DI workflow

[adamgarner]

I've been thinking about a 16mm DI workflow and wanted to know if anyone could help me understand something.

When a DI is shot on film, lets say 16mm in this case, the film is developed and transferred to (probably) uncompressed 1080 or even 2k.

Here's what I'm trying to understand. The film is then edited on Final Cut, for example, and is probably compressed to some intermediate flow, right? Like ProRes422 maybe?

Then, when the edits are completed, is the DI transferred back to film? If a film were going to be distributed, or even shown at a film festival, what happens here? I'd love as much technical detail as anyone has to offer.

I seem to recall a very expensive transfer process that takes the digital and puts it back on film.

[christoph]

hi adam,

it depends a bit on the on the definition of DI (which stands for digital intermediate). in the loose sense, every films which gets scanned to a digital format and then exposed back on a print film for projection can be considered a DI workflow, but depending which way and format one chooses the workflow and quality differs significantly.

the best results are achieved if the film gets scanned to a data format rather then telecined to a video format. this means the film scanner outputs uncompressed full resolution scans, usually as 2K (or sometimes 4K with 35mm) image sequences in a 10bit RGB log format. however, because this is slow to work with and very expensive, the film often gets first transfered to a lower quality video format and when the final edit is done, the selected takes get retransfered in higher quality. this retransfered data is then the "DI" and gets color corrected and exposed back on print film.

since this usually is not an option for low budget films, the next best thing is to transfer the film to a high quality video format. with the proper gear and experienced people, this can give very good results. sometimes the full source material gets transfered in zhe beginning, but again the film is usually edited from lower quality proxies. the final edit is then recaptured from the high quality video tapes and the master tape is used for making the final film print, even though that strictly speaking it's not a "DI".

the problem is that there are several different ways depending on money, time, gear available, post house etc.. here are a two examples reflecting the above:

"proper" DI:
negative film
-> transfer to low quality proxy video with keycodes
-> edit video
-> generate keycode list from EDL
-> retransfer selected takes to 2K log and reconform to final edit
or: cut negative according to EDL and retransfer
-> color correct data
-> print back to negative/internegative stock
-> make projection prints from new negative.

HD video "DI":
negative film
-> transfer to high quality HD video format (HDCAM or HDCAM SR) using a technical grade
-> make a low quality dub with identical timecodes
-> edit low quality proxy video and edit the film
-> recapture full quality HD video from original tapes on a high end system
-> color correct video (can also be done in tape to tape)
-> play out final video to master tape
-> make a new negative from the master tape/timeline
-> make projection prints from new negative.

all this is very expensive, one can try to save some money by doing direct to harddisk transfers and doing all the necessary data management by oneself.. however, you'll need quite indepht knownledge about the whole process to avoid the pitfalls.

++ christoph

[adamgarner]

Christoph:

Thanks for the in-depth reply. I googled "DI" and found some similar info, but yours is more what I was looking for. I understand the production to transfer portion quite well, and the post production in the digital format too. It's really the "going back to film" that is a bit of a mystery. It's becoming more clear though.

So,my final question is: how is the original film synced with the EDL? In a super-16 world the film wouldn't have any sort of time code. From what I've seen of an EDL, it's like XML with timecode markers right?

Though it seems technically possible, it seems problematic.

Thanks again!

Adam.

[Rick Palidwor]

Christoph:
So,my final question is: how is the original film synced with the EDL? In a super-16 world the film wouldn't have any sort of time code. From what I've seen of an EDL, it's like XML with timecode markers right?
Adam.

15mm (and 35mm) film has an "edgecode": every foot there is a sequence of letters and numbers printed along the edge. Film manufacturer puts them there when they make the film. So not every frame has a number but there are enough edgecodes to keep track of the exact frames. Negative cutters use these when conforming a negative.

When film is transferred to tape or file for editing, but the plan is to go back and cut the negative based on editing decisions made, you can ask for something like a FLX file (there are many variations on the file format and I don't know if I spelled flx correctly). This file keeps track of the edgecodes on the film and the time codes on the video copy and can be used to translate the video edit decisions back to a negative cut list.

Google edgecodes for more info. (A friend of mine made a documentary about editing called "Edgecodes". It's good. Rent it)

Rick

[christoph]

here in germany the frame numbers that are pre-exposed on the film are called keycodes. they are in a machine readable form as well, so when you transfer the film, the telecine machine will generate a database file which keeps track which frame number corresponds to which timecode on the tape.
when you have finished your edit you generate an EDL (editing decision list) which gives the tape timecodes of the edits, and the post-house can then generate the frame numbers of the original film to cut the neg or retransfer the important takes (you can also do this yourself using cinema tools or the like, but the risk of errors is usually too big, so the post-house does it).

putting the digital data back to film is done using by exposing it back on internegative film (or camera negative film if on a budget).
there are several ways to do this, from systems that basically use a normal 35mm camera (usually a mitchell or animation camera) to film a high-resolution monitor frame by frame up to machines that use a flying spot or laser system.
google for definity, celco or arrilaser for more information.

++ c.

[Nigel]

I can say first hand that the HD DI is very nice.

Whilst it is not 2/4k it is much cheaper and resulted in an excellent print.

It would be great to have a 2/4k scan done but none of my projects have found the money. In every other case the HD-->35mm has looked nice.

Good Luck

[adamgarner]

Thanks everyone. I will let this sink in. I would love to be able to go through this process at least once so I could understand it better, but I'm not shooting 16, and don't have a budget with 5 digits. I was really just curious, but think it'd be cool to do someday.

[christoph]

hi adam,

no problem, hope it gave you some ideas..
i've had film-outs of two films that i've done the camera and post-production work for. once it turned out fairly well, the other one was screwed up by the company who did the film-out to negative.

the problem is that good transfers are rather expensive already over here in europe, even if you only want SD resolution. with HD, prices go up a lot and proper data scans are just out of question for any non-commercial project.
that's the main reason i started into building a film scanner myself, i just want proper scans without having to pay 30'000EUR for scanning a feature film.
the main problems are that you need a very high quality camera, good post-production software and an in-depth understanding of image processing and color management... and lots of time of course (i've been working at it for two years ;)
but the results are 1.4K 10bit RGB log cineon scans that look very good, even with negative film.
++ christoph