Zodiac

[Fergus Meiklejohn]

I just watched the HD trailer for David Fincher's new film, Zodiac. Harris Savides was the DoP I think..? He basically used the RED workflow - not exactly the same of course, but all digital, no tape. Filmed on a Viper. I think it looks beautiful. And I guess if RED is used as intelligently as Harris Savides used the Viper, it could produce as beautiful a product?? :w00t:
That's an exciting prospect huh...
I'm just a poor Navajo that's been to town..; does anyone have any more info on how they filmed, lit, and graded it?
I love how "real" it looks, with characters sometimes allowed to become too dark and all the office shots graded to look flat nasty green.

Respect:love:

[Brandon Fraley]

American Cinematographer did a pretty hefty article on Zodiac last issue. not sure if you can find it online.

What I CAN say though, is that if RED does what they promise (like we all know they will) the RED ONE will be a much more significant tool than the Vipers used for Zodiac.

[Clayton Harper]

I think Zodiac turned out well DESPITE the Viper and totally to the credit of Savides.

Of course, Red will be the one true camera to rule them all. :devil:

[jamesedwelland]

I thought it looked good given that Savides wanted it to look straightforward and not anything like Seven.

I did notice detail in night sky (which I liked) and quite a few instances of blown out highlights. The greater DoF was not distracting, but noticeable in the newspaper meeting room. The lack of resolution was noticeable sometimes in wide shots, with close ups looking very crisp (reminds me of how good 16mm can look in c/u but fall apart in wides) I felt that a little texture from the print stock helped. A few shots did look a little two dimensional. Given that Savides said he wanted a slightly 'newsy' 'just tell the story' sort of look it worked.

Worth checking ot the Am Cine artice for workflow diagrams.

Worth remembering too that its not the technical stuff in itself that makes a good movie....

james w

[Tom Lowe]

Here is the AC ARTICLE (http://www.ascmag.com/magazine_dynamic/April2007/Zodiac/page1.php), which goes into a lot of detail.

I hope this comes out on HD-DVD or Blu-Ray soon so I can see it in HD.

[Jason Murphy]

Savides' work in Zodiac is fantastic, and last month's American Cinematographer is an excellent read on the process. Don't know how many people on this board have seen the movie, but with only a few exceptions, he and Fincher do their work to take advantage of almost all the Viper's strengths. The image is still of noticeably digital origins (mainly because of lack of grain), but it's the first time while watching a digitally-shot movie (which wasn't specifically going for some video aesthetic a la Miami Vice) that I felt it looked better acquired digitally than it would have on film.

Granted, much of that is because it was shot in RAW/uncompressed mode, but it's pretty clear that Savides had to do quite a bit of lighting (especially in interiors) to work around the camera's limitations. It was well worth it, though. The movie looks amazing. It's a great movie in its own right, too, which doesn't hurt.

[Fergus Meiklejohn]

thanks Tom

That's an interesting article. Key quotes would be:

"In the future, I see smaller, better cameras; fewer crewmembers needed to service them; and a method of production that’s freer and less encumbered by technology." - (Surely he means the RED 1..:usd: )

and

“With the Viper, low-angle shots aimed right up into the fluorescents were tough,” notes Savides. “The difficulty was in controlling the look and not letting those sources overexpose. We were shooting with multiple cameras, so I had to figure out how to beef up the exposure around the actors — allowing me to stop down a bit — while staying true to my sources and also keeping fixtures clear in the wider shots. It was a weird, frustrating balancing act. Incidentally, it would have been much easier to do if we’d been shooting on film — the shoulder is a weak spot on the Viper.”

