More Cameras, More Problems for Digital Cinematographers
As Vice President of Software Engineering, Leconte will develop custom solutions for Post Logic’s digital intermediate clients. He was Vice President and Head of Software Development at Pacific Title and Art Studio, where he co-developed the Rosetta digital YCM archival process, leading to an a Science & Technology Award from the Academy of Motion Picture Arts and Sciences.
Film & Video spoke to Bogdanowicz and Leconte about plans for the Image Science Division.
F&V: What was the idea behind forming the Imaging Science Division?
MB: The main thrust was that there’s a large change from film-only information and film-only deliverables to the screen, now going into something much more complex. You have half a dozen different digital cameras producing different data output with different bells and whistles that can get the cinematographer into a lot of trouble if he’s not careful. When you talk about using a specific film stock with a scan, people know what you’re talking about. When you say you’re using a Viper, there are so many outputs. It’s not just a large problem for the main production, but also for the post production houses. They have to know how to use the data to integrate that into the normal workflow that exists in the current post production pipeline.
It hasn’t been a major, major problem right now because there have been very scattered numbers of productions using all-digital output. But, rightly so, it was identified as a major problem with all the digital cameras coming out and different venues such as Digital Cinema.
DL: There’s another aspect of digital cameras that needs to be dealt with. There are very good workflows for film, where you go to your edit and select what you’re going to scan, so you limit the amount of data going through your facility. With digital, it’s a torrent of data, and you have to catalog the data, to try to help the filmmaker find their way through it. And it’s essential to have good access available to all the data.
The goal is to ensure consistent color throughout the workflow at every stage of a production from acquisition through to DI, distribution and archiving. Where do you start?
MB: At the beginning, at the digital camera or film camera end. Where we take the data in and render it such that it becomes suitable for putting into the color corrector so that if it needs to be cut into other material, it can be. And put it into a sweet spot so that the colorist can work on it.
F&V: What do you mean by the “sweet spot”?
MB: When you bring in raw data, you may be pushing the color corrector to a position where if you want to alter the image with creative intent, it may not have the room to move in the direction you want. But if you process the data beforehand, so the data is in the middle of where the color corrector can move, you have more chance to get the final look you want without stressing the limits of the color corrector.
DL: Raw data doesn’t usually look terribly good when it’s loaded raw. You try the best balance of presenting the colorist with images that are familiar, that the color is familiar looking and also preserve dynamic range, to pre-process it to a good position.
MB: Each different camera poses its own individual tasks or problems. We tried to take the different cameras and put them in a common position so the rest of the workflow is streamlined and then the major decisions after that are how you are going to display it, whether it’s on a digital cinema projector or CRT for HD, or of course traditional film.
F&V: Is there a philosophy behind your approach to color science?
MB: It isn’t a philosophical question; it’s more of a physical or mathematical type of thing. I tend personally to like to work from creating the transforms, whether it be 3D LUTs or just a pipeline of transforms from first principles of true, good science rather than just some empirical type of regression. That is probably okay for a one-off scenario, but you don’t have the freedom of manipulating things the way you may be able to if you have it broken down into true scientific principles.
DL: In terms of workflow, I will be looking for speed. Clients are always in a hurry, and the schedules are so tight. Efficiency is really essential in a workflow.
In looking at that workflow, do you identify what portions may particularly difficult or problematic?
MB: One of the biggest problems is essentially rapid access to many, many terabytes of data. And also, if you have one job, that’s one thing. But if you have multiple jobs, it’s cataloguing it, making sure it’s there for the colorist or the record-out. And you also have to make sure it’s packaged to go out to the VFX houses and make sure they get it in the form they can work with it. The actual management of the data becomes an extremely difficult thing to do rapidly and without losing any. If you lose the digits, there is no film to go back on. You can’t go back and rescan it.
Which also brings up the other problem with digital acquisition, which is archivability. That’s something that hasn’t really been solved yet. We’re looking at different ways, but we’re really tormented as to whether you want to do digital archiving in digital terms or go back to some kind of film output, and both of those are difficult. This is something the whole industry has to embrace. There are some ideas, but there is no good, concrete, well-accepted solution yet.
F&V: Does archiving impact the post house?
MB: It can. The word “post” is a holdover from the old days when you got it in film, to go out to video. It’s not the same anymore. We’re up front, in pre-production, before they start shooting. As time progresses, when we have to address how to output things with regard to archiving. If we do it digitally, then we have to produce that into a form and in disc arrays that will eventually be archival. So it does involve the post production house.
