outside the car, which required that I hang a big 20x20 solid over the whole area to keep the real daylight off (luckily the car was under some trees) and add lighting that would stay consistent over the hours it would take to shoot one half of the conversion for two angles, and then the other half after Toni changed hair, make-up, and wardrobe. Once we rehearsed the shot, with Toni playing both characters, we basically shot the foreground person against green and the background person against the real background.
 
Last edited:
This reminds me of a scene i watched the other day in Battlestar Galactica (Series 3 Exodus Part 2). I am rewatching it with a friend and there is a scene on a small set on their version of Airforce 1 essentially, there was a round table meeting of various cylon models. If anyone has seen the re-imagining there are multiple copies of the human looking cylons. Back to the point...there is a pan around the round table from the centre, one apparently handheld pan in complete 360 with at least three or more copies of each actor. They were all sat down so they didnt move or interact. If it was motion capture then it was one hell of a small set up. Unless they did multiple pans and tracked each actor on another layer...then composited them together.

EDIT: I just watched it again...it is not handheld. The camera just pans round on a centre pivot, although it does tilt up and down as it pans steadily around. Logically this would mean that you could composite three or four passes together only having to mildly adjust XY positioning on each layer to match the shots.
 
It's a lot easier when you are just duplicating the same actor in the same make-up, hair, and wardrobe, a LOT harder when there is a lengthy changeover to a new look for the actor -- if that change takes 45 minutes or so, invariably they want you to shoot something else in the meanwhile, but you don't want to breakdown anything between the two halves of the splitscreen, so you really have to plan it well, like have a different set with another camera for a different scene (with a different actor) that you can switch to shooting while you are waiting. Or you just wait and hope that hair & make-up estimate was correct. Sometimes a 20-minute estimate turns into 30 minutes, then 45 minutes, etc. so a decision you made based on the first estimate would have been different had a longer estimate been given. However, I don't blame anyone for underestimating how long something takes to do, often no one really wants to hear the truth.
 
Hi David,

I do VFX work at a post-production facility in Chicago (I think you were here just a few weeks ago!). We have a commercial project coming up which will be shot in a bar with a lot of televisions on the walls. The spot will be shot on 35, and the production company wants to playback stock footage in all of the televisions, but they are concerned about screen roll or flicker artifacts in the footage. Recently, I worked on several spots involving multiple flat-panel televisions and we would occasional see flicker on some screens but not others (those spots were also shot on 35). Fortunately, the screen footage was always going to be replaced in post and the flicker didn't get in the way. But the idea of shooting the screens practically and NOT having any visual artifacts has me very concerned, especially in this age of LCDs and plasma's possibly running at different refresh rates.

So, is there any fool-proof steps to shooting flat-panel televisions on film (or digitally) that will guarantee a clean, stable image?

Thanks,
Paul Rosckes
Optimus
 
LCD's are safer than Plasmas, which have a faint 60 Hz pulse going through them (as opposed to the roll bar from a CRT screen). The decay rate of LCD's tend to make them flicker-free when shot on film. Of course, you could take a 24P video camera with you to location to see if any monitors are acting up. I'm talking about TV monitors of course, not computer monitors (LCD computer screens tend to be safe as well, and I don't think anyone makes Plasma computer screens?)

It may or may not help if you used a 144 degree shutter (1/60th of a second exposure time) in terms of reducing a 60 Hz pulse. Again, a video camera may help you figure out a good shutter speed, assuming your film camera can be set to it.

If you end up with LCD's and only one or two have something funky going on in them, you probably could replace the image in post with a stable one.
 
David thank you so much for all of your patience even when answering some of my pretty basic ?'s. ... so here's another one.

I just want to be sure I understand some of the differences of timing footage for theatrical release versus broadcast versus DVD.

FWIU, both broadcast and DVD should have their color levels set to the studio swing scale which is 16 to 235 (using 8-bit values). So black is 16, 16, 16 and white is 235, 235, 235. And this holds true for BOTH broadcast and DVD.

Also, FWIU, theatrical release would require material to be graded using the full swing scale, so black is 0 and white is 255.

Is this correct and is this the "only" change that's done when re-timing for broadcast, DVD or theatrical release?

Thanks much. BTW, I've posted this ? on some other forums but could not get a clear answer, I believe b/c the number of people who have actually taken a film from origination to wide release is still rather small.

