Home of Professional Cinematography since 1996
Digital Film Grade
>OK. I've just seen a trailer for 'Oh Brother where art Thou', and the colouration of the backgrounds in all the exterior shots confounds me. The sky is blue, the skin tones are..... well they're skin toned but all the foilage and landscapes are a uniform saffron yellow. I know that no combination of filters, enhancers, cunning three light timing or other lab alchemy can produce such a pronounced and localised affect to the image so short of using crop dusters and lots of yellow paint I'm guessing this is produced by grading a digital intermediate. That or Roger Deakins is in league with the devil. Can anyone cast any light (sorry) on the subject for me.
>For the new Coen brothers film shot by Roger Deakins, I believe the entire film was digitized on a Spirit, manipulated as Deakins saw fit, then output back to film. Deakins wanted to achieve a certain look that he found impossible to do consistently in camera or through standard processing and timing, so he chose the digital route. If he's not too busy on his 24P feature, perhaps David Mullen could shed more light on this, as he actually met Deakins at a camera house last year and discussed the matter with him.
>Would that be "Oh Brother, Where Art Thou?" ? Curious, as it was shot about ten miles from my home here in Mississippi. Alot of my local crew worked on it. "Tweaking Deakins", they called him, but more in the vain of tweaking 18ks bounced into 20 bys.
>-- Jim Dollarhide
>Yep! and it's the centre of a lot of interest because of it. 2K transfer/scanning, then grading at Cinesite (not during the transfer as in DaVinci-style colour correction). There's some discussion at
Group Technology & Services Manager
>I don't have much more to add regarding the use of the Spirit/Cineon combination, except that the film was shot in Super-35 and transferred as a 2.35 anamorphic image. The film was desaturated digitally and output to film.
>Some of Deakins' casual comments to me that I hopefully remember correctly: when I first ran into him at Otto Nemenz, he was testing the lower-contrast 5277, hoping that it would give him the desaturated look he wanted. At that time, he was also concerned about the higher contrast of Vision print stock, which he felt was a problem for a more pastel-looking period film. Later, I talked to someone at the 1999 ShowBiz Expo at the Kodak Cinesite booth who told me that once the decision was made to do the desaturation digitally, Deakins decided to use 5293, which he was more comfortable with. That's also when I learned that the film was being shot in Super-35 and would be released in 2.35 anamorphic prints.
>Then later I spoke to Deakins on two occasions at events at the ASC clubhouse. I asked him if he felt that the 2K resolution was adequate for his project and his reaction seemed to be that it was OK but that 4K would be better for some shots needing more detail -- he might have said that some digital effects shots were being done at that resolution (not using the Spirit) at his request, but I'm not sure. He was very candid about the pluses and minuses of doing the digital intermediate process. He sort of scoffed at the notion that it was a "faster & better" method of producing intermediates, considering all the time (and money) that he was spending at Cinesite trying the get the color manipulation to his satisfaction. He also said that the more extreme the degree of color manipulation, the more certain digital artifacts would appear, like noise problems.
>I later asked how they were going to handle the bulk release prints. Since the 2K file could output as many negatives as needed, I thought that this would be the method of creating extra negatives for striking prints. But he said that while this was ideal, it was too expensive, so they were trying to now output a color interpositive so that they could strike dupe negatives (sort of eliminating the saving of in the number of generations). But the colors were coming out differently when output as a positive, so he was fiddling again with the color timing. So some "show" prints will be struck from the output negative, and the rest of the prints made from a dupe negative struck from the output interpositive.
>I then later asked him (I can be a pest...) about how they were going to handle the home video version. The 2.35 anamorphic film output (negative or positive) could be re-transferred or the 2K 2.35 digital version could be used for the widescreen home video version, but in order to use the whole Super-35 area for the 4:3 full-frame version, he was having to re-transfer the movie again from the original negative (I think), since the original Spirit transfer had cropped the image to 2.35 and converted it to an anamorphic image. (At the time I wondered why Cinesite didn't just transfer the whole Super-35 frame and then later digitally crop & convert it anamorphic, but then I realized that the 2K limit of the Spirit meant that it was better to use all the available resolution for the 2.35 picture area.)
>So my impression was that the whole digital intermediate enterprise was more complicated than it looked on paper, but Deakins still felt that this was the only way to create the colors he needed for the project. Hopefully there will be an article that will more accurately describe his techniques, because my memory is probably faulty. Deakins is a great guy to chat with, by the way -- very down-to-earth, very candid, opinionated in a refreshing way, and not pretentious at all, even though he is a great visual artist.
>David Mullen Cinematographer / L.A.
>These threads are really a mystery to me.
>On the one hand the CML/cinematography members fall over each other to decry originating their deathless epics on HDTV on the premise that film is by definition several millennia advanced over this lowly, crappy video.
