Awesome post Mike! Really appreciate your time explaining some of the issues and solutions you are facing - thankyou :)
In all serious though, if raw was preserved all the way through to grading (apart from vfx shots and others which are forced to bake in a look), then it doesn't matter how it's been originally shot (again, with exceptions), it doesn't matter if someones screwed up the RedLogFilm proxies by crushing your blacks prematurely, it doesn't matter if incorrect white balance has been set or if people have used RedGamma instead of RedLogFilm etc - all of this is correctable and should result in some of the best matching footage because of it's malleability. Obviously this doesn't work so well with your workflows for various reasons (as you have explained), but it's interesting to note that what you regard as a problem for R3D files is an advantage in my mind :)
Regarding a colorist not deciding what files they are given, I completely understand this, however I would have thought a post-production supervisor would have tried to create a workflow that hands them the best possible files to work with (I'm not sure if you work in this role, so this isn't a jab at you). Given what you have explained, I understand why passing a colorist R3D's in a TV episodic environment is not always (if ever) feasible. Still seems a bit of a shame to me though.
On another note, is 4k re-release in the mind set of TV content at the moment? If so, how is this being provisioned for? Are 4k ProRes proxies being used? I guess the fact that the "A cam" Alexa is shooting 1080p Prores, probably answers some of my question. I guess this is another area where maintaining an R3D pipeline seems advantageous - it would enable a really simple re-release in 4k down the track - but it seems that such a pipeline is not easily integrated into standard workflows. Maybe NAB shall shed some more light on where the industry is headed. Maybe we'll see Arri's 4k solution - who knows :)
Bear in mind that we worked with film for 20-30 years prior to going all digital. In the case of film for TV, we generally work(ed) with "best light" scans that essentially compare to RedLogFilm: images that are relatively balanced in terms of RGB levels, nothing crushed or clipped, in log space (or something approaching log space). A simple overall correction or LUT brings it into Rec709 color space and allows color correction relatively quickly, depending on the complexity of the look the client wants.
This is a time/money issue that generally can't be done as quickly or as easily as you might think. Look at a major motion picture, for example, that has maybe a 3-week schedule (at best) for color timing, with VFX coming in from 7 or 8 vendors in different countries. I guarantee you, all the effects will look completely different and will require effort to drop into the movie without major changes. In some cases, I've had situations where I've had to change the look of the scene because the visual effects were so far off, and there was just no time available for the VFX company to change the grade on their end. I once had a situation where the VFX company went out of business after they delivered the effects, so they couldn't even change them if they wanted to -- they no longer existed!I understand that the results from VFX houses can be unpredictable, but I don't understand how converting all your non-vfx material to the same format as the vfx house fixes this for you? It seems like it would be preferable to try and bring that vfx footage up to match the non-vfx footage rather than change all of your non-vfx footage to match the effects?
Not for time and money's sake in a longform project, no. My advice would be to go to NY or LA and sit in on an 8-hour color-timing session for a major TV show, and note how quickly the crew has to work in order to get a one-hour network show done and delivered. Typically, the colorist works by his/herself on the first day, timing the entire show. The second day, the client comes in and watches the show in real-time and makes notes and changes. If many alterations are needed, this session might go long. At the end of the session, the colorist outputs the final delivery version as a file or to tape, depending on the client's needs.This seems like advocacy for staying in raw all the way through to the grade!
Note, BTW, that even Social Network and Girl with the Dragon Tattoo were not done directly from the R3D files -- they went from intermediate DPX files instead, just to speed up the workflow and conform, and lighten the processor load on the color-correction system. Michael Cioni has gone into detail about this in past posts on his blog and here on Reduser.net.
Nobody cares yet. I bet 4K will be commercially viable about the time that 100% of every American cable channel is available in 1080 HD -- and we're a long way from that happening. My guess: maybe 7-10 years for 4K broadcast, but a lot will depend on the economy and what the networks and studios decide to do.On another note, is 4k re-release in the mind set of TV content at the moment?
agreed, keep up the good work and get some friggin sleep, you guys must be beyond overtime now.
I think people are only looking at the distribution methods we have today, without considering how that could change in the future. So far, the Internet has barely even touched cinema distribution.
I know that's going to change. It already is, not just on the high end (RED) but on the low end too. No one is going to be "waiting" on the broadcasters or cable; instead, new companies will route around them.
RED is part of making that happen. And 4K is where we want to be when it does (soon).
If I'm wrong about Social Network or Girl with the Dragon Tattoo being done from DPX files, perhaps I misunderstood Michael's lectures and blogs. One thing I will add is, in some cases, we get in DPX files for color correction that are "dodgy" and require rescanning (or reoutputting, in the case of digital files), just to give us more range. As long as it's a true log file -- like a RedLogFilm file -- it generally will work, though my preference is to start off with a balanced correction. I don't mind seeing the temp onset correct as an additional file to refer to, and a corrector like Baselight does allow dragging along two simultaneous files, using one for a "match to" comparison. But much too often, the DP will say, "just use that as a general ballpark -- don't match it 100%," and we wind up discarding it.
Congrats on the laser projector, BTW. I have no doubt there are many nervous Sony and Christie executives at the show. If you can get this out for $10K by the end of the year, it'll spread like wildfire in Hollywood.
|« Previous Thread | Next Thread »|