Thanks very much again.
 
David,
I do most of my own editing and minor effects work, and am by no means an expert. I therefore am very careful about green spill during green screen photography, you seem to have the same policy. How do post houses and effects artists feel about images with green spill? Do they routinely use post tools to get rid of the green spill and it's no big deal or should photographers strive for clean images?
 
I try to get rid of as much green spill as I can -- it helps that I try to work with the greenscreens farther back for starters. But I've generally had the efx supervisor there on the day and we look at the monitor and talk about what he's OK with. Sometimes I point out some green reflection in a desktop or something and he tells me he can handle that in post.

You do the best you can with the time & equipment you have, you try to not be sloppy about it. Most efx supervisors want an evenly-lit greenscreen at a decent exposure, and as little spill as possible, so that's what you try to give them.
 
Peter Moretti said:
I just want to be sure I understand some of the differences of timing footage for theatrical release versus broadcast versus DVD.

FWIU, both broadcast and DVD should have their color levels set to the studio swing scale which is 16 to 235 (using 8-bit values). So black is 16, 16, 16 and white is 235, 235, 235. And this holds true for BOTH broadcast and DVD.

Also, FWIU, theatrical release would require material to be graded using the full swing scale, so black is 0 and white is 255.

Is this correct and is this the "only" change that's done when re-timing for broadcast, DVD or theatrical release?
Click to expand...

Sorry, I don't know the bit values being used specifically, though I can tell you that most color timing for features is done in 10-bit Log, not 8-bit.

Hopefully Mike Most or some other post person will help answer this, but in general, the color gamut for film is wider than for broadcast video (Rec 709, which describes both a color gamut and gamma). Generally you would color-correct for this wider range for film-out and then do a pass working from the finished master that brings everything into Rec 709 space for a broadcast master; it's not completely automatic, you generally want to look at it to make sure it all looks good, making some adjustments when necessary.

If you deliver an 8-bit 4:2:2 video master in Rec 709 color space/gamma for film-out, basically a D.I. facility applies a LUT that corrects it for a laser recorder, including a conversion to Log, though ideally it would go through a D.I. suite once more for tweaks. The main thing is that you won't have the same dynamic range and color range as something corrected in 10-bit Cineon Log RGB space (assuming that the camera original had that much range.)

The general idea is to always work with a wider range and then deliver submasters in narrower ranges.

I don't recall what a D.I. facility calls the color space used for color-correcting in 10-bit Log 2K RGB, for example, for a film-out -- maybe Cineon or just Wide-Gamut RGB. P3 is the name of the color space used by digital projectors; some people have color-corrected in that space I believe. I have worked at somme D.I. facilities that insist on timing digital camera material in Rec 709 color space because they don't believe the camera recorded any colors outside that gamut anyway. Not sure I believe them -- the sensors in some of those cameras allow wider-then-Rec709 colors to be recorded.

In terms of the values used for picture information (black to white) in 8-bit and 10-bit, film-outs and broadcast, I'm sure someone else can chime in. Or find it online.

Here is an article that touches on some of these issues:
http://www.theasc.com/magazine/april05/conundrum2/index.html
 
Thanks David! I'll try to have the production company shoot some test with a 24p camera at the tech scout. And if we do have to replace some screen footage, it will undoubtedly be on the only moving camera shots, where the television is behind an actress with big frizzy hair, and there will be no usable tracking marks! :)

-Paul

David Mullen ASC said:
LCD's are safer than Plasmas, which have a faint 60 Hz pulse going through them (as opposed to the roll bar from a CRT screen). The decay rate of LCD's tend to make them flicker-free when shot on film. Of course, you could take a 24P video camera with you to location to see if any monitors are acting up. I'm talking about TV monitors of course, not computer monitors (LCD computer screens tend to be safe as well, and I don't think anyone makes Plasma computer screens?)

It may or may not help if you used a 144 degree shutter (1/60th of a second exposure time) in terms of reducing a 60 Hz pulse. Again, a video camera may help you figure out a good shutter speed, assuming your film camera can be set to it.

If you end up with LCD's and only one or two have something funky going on in them, you probably could replace the image in post with a stable one.
Click to expand...
 