>On the other hand their films blithely make a detour through a Spirit or DaVinci, mostly to ENHANCE whatever appeared on their film stock in the first place. And then is dumped back on film.
>Pray, what IS the secret?
>If this detour does not deteriorate the film image, what stops Sony or Pana to take whatever wondrous chips the Spirit posesses and slam them in the nearest Ikegami or whatever.
>The argument that HDTV cannot digest the enormous latitude of film SEEMS to be erroneous, if that same width is accommodated easily and routinely by that Spirit.
>Where am I wrong?
>Robert Rouveroy csc The Hague, Holland
>Where you are wrong is in believing Sony. Despite their PR machine claims, the current HDTV camera head and imager, in practice, simply cannot capture the dynamic range, resolution and bandwidth that 35mm film can handle. Period. The CCD creates artifacts when pushed that do not look natural nor organic, as we're used to seeing (clipping, for example). So the electronic "camera original" is inferior to film negative in the first place, at least with the present generation of HDTV and 24P electronics. That same signal is then compressed severely just to make it recordable on tape. Another minus for your original image.
>However, in post-production the Spirit and other datacines (1) don't have to capture your image back in real time, and (2) don't have to use small, portable, battery-powered tape drives for storage. You're talking about a combination "electronics package" that only has to fit in a room, not on a tripod. This allows for bandwidth recapture and data storage solutions impossible in a portable camera.
>Jim Furrer VGG Systems, Inc. Dark Street Films
>Mabye not a cml question... but related to this topic.
>What is the bit-debth of a Spirit transfer? I see telecine and scanner used interchangeably and it confuses me. I understand that they are converging technologies, but a true 14-bit system and a sub-sampeled system seem quite different when you consider swinging your timing from an OCN.
>Doug "really HD-2-film?" Delaney
>Robert Rouveroy writes :
>Where am I wrong? . . . . .let me count the ways:
>(I've just deleted the first few, because we've read them all time and again.)
>But the main point is: "horses for courses". Some projects demand exotic colour corrections, as in "Brother where art thou", (apparently - I have yet to see it), which must be at the cost of some other factor. Another will look for needle-sharp images with perfectly natural colour "as photographed". Great - that's what Kodak and Fuji make the stock for. Yet others gain some other digital benefit, which at this stage comes at the cost of sacrificing ultimate quality.
>It's interesting to look at the few recent pictures shot or post produced digitally.
>There is a reason for every one, never to do with image quality (resolution). Pleaseantville - Spirit transfer and grade for the colour drain and partial b/w. Chicken Run - Cineon (not Spirit) transfer for compositing and all the necessary work to make 100 minutes of claymation achievable. Timecode - shot on Digibeta for the 92 minute continuous takes. Blair Witch Project shot on DV (format?) - no-budget, home movie look and feel. Star Wars Ep 2 - studio shoot on HD, backgrounds on ???. To make a point. All-action movie, doesn't really rely on the full gamut of colour and resolution.
>The list of films that were shot on 35mm film and stayed on 35mm film throughout the process because they needed to, is much longer. But none of the above films could have been made without the digital phases they went through.
Group Technology & Services Manager
>The argument that HDTV cannot digest the enormous latitude of film >SEEMS to be erroneous, if that same width is accommodated easily and >routinely by that Spirit.
>>Where am I wrong?
Where should I start?
>Remember, one of the great things that film does is compress the dynamic range of real life into something a bit narrower. When the Spirit (or other telecine) gets hold of a piece of film, the range has been greatly compressed in a non-linear manner that most consider very pleasing. There are no chips out there yet (CMOS chips show promise, but have other problems) that can handle this extreme dynamic range.
>Remember also that the Spirit isn't sampling an entire frame at a time. It uses a linear array CCD which samples a single line of data at a time... far simpler than capturing an entire frame.
>Also, when you mention HD and mention the Spirit/daVinci combo, you're not talking (in this instance) about the same thing. People here are speaking of using a Spirit in DATA mode, scanning neg at 2K res. HDTV is roughly half that... the bad half.
>Jeff "HD patrol" Kreines
>Robert Rouveroy wrote :
>On the other hand their films blithely make a detour through a Spirit or DaVinci, >mostly to ENHANCE whatever appeared on their film stock in the first place.
>> Where am I wrong?
>Film excels as a capture medium. There is more in a 35mm neg, from the deepest shadows to the brightest highlights than you can even see ALL AT ONE TIME in a release print, or in a video transfer. What Deakins seems to have done (I haven't seen the movie) is use the Spirit (and whatever grading system was used subsequently) to extract information from the neg selectively, to give a look that could never have been achieved either with an HD camaera looking at the world, or from using raw HD footage and then subsequently playing with the colours.