F&V: What about standards? Isn’t the industry moving towards standards for DI and archiving at least-and don’t we need standards to ensure color consistency between facilities and over time?
DL: On the acquisition side, each one of these digital cameras is different in how it captures and stores data, so there is definitely a need for standards. At this point, the field is wide open.
MB: Each one of the cameras tries to show off its own brand of dynamic range and color, so the manufacturer puts a new metric together for his data output that highlights his camera. Eventually, when we get the Holy Grail of the digital camera with high dynamic range, then there’s no need to have special formats. You can go to one agreed-upon format. It doesn’t have to be one digital camera, but all the different manufacturers may migrate to a level of excellence, a standard format. That would be nice, but it’s not here yet.
What are the most significant hurdles in ensuring color consistency throughout production, post and DI?
MB: The biggest differences are the different cameras, and the spectral response of the cameras and how they see the scene. Then we have to translate that into data that represents how a human would see things, and then put that into the pipeline, into the proper data to go out either to a digital projector or film, just to make an adequate representation of what a human would see.
DL: The film cameras also had differences, but they were brothers and sisters instead of being uncles and third cousins. It’s also how the camera stores what it sees and what transforms it may be applying to the data that the sensor perceives. That’s another big challenge.
MB: The one thing about cameras is that a lot of cameras are putting things out onto tape in a compressed form. Once you compress, you’re losing something so you’re starting off by losing. It’s just like taking film and doing a 1K scan and eventually wanting to go out to 4K digital cinema. There’s no way you can get the quality by doing that.
F&V: Your aim is described as creating project-based customized digital workflow. Will they be customized based on the digital camera used?
MB: Yes, based on the camera. But even in the film world, almost every production is a research project in itself. And so to that extent, for example if we deal with a Dalsa camera on one project, then we would know how to best treat that on the next project, but we may need to do some specific in-house rendering because of the type of content, or deliverable or intent of the project producers on the data.
DL: It’s a plug-in kind of architecture where you have guidelines based on the camera but additional customizations based on a project-by-project basis.
F&V: There are other solutions out there for color consistency from Kodak/LaserPacific, Technicolor, Gamma & Density. What’s missing there that you are addressing?
MB: The Technicolor system is essentially giving you printer lights, which is just like when you took a negative and went to print. That gets something close. But it doesn’t impart any color rotation or any specific type of color transforms; it’s color changes in the simplest form in a way. The Gamma & Density system sets things up in a certain way, and again, that’s fine but it’s setting things up but doesn’t transform the data, it just sets you up in a position. And the Kodak/LaserPacific aim system is very similar to the Technicolor system, with slight differences. They’re using it as a communication tool.
We’re essentially color-system agnostic. We don’t care which system you want to use, we can deal with all of those, or someone could come in and we could create a look here on the computer and in his dailies or in the actual suite, he will see that same type of look.
These other systems can be somewhat complementary. We’re doing what they do, but we can go a little bit further. If someone wants to shoot a day-for-night scene, for example, we can create a 3D LUT that can take the data of a full negative and make it look like it was shot day-for-night. And if they want to change the intent for some reason, we have the original data and can process it another way. And it works on digital cameras.
F&V: How much of the pipeline will you be creating in-house?
MB: A lot of the color science part is in my mind, so that we have to put more into code and of course Denis will be doing a lot of the pipeline and rendering code. That’s only part of it. Another part will be dealing with the production and the cinematographers and directors early on in the process and essentially taking their ideas of how they want things to look to make sure that we can deliver that. And also, to take some of the potential “gotchas” so they know they can go a certain way but they’ll have these problems. We’ll try to come to a happy medium to minimize the errors or problems in the system early on while we’re trying to manipulate the data and get the best look for the client.
DL: With so many new digital cameras, it’s very confusing and I’d imagine that productions are in a serious quandary about which camera to use. We hope to enable them to reach a level of comfort that they’ll have what they want no matter which camera they use. We’re collecting the list of pluses and minuses for each of these digital cameras and will be able to help them decide or to help them avoid problems to get the best out of the camera.
F&V: Are you working on a project now?
MB: I’ve been consulting here for a while, and Denis has just joined as of last Monday. We’re looking at a project starting this week, but I can’t say much about it.
Did you enjoy this article? Sign up to receive the StudioDaily Fix eletter containing the latest stories, including news, videos, interviews, reviews and more.