S16mm 2k workflow/VFX quality control

[Liam Kavanagh]

Hello,

 

The project we are producing is an 18minute short film photographed in super 16mm on Kodak?s 7219 stock.

Savings on cost is a large factor in this production so my basic assumption for the workflow consists of the following;

-Scanning the processed negative via a Quadra telecine to a DVCAM format for an offline picture edit on adobe premiere cs3.

-Export EDL and return to the lab to have our 18-20 min decision scanned in high definition for online assembly and color compositing in After Effects CS3.

Besides the color corrections that the picture will go through ? we have about a 30 second segment that will require heavy compositing (roto-scoping, keying and 3d modeling).

Colleagues have suggested photographing the FX sequences in super35mm to have higher res footage to work with in the compositing suites ? yet others have recommended having the 16mm prints scanned at 2k to have 2048x1556 .dpx files created on an external HDD (the 35mm print would go through the same process) and should be sufficient in terms of image info for a compositor to work with. The final output will be 1080p mastered on D5 with HDCAM being the media that will be sent out to the festival circuit. 16mm originating footage doesn?t compare to 35mm in terms of resolution quality, of course, but for a project that isn?t going to finish on film... (yet will be digitally project at the festivals), I'm debating between the two choices.

Costs savings is an important issue ? yet quality fortunately comes first as the VFX segments are a crucial point to the plot.

What advice would you have concerning these choices for our VFX segments and the workflow in the 2k medium for a 16mm originated project? And if you see issues with the described workflow, or have worked with similar workflows in the past ? please share your advice as this is my first attempt using these methods.

Thanks for reading and I apologize if perhaps this was covered in another thread.

Have a great day!

Liam

[Guest tylerhawes]

We recently did a film shot on Super 16 and the workflow was Spirit 2K > HDCAM-SR 4:4:4 log > DPX finish. We also performed over 400 visual effects on the film.

 

Before doing the FX, we did a test scanning on Northlight 2 at Super 3K (6K scan scaled to 3K) 16-bit TIFF to see if it would improve the FX process. We found that the extra resolution was wasted and didn't help the shots. In fact, the director preferred the Spirit path because it was a little flattering toward the grain, while all the scan did was help it stand out more (that's personal preference, I didn't mind the sharpness of the grain myself). The only real improvement was the stability of the pin-registered scan, however the Spirit is so smooth it very rarely was an issue for FX that we were doing, esp. since there was a little wavering in-camera anyway.

 

The HDCAM-SR format was perfect for what we were doing and is what I would recommend for most S16 films. There is plenty of resolution and 4:4:4 log is virtually the same quality as DPX. I only recommend making sure you work with a DI house that is actually setup to do that right, because there's a lot of telecines out there that will output RGB but have actually dithered it down to 4:2:2 video somewhere in the path, negating the benefit. Also not everyone can do a log transfer, and there are ways you can adjust the log curve to a custom curve that will actually get a lot more precision out of the transfer, but it's not off-the-shelf so not every company does it.

 

For your final output I would also recommend considering HDCAM-SR instead of D5, as the 4:4:4 master is important and SR is a lot easier to manage than all the data files.

 

If this were my film, I'd try to just Spirit 2K *everything* to HDCAM-SR 4:4:4, apply a log/lin LUT and make dailies files for editor, edit, then reconform from SR tape, color & finish. You only transfer once, you get everything you could want at high quality, and you can always make a change later at much less incremental cost. The conform is a very predictable and low amount of time. If you telecine everything, you can save money as long as you have a reasonably low shoot ratio because the TK operator doesn't need to stop and start, he just lets it roll.

 

If you can afford to shoot the FX sequences on 35mm and you anticipate a lot of cropping, I'd do it. Also, if you really want the look of 35mm in terms of grain structure, there are some new emulsion-specific motion analytical grain removal processes that can be pretty amazing at stripping away the grain and we've had great success with those. Of course it would be best to just shoot 35mm, and somebody else will say it if i don't, so I'll also mention RED.

[Chris Burke]

I will second what Tyler has already said, transfer all your flats to HDCAM-SR as a technical grade, which is very broad grade that gives you everything on the negative. Time in the tk suite is what cost big money, so the less time there the better. It will cost you more to do the traditional one light for editorial and then rescanning of selects, in most cases. I should add that deals can and often are had with the lab you are working with, so ask around. I did the later workflow and while I will say I don't have any regrets, I now see how much more efficient and economical it is to do the scan once to SR route. Your negative is precious, the less you touch it the better. This work flow seems to be the best for Super 16, especially with a short film such as yours. If the SR deck rental is also too much, see if the added cost of going to hard drive is cheaper than the deck rental. If you do the hard drive dump, then working with some sort of offline codec is best. In the Final Cut Pro world, ProRes HQ would be recommended. In Adobe land I am a foreigner, but I am sure there is an equivalent. I would also recommend the SR finish over D5, with it's added color resolution, it makes for a better master that can even be printed to film with amazing results. :rolleyes:

[John Brawley]

Hello,

 

 

Colleagues have suggested photographing the FX sequences in super35mm to have higher res footage to work with in the compositing suites ? yet others have recommended having the 16mm prints scanned at 2k to have 2048x1556 .dpx files created on an external HDD (the 35mm print would go through the same process) and should be sufficient in terms of image info for a compositor to work with.

 

 

Hi liam.

 

The main issue is actually that many super 16 cameras aren't very steady for VFX work. Image instability makes it much more difficult to roto and work with images, no matter how high the scanning resolution.

 

There are better super 16 cameras to use for VFX work than others.

 

In my own tests recently using an arriscanner, I found there to be little difference to images scanned at 2K and those transferred to HDACAM SR. (when shooting super 16)

 

The main cost with 35mm is stock and processing. For VFX, most of the shots tend to be short so it's usually not going to cost a WHOLE lot more to shoot 35 just for VFX.

 

I would recommend it if you can as you will definitely see the difference.

 

If you have to shoot 16 then use one of the more modern and more stable cameras. Most of the Aaton's are good, or an Arri SR3 ADVANCE or a 416 are good too.

 

If it were me, I'd probably consider shooting RED for VFX. No stability issues and very clean images for kes and comps (provided it's shot correctly.)

 

I love super 16 a lot, but VFX is not is strong point...

 

jb

[Scott Fritzshall]

S16 is painful for VFX work, especially when you've got to do keying, and especially when you're shooting on fast stock. I would strongly recommend shooting the VFX shots on 35mm or some other high resolution, low compression format. You can always degrade high quality footage to make it look more like a smaller format. I actually did this for a film a little while back- most of the film was shot in S16, but the VFX sequences were done on 35mm and scanned at 2k. I am a really big fan of Super16, but I hope I never have to do compositing on it.