Which does lead me to question how the RED will perform. I like the incredible, grain-less, detail; that crisp reality. But the luminescence curve is video's achilles heel. I even thought I saw traces of that video look in Peter Jackson's test film. I'd be very interested to know what others thought about it and to hear any suggestions for of how to deal with it.:cold:

[Ace]

thanks Tom

That's an interesting article. Key quotes would be:

"In the future, I see smaller, better cameras; fewer crewmembers needed to service them; and a method of production that’s freer and less encumbered by technology." - (Surely he means the RED 1..:usd: )

and

“With the Viper, low-angle shots aimed right up into the fluorescents were tough,” notes Savides. “The difficulty was in controlling the look and not letting those sources overexpose. We were shooting with multiple cameras, so I had to figure out how to beef up the exposure around the actors — allowing me to stop down a bit — while staying true to my sources and also keeping fixtures clear in the wider shots. It was a weird, frustrating balancing act. Incidentally, it would have been much easier to do if we’d been shooting on film — the shoulder is a weak spot on the Viper.”

Which does lead me to question how the RED will perform. I like the incredible, grain-less, detail; that crisp reality. But the luminescence curve is video's achilles heel. I even thought I saw traces of that video look in Peter Jackson's test film. I'd be very interested to know what others thought about it and to hear any suggestions for of how to deal with it.:cold:


I think the secret to shooting with red would be to underexpose slightly.. The amount of detail that is recoverable from underexposed areas is pretty astonishing as demonstrated by David Stumps tests.. I think it was Boothba who demonstrated a completely recovered image from 4 stops underexposure (albeit a bit grainy).

[Alex Boothby]

Ya - way back in November '06.

http://www.dvxuser.com/V6/showthread.php?t=78368

Sadly I think I'm the jerk that accidently got "8 1/2 stops of latitude" in some people's heads by wording the thread title very poorly - my bad. Anyway this seems but a footnote, as recent improvements have promised to yield greater dynamic range.

On a similar topic I was surprised to read an article on "Balls of Fury", an upcoming Genesis pic, on which the DOP decided to OVER expose slightly as the Genesis was quite noisy in the blacks.

[David Mullen ASC]

The terms "over-exposure" and "under-exposure" are a bit inaccurate (even though we all use them) in some ways because the "correct" exposure is the one that gets you the look you want.

Especially in the case of digital cameras were there really isn't an exact ASA, although film doesn't really have one either (the "E.I." rating was actually invented as a way of bending the sensitivity rating from a technical measurement of density per exposure to some sort of "optimal" rating determined by the manufacturer.)

I can see one person preferring the look of the RED camera at 250 ASA generally while another preferring the look at 400 ASA, for example.

I've been seeing a lot of Genesis-shot features in theaters lately ("The Lookout", "Planet Terror", "Reign Over Me", "Next", trailers for "SuperBad" and "Balls of Fury", etc.) and noticed that many of them have a consistent noise level that's not unpleasant, sort of like faster film, but I can see wanting a cleaner signal at the expense of some bright highlight detail, especially in a comedy (like "Balls of Fury") with a lot of sets being lit where you can control the highlights with lighting and art direction.

I'd also point out, regarding "Zodiak" (which I thought was excellently photography by Savides and perhaps the best looking HD feature shot to date), that the whole movie ended up going through an expensive noise-reduction process at John Lowry's company, which specializes in film restoration work. This accounts for the very consistent noise level throughout the movie. All of Fincher's movies have some very underexposed scenes in them -- the people doing the D.I. for "Panic Room" have mentioned in the past about the amount of grain reduction they had to do for that movie.

[laguun]

The terms "over-exposure" and "under-exposure" are a bit inaccurate (even though we all use them) in some ways because the "correct" exposure is the one that gets you the look you want.

Especially in the case of digital cameras were there really isn't an exact ASA, although film doesn't really have one either (the "E.I." rating was actually invented as a way of bending the sensitivity rating from a technical measurement of density per exposure to some sort of "optimal" rating determined by the manufacturer.)

agreed.



