Digital Capture From 35mm camera
Posted 02 February 2011 - 08:13 PM
Testing is a few months away and we are simply trying to beat this thing out in theory.
Our previous projects have all been photographed digitally (ex3 and Red) and are very comfortable with our post workflow on these medias. For aesthetic reasons, we have decided to go the film route. Originally we had opted for super 16mm except the film itself requires numerous motion graphic and compositing effects done in the digital intermediate and demanded a little more wiggle room from the acquisition format. The show itself will already have two cameras; the principle sync camera (super 16mm) and the phantom HD for high speed shots. Lugging a third camera (35mm) into the mix for the SFX shots is currently out of the questions. So we have decided to go with an Arricam ST configured for 2-perf acquisition as our principle camera.
What we have done on past film shoots is to have the film telecined to low resolution proxy files (QT 720p) for dailies and offline editing. Once we have locked picture, we would return the EDL to the post house and then commence our 2k scan. We’d then conform the 2k files to our edits and composite/grade for the final package.
I’d like to bypass the first step after processing by doing the following;
Arri now has the HD-IVS video assist that can stream and playback 1080p video. While using this device to monitor shots and playback, I’d also like to capture from this source and use the captured footage as our dailies and offline proxy files. We would use the NanoFlash to capture the proxies to CF cards and then ingest them into our offline system. So instead of scanning everything in the lab, we do it on set and potentially save thousands of dollars in lab fees.
Once we have picture lock, we go to the lab and have the footage we need scanned at 4k and then we finish up with our motion graphics and grading before the show is packaged.
The idea dawned on me after watching some of the NAB videos for the Alexa and Penelope delta. Both camera’s can simultaneously record their uncompressed data (for online) and a compressed proxy version for offline and playback. Why can’t the same be done for a 35mm show? I’m sure someone working today has thought of this or, what I’m hoping for, already applied it to their workflow. Any feed back or alternatives would greatly be appreciated and if this topic has been covered somewhere else, please forgive me for not having the time to of searched for it.
Posted 02 February 2011 - 08:29 PM
Please post your findings here after deciding what to go with.
Posted 02 February 2011 - 08:35 PM
There were in the past some attempts to expose time code info onto the edge of the film in the camera. If you could find such a system and also put the same TC on the video, you'd be good to go. Otherwise, you'd have to do some very labor intensive hand tracing.
Another thing to be aware of in editing is that the finder tap recording won't be what you have on the film. Light goes to the finder when the shutter is closed, but not when it's open and exposing the film. So, in terms of time and frames, what you have in the finder is exactly what you *don't* get on the film. Your cut from the tap won't be frame accurate. It'll let you cull selects, and do a 2K scan with handles, from which you can make your fine cut.
Posted 02 February 2011 - 09:17 PM
Even if you could use ARRICODE or AATONCODE to burn the same TC onto the film and the videotap signal, there is a fundamental problem: the video tap image isn't the film image. You could be recording a blank image for all you know and still get a video tap image. The point of transferring to video or printing the negative is to SEE what actually got shot.
The video tap image is only a capture of the light that also hit the film, but it's not the image on the film. The film image could be bright green with a scratch running down the center for all you know, it could be really under or overexposed, etc.
Of course, you could have someone inspect the whole negative everyday at the lab on some sort of hazeltine or something and just hope they catch something wrong, but honestly, you want to see what you actually shot to know if it is usable.
And since a video tap image is photographed off of the groundglass, there are limits to its resolution whether or not the sensor or the recording is HD or SD. So some focus problems may not be caught if that's all you ever see before you lock picture.
So while the plan could work if you could get an EDL that matches the film without somehow actually having edited a transfer of the film, you are taking a big risk, basically editing blind and assuming that the negative is OK.
Posted 02 February 2011 - 11:21 PM
Also, as for "lugging" around a 35mm package, there are plenty of small light 35mm MOS cameras, like an Arri 2C, or 35-III or even something really cheap (which I own and love) like a Konvas 2M, or if you want best of best, the 235 from Arri. Or, for another take, why not use the Phantom you're already bringing for the FXs shots?
Posted 03 February 2011 - 01:21 PM
Yes, there are issues with synching, but the same thing applied to the early days of sound. As for the tap recordings not matching up with the film, again, this is correct. The tap footage is always off. . . by exactly 1/2 frame.
Focus issues? If you have an HD tap, and a puller and operator who are being hypervigilant, they should be able to at least say when they are unsure about focus. Then you can shoot some more safety in this situation and be OK.
THen one needs to be hypervigilant for problems with the negative itself, especially emulsion scratches, any hair that was in the gate that didn't show up checking the gate (it does happen).
All the "potential mishaps" of using tap footage are still encountered on the productions I've worked with that relied on film dailies. WHen there were issues with scratching, they have often already rapped shooting so either cut around it or go back if insurance pays for it. Ultimately, I see the biggest issue with using tap tapes as being that, if there is a problem, the insurance company can argue, whether this is correct or not, that if you had been relying on dailies from the film there would be no need for the reshoot you now need. Even if this is totally false, you have to then prove otherwise.
I don't see a problem with exposure. You can even rig the gamma of a video camera to match surprisingly well to the characteristic curve of any film negative, shoulder and heel information included. I mean, they've been doing it since 1969 now with video analyzers.
The problem would be awkward blinks, looks that occur in the half of the shooting time that the video tap doesn't see. But there it really is a crap shoot, just like with the 50% of the shooting that the negative doesn't see.
The problem is, even though all of this has been addressed and supported by several camera systems, it runs into the "Hoosiers" syndrome. "We don't like change here too much. So we'll ease 'em into it REAL slow." Not that that is what Adrian, David, and Richard are doing. I'm saying you'll run into people in labs, post-houses, editing suites, but I would say especially, especially in camera crews, who view any little change to any technique as the equivalent of you breaking one of The Ten Commandments. With some of these people, God forbid you have the courtesy tab on the magazine tape going clockwise instead of counterclockwise (seriously).
Edited by K Borowski, 03 February 2011 - 01:23 PM.
Posted 04 February 2011 - 04:06 PM
And yes, we have them available for rental, along with the Phantom and anything else you might need.
Posted 17 February 2011 - 08:55 PM
People have used an AJA Ki Pro recorder to record from an Arri video tap:
Here's a link to a demo of Arri's HD-IVS High Def video assist:
As several smart people have stated, there are some risks involved with the workflow you've suggested, but with proper testing you might be able to make it work for your project.
You'll need to make sure you can get reliable timecode from the video tap to your recording device and that the method of transferring that timecode is compatible. Having a recording device hardwired to your 35mm camera could prove to be an annoyance if you'll be doing a lot of handheld camera work.
Give Arri a call and talk to them.
If you'd consider the Aaton route, you should definitely give Mitch Gross a call. The Penelope is supposed to have a good progressive scan video tap (with timecode).
Let us know what you find out and best of luck with your project!