>In a nutshell, a film > digital grade > film path allows your to MANIPULATE information, not to ENHANCE (as you suggested).
>Also, the work would have been done (I imagine) using the Spirit in data mode - so the images would have been 2k 30bit log - which has considerably more colour depth than HD.
Computer Film Company,
> "So, laying eggs all your life and then getting stuffed and roasted, that's good enough for you, is it?"..."It's a living."
> Robert Rouveroy wrote :
>These threads are really a mystery to me.
>Robert, please stop being deliberately obtuSe.
>A number of us have said on many occasions that digital origin is crap, digital post is great, digital projection id getting there.
>I've also said that some people, I think that was the word that I used, deliberately misunderstand this difference to advance digital imaging as a whole.
>Once again, most of us have to deal with reality, we don't have the luxury of retired theorising.
>I realise that this may be offensive but PLEASE!
>Dominic Case wrote:
>Chicken Run - Cineon (not Spirit) transfer for compositing and all the necessary work to make 100 minutes of claymation achievable.
>Can I just - for the record - point out that the digital mastering for "Chicken Run" used no Cineon/Kodak technology other than the basic image format (30 bit log - with some CFC mods to the Kodak standard) and the 5244 intermediate stock for the laser recording. All else - the pin reg film scanner, the colour grading system and the film recorder were either CFC Digital Lab proprietary or non-Cineon/Kodak off-the-shelf (eg ARRI).
>But thanks for the mention of the movie, Dominic! We're kind of pleased that "Chicken Run" has turned out to be the first major 35mm movie release to have 100% digital film mastering (if you ignore the Cannes screening for "Brother Where Art Thou" and you also ignore "Phantom Menace" (it's best to) :)
Computer Film Company,
> "So, laying eggs all your life and then getting stuffed and roasted, > that's good enough for you, is it?"..."It's a living."
>Robert Rouveroy wrote :
>This video technology is unfolding exponentially
>This just isn't true.
>The quality of images produced using video/digital origin techniques has, if anything, dropped in the last 20 years.
>HDVS when introduced in '82 was a huge move backwards, HDCam doesn't have the same image quality now.
>So where is the exponential increase, it doesn't exist!
>For anyone actually earning a living in this field, or wanting to, YES, keep your eyes open but don't be conned by the hard sell/theorising.
>It isn't happening that fast.
>Or it's not happening that fast in origination, there isn't the size of market for the H&D required.
>The post and projection parts of the equation are subsidised by other industries, where do you think motion tracking came from?
>Robert Rouveroy wrote:
>On the other hand their films blithely make a detour through a Spirit or DaVinci, >mostly to ENHANCE whatever appeared on their film stock in the first place. And >then is dumped back on film. Pray, what IS the secret?
>For me, the secret with originating on film, even for conventional television, is that the negative captures so much more information than you'll ever use. Yes, in the post process, you have to decide which bits to use, but if you need it there is a lot more there. In electronic imaging you have to make the same decisions, but on set when you are making most of the rest of the decisions about the look of the image. Film gives you the luxury to make those decisions in the relative comfort and detatchment of the lab or telecine suite.
>Yes, video does give you some room to move, but not nearly as much.
>Mitch Gross wrote :
>Deakins wanted to achieve a certain look that he found impossible to do consistently >in camera or through standard processing and timing, so he chose the digital route.
>This is one of the reasons Roger Deakins approached Cinesite for help. He wanted a dry, dusty look for the new Coen brothers' film, "O Brother, Where Art Thou." Shooting in Mississippi where the vegetation was very green and lush, he hadn't been able to achieve the look he wanted by conventional means. He was shooting Super 35mm for an anamorphic release. The people at Cinesite advised him to expose the film straight with no filters and then do the color grading digitally.
>Here's the post path according to the people who worked on the project: The cut negative (A and B rolls) was graded using a Pandora MegaDef color corrector and transferred on the Spirit Datacine at 1828x1556. The anamorphic squeeze was applied digitally in the Spirit for efficiency and to save a generation. Color grading was done using a Cineon-calibrated SGI display. The opticals were all done digitally using Cineon software. The digital effects shots were graded on a Philips Specter Virtual Datacine at Warner Brothers and then tweaked in Cineon.
>Grading was a real challenge, according to Julius Friede, senior colorist at Cinesite. Roger wanted some fairly extreme changes, particularly in the greens.
>Using a Kodak Lightning 2 film recorder, the completed reels were recorded to film -- an internegative for making show prints, including what was shown at the Cannes Film Festival, and an IP for subsequent release printing.
>The Spirit transfers were done to data, as 10 bit logarithmic Cineon files -- in other words NOT HD. Cinesite captures and manipulates the density range of the negative. Because of that, the range was there for Roger Deakins to do additional tweaking in the lab if he wanted to.
Don Ver Ploeg