I can see one person preferring the look of the RED camera at 250 ASA generally while another preferring the look at 400 ASA, for example.

i am not agreeing here.

not at all, to be precise.
it is interesting how fast quality issues can shift.
when i started in the biz (35mm, late 80ties) grain was evil.
when i started digital hdcam (2/3, ~02/03) grain suddenly was cool, and noise was bad.

noise and grain are pretty much the -exact- same thing.

later on there have been several discussion regarding the lattitude of mechanical aquisition (film) and digital aquisition (sensor). lattitude basicly is all about the difference between grain/noise and signal.

this clearly demonstrates the amount of uneducated workforce in the industry. emotionally calling it noise OR grain doesn´t change the fact. they are bot the same. s/n.

however, its easy to add some 500 grain to a 200 or hdcam noise level.
it is hard, however, to reduce the noise (aka grain) of a 500 to 200 or hdcam levels.

[David Mullen ASC]

I'm not exactly sure what you're disagreeing with when I say that people are going to rate the sensitivity of the RED camera to suit their tastes... as was pointed out, people aren't rating the Genesis camera the same way.

Some people are going to prefer more speed and live with some noise, while others are going to prefer less noise and live with less speed. This is one reason why digital camera manufacturers are loathe to say that the camera has a specific "ASA" rating.

Noise and grain aren't exactly the same thing -- the film image is basically made up of grains (or dye clouds in color), but the digital picture isn't made up of noise.

[Alex Boothby]

All of Fincher's movies have some very underexposed scenes in them -- the people doing the D.I. for "Panic Room" have mentioned in the past about the amount of grain reduction they had to do for that movie.

I worked on Panic Room and can certainly attest to that. Fincher loves the toe. Shame that Darius Khondji got sacked early - I believe that he was fighting for a bit fatter negative.

[Tom Lowe]

Noise and grain aren't exactly the same thing -- the film image is basically made up of grains (or dye clouds in color), but the digital picture isn't made up of noise.

True.

I think people are going to be a little shocked when they see how much noise they are going to get in low-light conditions when shooting with anything over 400 or 600 ISO on the RED, especially if they are trying to maintain full 4K res.

Just go to Flickr and take a look at some of the sunset and low-light shots people have uploaded from their DLSRs at near full resolution -- there is a lot of noise, and it aint "grain." If you look on "more properties" you can see the EXIF data, which tells you the ISO and shutter speed, etc, that the image was taken at. Essentially the RED sensor is a sooped up DLSR-style CMOS sensor, so you can imagine that similar results are going to be had under similar conditions. Every DP who is planning to use the RED, I hope, already owns and is using a DLSR camera, to make themselves familiar with that type of sensor and shooting RAW.

[Jason Murphy]

I'd also point out, regarding "Zodiak" (which I thought was excellently photographed by Savides and perhaps the best looking HD feature shot to date), that the whole movie ended up going through an expensive noise-reduction process at John Lowry's company, which specializes in film restoration work. This accounts for the very consistent noise level throughout the movie. All of Fincher's movies have some very underexposed scenes in them -- the people doing the D.I. for "Panic Room" have mentioned in the past about the amount of grain reduction they had to do for that movie.

When I saw the movie, I was really surprised at how clean it looked; there was very little noise even in scenes that looked as if they were lit by only a flashlight (or by the LCD on some PA's cell phone - it's true - Fincher does love that toe). This would explain it.

Also, while a lot of attention has been paid (and rightly so) to some of the stunning scenes shot in low light, one of my favorite sequences in the film (from a photographic perspective, anyway) was the murder at the lake early on in the film; the colors and feel of the waving grass and the ripples on the lake were both subtle and beautiful, and it looked quite unlike any film stock I've seen. Shame the movie went off of most screens in only a few weeks; I meant to see it a second time, but by the time I got around to it, it wasn't playing in any of the local theatres anymore. Such is life, I guess.

[Keith Alan Morris]

thanks Tom

...the shoulder is a weak spot on the Viper.”

Which does lead me to question how the RED will perform. I like the incredible, grain-less, detail; that crisp reality. But the luminescence curve is video's achilles heel. I even thought I saw traces of that video look in Peter Jackson's test film. I'd be very interested to know what others thought about it and to hear any suggestions for of how to deal with it.:cold:

YEP. It's all about the SHOULDER.

