Struggling with digital movie-making

For the past few months I've been working on setting up and executing a simple and efficient workflow for the post of a delightful little movie called 'Stanley ka Dabba'. This movie was shot entirely on a Canon 7D DSLR camera. I've posted many a post of still cameras that shoot videos right here... 

The Canon 7D DSLR camera shoots beautiful and natural stills. And the same sensor shoots HD video. These are saved on CF cards. As Quicktime movies encoded H.264 with PCM audio.

But there are two major problems. One is, the Canon 7D shoots 23.976 fps, not true 24 fps that film in theaters runs at. And the other is that the files have no timecode (TC in short) or reel/source associated with them. HD movie files out of a Canon are called MVI_xxxx where xxxx is a number serially added by the camera.

So what happens is that after one finishes the edit, and needs to send it for further post, like colour correction, VFX etc, there is no EDL based route to sending the edit data onward.

For 'Stanley ka Dabba', I had devised a workflow where I converted the files from 23.98 to 24fps without any frame loss or motion conversion. I then striped all the files with timecode with one following the next, sort of like they were captured off tape. And I assigned a reel name or source for all the files as per the folder they resided in. 

Deepa Bhatia, the editor and co-producer of 'Stanley ka Dabba' edited on an Avid Media Composer, I made ALE files for her assistants to import into a bin and then batch imported the original (striped with TC) files from the Canon. That way Avid also 'gets' the TC from the Canon files.

After Deepa's edit, the colour correction happened at Reliance MediaWorks. On the Baselight at Reliance, the Avid EDLs that Deepa made out, perfectly conformed the original Canon files and they were on their way to colour grading within minutes. There was no struggle whatsoever.

So what's with the title of this post? 'Struggling with digital movie-making'. Here's what...

On the same day as 'Stanley ka Dabba' was released, there was another movie released, a horror movie about text messages. This too was shot (partly or entirely) on a Canon DSLR. But they chose to do their post in a different way. They directly imported the Canon files into Avid Media Composer to edit them. No 24fps, no TC.

Long story short, that movie had to be re-assembled manually shot by shot, eye-matched on an Autodesk Smoke system. The post facility concerned, simply couldn't fathom a correct method for reproducing the original edit.

Similarly, there are other post houses which have yet another interesting workflow for Canon files. They advocate (or so I am told by clients who've been there) first transferring all the Canon 7D files to HDCam tape (eeks 3:1:1 colour sampling) or HDCamSR tape. 

In a period of nearly extinct and horrendously expensive HDCamSR tapes (after the Japan tsunami-earthquake), this isn't a rather bright suggestion.

I'm getting a bit snarky and sarcastic about all this because, as an editor it bothers me no end that people with no desire to find out new solutions, have such a say in the way post-production is (man)handled.

There are also other struggles and horror stories - with Alexa files, bad colour space conversions, loss of timecode, Red R3D file conversions that take days to get done, Weisscam files that just don't open and many such 'Adventures of a digital rebel'.

In my opinion 'Stanley ka Dabba showed that if one has a story with a heart, a positive attitude, and the ability to take a calculated risk in doing something for the first time, then there is a medium that's waiting to help you tell your story. There are challenges for sure, and shooting Canon is not exactly simple.

So just go to a theatre near you where 'Stanley' is showing, and take a look for yourself. Cinematographers and even DOPs (are they different?) will of course find issues with latitude, resolution, gamma, etc etc. 

But you know what. With some 'careful' lighting it is entirely possible to make even 35mm film look flat. And some 'adventurous' DI colour correction can easily make film look synthetic and 'inorganic' too.

About operations on the Canon 7D, not being able to pull focus or aperture, no optical viewfinder (for video), and other whines from movie film cinematographers. I got an interesting insight at a fashion show a few weeks ago at the Leela. 

At the head of the ramp during the show, there were photographers taking stills on their DSLRs. How do they focus, without a tape measure, with the artist walking toward them with no 'mark' to stop on, no rehearsal 'for movement'. And stills photographers have to take a frozen moment, so going out of focus is also out of the question. 
And with the light at fashion shows they are often shooting at wide apertures, so depth of field is shallow.

Bottom line. Digital 'movie-making' is on it way to replacing the term 'film-making'. These are early days, of course. Those who forge ahead are the positive ones with guts. The rest will whine and sigh and hang on to film with a naive hope that digital is a fad that will go away and one day we will all go back to film. 