David Mullen ASC said:
... I have worked at somme D.I. facilities that insist on timing digital camera material in Rec 709 color space because they don't believe the camera recorded any colors outside that gamut anyway. Not sure I believe them -- the sensors in some of those cameras allow wider-then-Rec709 colors to be recorded.

...
Click to expand...

David thanks very much for the info. and the link.

I do hope someone like Mike chimes in. BTW, I know for a fact that the F35 records a wide gamut of colors, wider than Rec. 709 and wider than P3. I was at a Sony presentation earlier this week and that was one of the points they emphasized.

Another was that the SLog curve was/is optimized get the most out of the F35's sensor. So while it's a log curve it's different enough from a Cineon curve that an (easy to do) SLog to Cineon conversion needs to be applied to properly bring SLog into a post production workflow.
 
Peter Moretti said:
FWIU, both broadcast and DVD should have their color levels set to the studio swing scale which is 16 to 235 (using 8-bit values). So black is 16, 16, 16 and white is 235, 235, 235. And this holds true for BOTH broadcast and DVD.
Click to expand...

I'm not Mike Most, but if I may I'll chip in from a post perspective.

Although 'video' formats such as tape and DVD work internally in a restricted range (16-235 for 8 bit and 64-940 for 10 bit) the situation is unfortunately not as simple as "grade in restricted range for video".

You can in fact grade for video in either 'full range' or 'legal range' (Baselight's terminology for the two, although other software uses different names) so long as you transform on the way in from the format your footage is in into your working format, and transform on the way out into your deliverable format.

However the situation is complicated by the fact that some software allows you to choose a working format, and tell it what format your footage is in, and what format you would like it output in. Other software makes assumptions about footage formats, and tries to hide the internal workings from you. This is intended to make things easier for the non-technical, and if the assumptions the software made were right every time it might work. Sadly this is not the case, and incorrect assumptions (QuickTime I'm looking at you here) are the root of all sorts of problems.

So taking DVD as an example, if you are encoding for DVD in Apple Compressor, the footage needs to be converted from whatever format it is in to 16-235 Y'CbCr. But if you give Compressor 16-235 RGB footage, there is no way of telling it that is what it is, so it will convert it wrongly, assuming it is 0-255!

And as for gamma, don't get me started!

Confused? You will be!

I hope I have helped a bit. My preference is to work with software which gives the user full control of interpretation and conversion of footage. With QuickTime and Final Cut Studio this is not easy. My personal experience is that Baselight (with Kompressor) is pretty much unique in being able to convert RGB footage to ProRes QuickTime for FCP and do so correctly.
 
David Mullen ASC said:
Hopefully Mike Most or some other post person will help answer this, but in general, the color gamut for film is wider than for broadcast video (Rec 709, which describes both a color gamut and gamma). Generally you would color-correct for this wider range for film-out and then do a pass working from the finished master that brings everything into Rec 709 space for a broadcast master; it's not completely automatic, you generally want to look at it to make sure it all looks good, making some adjustments when necessary.

If you deliver an 8-bit 4:2:2 video master in Rec 709 color space/gamma for film-out, basically a D.I. facility applies a LUT that corrects it for a laser recorder, including a conversion to Log, though ideally it would go through a D.I. suite once more for tweaks. The main thing is that you won't have the same dynamic range and color range as something corrected in 10-bit Cineon Log RGB space (assuming that the camera original had that much range.)

The general idea is to always work with a wider range and then deliver submasters in narrower ranges.
Click to expand...

I don't recall what a D.I. facility calls the color space used for color-correcting in 10-bit Log 2K RGB, for example, for a film-out -- maybe Cineon or just Wide-Gamut RGB. P3 is the name of the color space used by digital projectors; some people have color-corrected in that space I believe.
Click to expand...

The color space normally used for film targeted DI work is basically film print density space, and is usually simulated on a digital cinema projector by using a LUT, either in the color grading system, in a box prior to the projector, or in the projector itself through use of projector configuration file. In some facilities, digital cinema is now being used as the primary target, so the system is set up to monitor in P3 (digital cinema) color space, with a P3 to Rec709 conversion LUT used for creating video deliverables, and an inverse LUT used to create proper Cineon files for the film recorder. This approach is becoming quite popular for digitally originated material as it retains much better color purity from the original values than is obtained by forcing the material into a film centric pipeline and a non-native color space.