[Keith Alan Morris]

AC on Zodiac: “[The concern] was mostly about workflow and how we were going to ingest the data,” says the director. “It was also about getting used to watching things through a 32-inch LCD HD monitor and being able to direct from that. The image is kind of flat and green, but it offers an immediacy that allows you to hone and shape the image. And it’s all in real time, so when you’ve shot the shot, you know you have it.”

Why do you guys think it was "flat and green?"

[P Andersson]

everybody now:

head and shoulders knees and toes, knees and toes
head and shoulders knees and toes, knees and toes

[David Mullen ASC]

Why do you guys think it was "flat and green?"

It's flat because that's the look of a 10-bit Log image unless it has a LUT applied to give it the correct gamma for a video display device.

As for the green look, Dave Stump's explanation to me is something like "the Viper isn't green -- reality is green and the Viper is just accurately reflecting that." In other words, there a lot of green in reality and either you can suppress it by making the green CCD slower or lower the green channel signal (or using a magenta filter on the camera, as they did for the day scenes in "Zodiak") or you can live with it and correct the image in post.

There is a lot of controversy surrounding the need for the magenta filter on a Viper. Some people feel it helps the three channels clip more evenly in bright sunny scenes, others feel it can actually reduce dynamic range a little. It's unclear what the best approach is, but the other issue is the monitoring -- besides the LUT to correct the gamma for video display, you have to design a LUT or something else to correct the green cast out. This only affects the monitoring, not the signal going to the recorder (unless you use a magenta filter on the camera.)

[Keith Alan Morris]

will footage to a monitor onset look "flat and green" on the RED also? I'm planning on hooking up my 35 inch HD monitor for a one location movie... Just dont want any surprises...



It's flat because that's the look of a 10-bit Log image unless it has a LUT applied to give it the correct gamma for a video display device.

As for the green look, Dave Stump's explanation to me is something like "the Viper isn't green -- reality is green and the Viper is just accurately reflecting that." In other words, there a lot of green in reality and either you can suppress it by making the green CCD slower or lower the green channel signal (or using a magenta filter on the camera, as they did for the day scenes in "Zodiak") or you can live with it and correct the image in post.

There is a lot of controversy surrounding the need for the magenta filter on a Viper. Some people feel it helps the three channels clip more evenly in bright sunny scenes, others feel it can actually reduce dynamic range a little. It's unclear what the best approach is, but the other issue is the monitoring -- besides the LUT to correct the gamma for video display, you have to design a LUT or something else to correct the green cast out. This only affects the monitoring, not the signal going to the recorder (unless you use a magenta filter on the camera.)

[David Mullen ASC]

(Someone correct me if I'm wrong...)

The RED camera will send out a downconverted HD signal (720P, 1080i, 1080P) -- for monitoring, you'd probably use single-link HDSDI 4:2:2 output. Now as to whether that output signal is Log (flat) or Linear (for display), I think that is a menu choice, although probably single-link HDSDI is always linear video and you can only choose Log if doing dual-link 4:4:4.
http://www.red.com/techspecs.shtml

If you de-bayered the recorded 4K RAW image on a computer and looked at a 10-bit Log 4K RGB file, it would look somewhat milky and low-con unless you applied a viewing LUT, which is all part of the REDCINE features.

The green thing is only particular to the Viper camera. Other cameras that create a 10-bit Log file do not have the green issue.

[Adam Jeal]

I just watched the HD trailer for David Fincher's new film, Zodiac. does anyone have any more info on how they filmed, lit, and graded it?

Here's the workflow that they used both in the shoot and post.

695

696
I can't seem to upload these jpeg's properly, so here's the links to the original PDF's;

http://digitalcontentproducer.com/images/608MIL_zodiac_viperonsetacquisitionworkflow.pdf

http://digitalcontentproducer.com/images/608MIL_zodiac_editorialpostdailiesworkflow.pdf


adam