As far as post production is concerned - editing, colour correction, sound editing and mixing - most of this is digital for a decade or more. Doing post on digital is no more complex than post for film - managing logs, key codes, cut lists, pull lists, making cement joints, Hamman joints, winding and rewinding film on a winder - I'm glad we got rid of all that.

 

 

Movieola, Steenbeck, Hamman splicer... digital is vastly simpler than working on these

Red One, Red MX, Epic, Alexa, Canon 5D/7D, AVCHD, XDCamEx, HDCamSR, Viper, D-21, Genesis, ArriRAW, Canon MPEG2 422, AVC-Intra, P2, XDCamHD, Varicam, DVCProHD, KiPro, Gemini, Hyperdeck, Nanoflash, GoPro... the list goes on...

There are a large number of resources and personal experiences for each of these. Forums, blogs, web zines are oozing with information. And simple common sense, along with some basic knowledge of how computers work and store data, also helps. For my part, I've worked hard to be ready for each and every one of these new digital formats. 

Store, Copy, Convert, Edit, colour grade, finish, backup, archive, re-purpose into new formats... the whole nine yards. 
So digital doesn't have to be struggle, unless you try hard, do the wrong things, or go to the wrong places.

Editing HDCamSR 24p in an Avid Media Composer Mojo

A colleague came up with an interesting situation. 

A client of ours is proposing to shoot a film on HDCamSR at 1080p24, meaning real 24 fps. The final output will be 35mm film also real 24fps. Being from a PAL country we see no utility on doing the 23.976 - 29.97 NTSC workflow. 

In fact, when I researched this on the 'net and specifically the Avid Community forum, I came up with mostly NTSC based workflows. Even the Avid MC manual doesn't have any explanation on this. Not surprising, editing system manuals seldom address the PAL 24-25 workflow. 

I've found that in the non-PAL post world, people mostly work in 23.976 fps. For shooting HD, telecine to tape whatever. No 24fps. Or in variants of 23.98 like 59.94. Incidentally, on FCP this would be easy. All our FCP systems have Blackmagic HD capable cards. And in FCP one can capture the HD as ProRes or DVCProHD and save space. 

Either way we would be capturing 24fps and editing true 24fps HD. We can then make out an EDL and recapture selects for conform and grading in our normal film grading systems - like Quantel iQ or Resolve. 

Even Autodesk Smoke or a Quantel eQ could be used for the recapture. We then export as DPX for further processing. All of this is at 24fps. Tried and tested. Works. But on an Avid there is a problem. 

All our Avids are Mojo SDI Avids that cannot capture HD. So we need to output from the HDCamSR VTR as SD PAL video. And an HDCamSR VTR, when outputting an HDCamSR tape as SD PAL video does so with a speed change. Meaning 24p HD plays out, not at 24fps, but at 25fps. 4% faster. 

Of course, Avid can capture 25fps PAL material in a film project and make it play at 24fps. But in this process of making it 25fps in the HD VTR and converting to 24fps in the Avid one loses the original HD timecode. So no EDLs, no recaptures. 

So I did some trials to find a fix. 

I've tried on a small test clip and sequence, so please do try it out and check if it works. Here it is. 

First set the HD VTR to do a burn-in TC (also called window burn or TC super). Make it show the original 24fps TC and the 'derived' 25fps TC. It looks like this.

 

 TCR is the 25fps 'derived' timecode that Avid will get when capturing, and 'ORG' is the original 24fps timecode. Next, open a film project in Avid Media Composer (this applies to MC MojoSDI ver 2.6 upwards). Capture the HD with deck control normally as you would. After capture set the bin headings like this. Basically you need to be able to see 'Start' and 'TC 24'. Initially, the TC24 column will be blank in the bin. Now open each clip from the bin to the source window and park at the first frame. In the bin enter the timecode you see in your rushes as 'ORG' in the 'TC 24' column. 

So now, you've 'assigned' a 24fps timecode to each captured clip. Avid will now track both the 'derived' 25fps timecode as well as the original 24fps timecode through your edits. Now you can edit normally. In the 'Film and 24p' settings you can set the edit play rate as 24fps so that the sequence clips will sync with audio if any was recorded separately on location. 

After the edit is complete, you can export the EDL using EDL manager. But with two small changes. In the EDL manager main window (not settings window), set the 'Source TC' to 24 and 'Record TC' to 24. Now open the seq you've edited and make an EDL. 