I have worked at somme D.I. facilities that insist on timing digital camera material in Rec 709 color space because they don't believe the camera recorded any colors outside that gamut anyway. Not sure I believe them -- the sensors in some of those cameras allow wider-then-Rec709 colors to be recorded.
Click to expand...

Not speaking for any facilities that do this (the ones I've worked for don't, and the one I'm working for now doesn't), but that sounds a bit more like an excuse for not wanting or perhaps being able to set it up correctly than any kind of reasonable technical explanation.

In terms of the values used for picture information (black to white) in 8-bit and 10-bit, film-outs and broadcast, I'm sure someone else can chime in. Or find it online.
Click to expand...

The difference between full range and SMPTE scaled values is based primarily on whether you are dealing with video devices (i.e., videotape machines and HD-SDI feeds) or computer graphics devices (i.e., computers). DI work is done using "full range" and video deliverables are created by using SMPTE scaling to create something the VTR (usually HDCam SR, or sometimes D5) expects. There are both hardware and software scalers available, but SMPTE scaling is done as a last step on the way to the VTR. All original files are kept in their original full range values if they're created as DPX sequences, as they usually are.
 
Peter Moretti said:
Another was that the SLog curve was/is optimized get the most out of the F35's sensor. So while it's a log curve it's different enough from a Cineon curve that an (easy to do) SLog to Cineon conversion needs to be applied to properly bring SLog into a post production workflow.
Click to expand...

That's somewhat true for a film targeted DI finishing path, but is is not the case for television work. Both SLog and Panalog are designed to be gentle enough log curves that they can be used directly - without a LUT - for color grading in a "normal" grading system. The vast majority of shows being shot on Genesis, for instance, are recorded using Panalog, and they are color corrected directly from that using a DaVinci 2K or, in some recent cases, systems such as Lustre, Baselight, or Digital Vision Film Master.
 
Hi David,
I'm producing a feature to shoot later this summer and really have this in the back of my head constantly.
After watching films shot on RED such as "Knowing", I see it is virtually indiscernible to film.
However I just watched an independent film with a great DP, great lighting, great lenses, shot on 4K Red. And although the footage looks just beautiful, it doesn't have the same film feel to it. I can tell it is RED footage.

My question:
Is this something that has to do entirely with the post process, like a film out then rescanned for digital distribution then it has the film feel, or added post film grain?
Or is this just the work of a knowledgeable DP who knows how to set up the cameras shutter or has some tricks to actually make it look and feel like its indiscernible to film?

Sorry if this is a daft question, just something that I have noticed.
 
Josiah Spencer said:
Hi David,
I'm producing a feature to shoot later this summer and really have this in the back of my head constantly.
After watching films shot on RED such as "Knowing", I see it is virtually indiscernible to film.
However I just watched an independent film with a great DP, great lighting, great lenses, shot on 4K Red. And although the footage looks just beautiful, it doesn't have the same film feel to it. I can tell it is RED footage.

My question:
Is this something that has to do entirely with the post process, like a film out then rescanned for digital distribution then it has the film feel, or added post film grain?
Or is this just the work of a knowledgeable DP who knows how to set up the cameras shutter or has some tricks to actually make it look and feel like its indiscernible to film?

Sorry if this is a daft question, just something that I have noticed.
Click to expand...

Probably many subtle issues involved that makes one Red-shot movie more film-ish than another. Some technical... I think "Knowing" went through post at Weta or some other efx/post company willing to spend a lot of time to get things right. I'm sure the indie film had a basic color-correction pass, nothing more. Some psychological -- "Knowing" looks like a "real" big-budget Hollywood movie because it is, whereas an indie movie has to make compromises to get made, some technical. Mistakes are more often left in, not reshot, you get rushed more, etc.

My basic belief is that when you make a mistake on film, it still looks like film, the artifacts generated are traditional film ones... when you make a mistake on digital, the artifacts look digital. Whenever you push an image to its limits for whatever reason, its underlying architecture comes out, so to speak.

But there are even more factors, like shutter speed chosen, type of display device used to view the movie on, on and on and on.

But it could also be philosophical -- I know when I shoot digital, I don't try to make it look 100% like film, I don't limit myself that way. I try to find reasons to be shooting digital and take advantage of what the technology can do that a film camera can't do.
 
You must log in or register to reply here.
Back
Top