Sunday, 25 February, 2007

Shoot HDV release in a theatre

Some months ago we completed the entire digital post-production for a film called 'Shoonya' (literally meaning zero). This movie has been shown at festivals overeas, Rotterdam for one.

What made Shoonya different and probably unique is that it was shot not on film, but tape, Sony HDV tape. And using a Sony Z1 camcorder too. on small DV sized tapes.

When we first saw the footage on tape we were sceptical about how it would look. Particularly considering that the producers wanted to grade it and finally record it out to 35 mm film to show in theatres. Very challenging. Fortunately we were involved right from the beginning so we could specify a correct workflow.

I first tried out native HDV editing, which was then possible in FCP 5 (the producers chose to edit with FCP - Final Cut Pro). But the 'conform' process discouraged me as it was way too slow. For a 2 hr plus film it would be a drudgery. So I advised working with an 'intermediate codec' - Apple Intermediate.

Working with AIC assured discrete frames instead of B and P frames but it had two down sides. One, time code was not preserved after capture. And two, AIC takes up much more space than HDV-native's 12 Gb/hour.

Yet, we eventually went with AIC. We did small trials and viewed them in theatres to be convinced about the quality. The final result, though not exactly like film was definitely better than shooting SD (DV/Beta/DBeta)

After editing was complete, we collected the edit on a disk. We broke it into reels corresponding to 2000 feet in 35mm film. The maximum length most cine projectors can show at a stretch. Remember they shot 1080i50, that is 25 fps.

After that, I applied a 'secret sauce' to get it into a Lustre grading system. I created DPX file sequences from their media since Lustre is happiest working with DPX. I experimented with 8bit and 10bit and came o the conclusion that since the acquisition process worked at 8bit, there was little to gain witn 10bit.

Grading in Lustre was a bit challenging since the material lacked the latitude that 35mm film has. Here's where on-set monitoring plays a crucial role. Because if your on-set monitoring is correct in black and white levels, you'd set your exposure just right and so any exaggerated correction would be unnecessary.

If you plan on shooting HD in a big way, invest in an HD monitor of your own. Worth it.

We found HDV originated footage is less tolerant to exposure correction than 35mm originated film. Shoonya was shot pretty precisely so this kind of correction wasn't necessary.

So in the final analysis, is it really a good idea to not shoot film and yet aspire to have film as the final exhibition format?

No, I'm not about to deliver a final judgment on this. I'll just put forth some issues to keep in mind.

Not shooting film can (or could) definitely save money.
Shooting 'prosumer' HD like HDV is better than shooting SD - Beta/DBeta/DV. HDV at least has the resolution.

Shooting 'advanced' HD like HDCamSR/Varicam/Genesis can actually be more complex than shooting film. And these may actually even be costlier than film. Yes!
Avoid HDV camcorders with average lenses. The Canon XL-H1 and JVC HD250 producer much better mages than Sony's Z1 and such like just on account of their lenses.
The Z1's lens 'performance' has to be seen to be believed.

The Red camera. Keep a close watch on this.
Sony HDCamSR over Panasonic DVCProHD. Varicam shoots on highly compressed DVCProHD so its half-way between the highly compressed HDV and the relatively uncompressed HDCam-SR. HDCamSR rules. Period.

HD tapes (except HDV and DVCProHD) arent's exactly cheap. How's 25 grand for 90 mins?
For post noone's giving you an HD recorder at less than 7k and hour. Yes you'll need to factor that for editing.
Shooting film means processing delays. Shooting HDCamSR or DVCProHD also means transfer delays. You need to transfer to DV so you can edit at your place.

If you want a 'film look', currently only film will give it to you.
Not one single HD format looks exactly like film. But that's missing the point completely. Do a trial and convince yourself that the 'not-looking-like-film' really affects your story. For some stories, it might just not make a difference.

Most editors have not equipped themselves with the finer nuances of technology that HD working necessitates.
A final post-production workflow has to be worked out before you begin post production. Do yourself a favour and please hire a consultant just for this. It can save you a lot of time and money. And keep you sane.

Did I mention that Shoonya was shot with sync sound mostly. And it was shot as 1080i50 meaning 25 fps. Yet, the theatrical film version is 24 fps 35mm film. And the TV version is tape at 25fps. Both same length. And sound gave no trouble.

Now you know why you need a consultant.

Saturday, 17 February, 2007

The 24-25 issue in PAL film editing

There is this '24-25 thing' that crops up every now and then, in film post in PAL countries, especially India.

I was at the Institute last weekend. The Institute in the Indian film industry refers to the Film and Television Institute of India. Its at Pune and many of India's finest Film and TV personnel and artistes passed out of there. Along with my partner Anand Subaya we did a short workshop on editing.