And save it. If your finishing system needs a reference cut, export the sequence as a Quicktime movie. Make sure the Quicktime movie is 24fps. Mind that Avid , when doing a fast export from a 24fps timeline, makes a movie as 23.98 fps. If you have an FCP system handy you can convert this movie to DV-PAL and then use CinemaTools to conform it to 24fps. Else just do a custom export from avid as DV-PAL at 24fps. 

Quantel iQ and the new Smoke and Resolve accept these as reference movies. This reference movie will match clips recaptured from the original HDCamSR tapes at 24fps using the EDL we just made. 

Please mind that I've tried this for a few clips and an EDL. You should test this thoroughly with your source material before committing to this workflow. And as I said before, if you have the option to work in an FCP system, none of this TC wrangling is necessary. 

Or, if you haven't got a Avid system yet and have to work in Avid, then get a MojoDX. This allows HD capture. And no TC wrangling. If this method has saved you some labour or made you some money, then consider pressing that yellow button in the margin that says 'Donate'.

I'm shooting HD

I keep hearing this at weekly and even twice, thrice a week now. "I'm shooting my next film on HD". "How much difference would it make if I shot on HD?". "I've read that HD cannot be made to look like film" Or, "I've read that HD looks as good as film".

I'm not about to answer or refute any of these. What I'd like to put forth simply is that HD is not a single monolithic format like 35mm film is. So if you used the phrase "...shooting on HD..." in a sentence, you could be shooting on...
HDCam, HDCamSR, HDV, DVCProHD, Varicam, CineAlta, P2, XDCam, XDCamHD now.

Further in many of these you could be shooting as...
720p24, 720p25, 720p60, 1080p24, 1080p25, 1080p30, 1080i50, 1080i59.94 etc etc.

So what's your flavour of HD?

A quick run-down on what the figures mean.
Normal or SD TV meaning what you shoot on DV, DigiBeta, play back on DVD, watch TV on, etc in India is...
Frame size 720x576 at 25 fps. So this will be called 576i50.

In HD the numbers are
720p24 meaning frame size 1280x720 pixels at 24 fps progressive (no fields)
720p25 meaning frame size 1280x720 pixels at 25 fps progressive (no fields)
1080i50 meaning frame size 1920x1080 pixels at 25 fps interlaced
1080p24 meaning frame size 1920x1080 pixels at 24 fps progressive (no fields)
1080p25 meaning frame size 1920x1080 pixels at 25 fps progressive (no fields)
 
These are the numbers we would normally encounter in India.
If what you shoot will be used for TV, then do 'i'. For film outputs or mixing with film do 'p'

Outside India there would be other formats like 720p30, 1080i60, etc.

What about shooting formats?

If you're in India, in Mumbai, here's the deal.

If the camera is large and Panasonic and takes prime lenses you most likely are using a Varicam. So you're shooting on small DVCProHD tapes at 720p something. Frame size is 1280x720 pixels and compression is 12:1. In each frame 8% data is kept, 92% thrown away.
If you shoot on a small hand-held Sony camcorder you're shooting HDV. Frame size is 1920x1080 and compression is MPEG-2 50:1 Yes that's 50:1. In each frame 2% data is kept, 98% thrown away. Camera would be HVR-Z1
If you shoot on a small hand-held Panasonic camcorder and on small flash cards not tape, you're shooting P2. Frame size is 1920x1080 or 1280x720 pixels and compression is 12:1. In each frame 8% data is kept, 92% thrown away. Camera would be HVX 200.
If the camera is large and Sony and takes prime lenses you most likely are using a CineAlta. So you're shooting on DigiBeta sized HDCam or HDCamSR tapes at 1080p something. Frame size is 1920x1080 pixels and compression is 5:1 or 3:1. In each frame 20% to 33% data is kept, 80% to 66% thrown away.

If the camera is large and Sony and shoots on disks you most likely are using a XDCamHD. So you're shooting on Blu-Ray disks 1080p something. Frame size is 1920x1080 pixels and compression is Mpeg-2 at about 20:1 to 30:1 depending on format. So, in each frame 5% data is kept, 95% thrown away.

As you can see there's no real uncompressed HD format. All use some or other kind of compression. But they are all very good compressors. Really can't see defects unless you see the image up close.

Cost. Just find out what tape costs. And then what it costs to hire the camera. Now do the math. Is film costlier by a huge margin? You'll be surprised by the figures.

And does it/can it look like film? Very subjective. Very. My take on this is "Why bother?" Just go ahead and shoot for some other reason. And check if the image you've shot goes with the story. If it does, then like I said "Why bother?"