One of the things that came up with the students one evening was this '24-25 thing'. The students there have some rather interesting issues, some I'd never thought of.

Anyway, this '24-25 thing' comes about because...
1. Film is shot at 24 fps.
2. Sound is recorded separately on Nagra/DAT/disk
3. Film is edited on a computer.
4. But it needs to be transferred to a tape to be transferred a computer.
5. Tape runs at 25 fps in PAL countries.

So...
6. Film is transferred to tape at 25 fps.
7. Because an accurate relationship has to be set up between film and tape.
8. So that in the end a cut list can be made to cut the negative accurately.
9. This transfer from film to tape is at 25 fps
10. That speeds up the film. (by 4%)
11. But editing systems need to run (play back) at 24 fps to stay in sync with sound.

So where's the problem?

As long as you're editing the film, and making cut lists, there is no problem. It works and we've been doing it for over a decade now.

But you need to pass on the edited picture as video on a tape for sound track laying and mixing. Since in Avid or FCP (computer-based) editing, you don't cut the negative till after the sound is done.

This transfer from computer to tape needs to be at the right frame rate so that after mixing and recording to optical neg, the optical neg matches the cut picture neg.

So that's the problem. Most people, especially editors from the video world simply don't handle this correctly.

And to make it worse, no system except telecines, Avid Film composers (and new Media Composers), and FCP systems - only these systems are capable of playing a 24 fps video source to a 25 fps video monitor or tape recorder (Beta/Digi/DV whatever) while keeping the duration intact.

Smoke, Quantel eQ, etc. nothing at all can do this.

What's the problem again?

1. Film is shot at 24 fps. So 10 sec of time is recorded to 240 frames on film.
2. Film is transferred to tape at 25 fps. So all 240 frames of film are transferred to 240 frames on tape. But since tape runs at 25 fps, these 240 frames run in 240/25 that is 9.6 sec.
3. Avid and FCP systems capture this at 25 fps. But they convert captured footage to the frame rate of 24 fps. So the original 240 frames are transferred to Avid/FCP.
4. And these (Avid/FCP) machines play them back 24 per sec. So 240 frames now play out in 10 sec.
5. That's how they sync with sound that was recorded.
5. But if after capture and editing, you play them back on an Avid or FCP and record that to tape, you'll get a 10 sec shot on tape which will be 250 frames.
6. So 10 sec on a timeline is 10 sec on tape. 10 sec on video tape is 250 frames.

So where do these extra 250-240=10 frames come from?

Avid and FCP systems repeat one frame every second so they add 1 frame each second. 10 secs = 10 frames. So 240 frames become 250 frames.

As an aside, the problem that students at FTII have is more complex.

They shoot at 24 fps. Process and print. Then cut these rush prints on a flatbed or Steenbeck. Steenbeck runs picture and sound (on mag film or sepmag) together at 24 fps.

But - and this is interesting - budding film-makers there shoot this edited rush film off the Steenbeck screen to a Handycam. They then capture these Handycam DV tapes to a sound software to do sound sweetening, over-dubbing, and other sound post.

And - this is serious stuff - their sound software outputs a wav or aiff file that doesn't sync with the visual. On the face of it, this seems more likely to be an issue of the Steenbeck not playing back at a tight 24 fps rather than a '24-25 thing'.

Anyway, this FTII 24-25 Steenbeck telecine issue is just an aside.

Even seasoned film-makers don't have much of a clue about this 24-25 thing. At a trial screening for a forthcoming feature someone noticed a small part of a scene being 'out-sync' At the interval, he promptly asked me if it was the '24-25 thing'

One thing to get very clearly. The '24-25 thing' invariably creates a mis-match between video and audio which is perfectly 1 frame per sec. So after 5 secs you're 5 frames out-sync. After 20 secs you're 20 frames out-sync. And if a 2000 feet reel is mis-matched because someone messed up the '24-25 thing' you'll be out-sync by a whopping 1 minute! Meaning at the end of the reel, picture will end a whole minute after sound or vice-versa.

So what's the best solution to this '24-25 thing'?

Things to remember

1. Understand how each stage of the film was handled.
2. Convert from 25 to 24 and vice versa if necessary (only possible in FCP)
3. Never capture an Avid or FCP dump back for editing.
4. Prefer handing out Quicktime movies rather than tape dumps for sound post.
5. Sound has no fps. It has a duration in seconds only. (Read this twice)
6. EDLs have no fps. They generate timelines of the fps of the source fps. (Read this thrice)

And if all this fails and you're completely stuck with a picture and sound that doesn't match, then I'm available for consulting. I've figured that so much time and energy has been spent on this '24-25 thing' that there's some commercial value in helping out. So write me and we can talk.

Else figure out who's gonna pay for the stock and studio time if you mess up.

I'm still available.