Editing / Cutting the film

#7

Another question, workflow. When we (the editors) come into the picture I’m invisioning this:

  1. Rendered frames are generated. (Each frame one image)
  2. Frames are stored somewhere so we can retrieve them.
  3. After retrieving frames we cut and edit the film.
  4. Finished film is output to the various formats.
  5. We deliver the various outputs as is appropriate. (DVD, Mov)

First, does this sound accurate. Frame rendering is the way I’m used to working on my own and customers projects.

Second, who is rendering, where will they store the finished frames, and then the other questions from before.

#8

The frame format is important too. JPG, TIFF etc. I’ve done animations with JPG and TIFF. TIFF is better quality no compression, large files. JPG is lesser quality, high compression, smaller size. Why does it matter. The storage location.

Lets run some numbers, on a 5 min short only.
30 frames/sec
60 sec/min
5 min
9000 fames total

1440 x 972/frame
4MB/Tiff
~40kB-400kB/Jpg (400 kB is probably high since most of the images will be very basic)

Total 5 min. short
36GB Tiff
~360MB-3.6GB Jpg

I think Jpg is the way to go for this one since we will be working over the web. I’ve got enough space but someone needs to get the files to me. We need web space storage for that. Can CGTalk donate space temporarily till we download images? Leonard would need to answer that I guess.

I’ll update the first message with these questions.

#9

Folks, we have FTP space reserved for this project. Don’t worry about finding space.
When we get to the rendering phase, FTP info will be shared with those who need to know.

#10

For the final rendered size, I think it’d be best to go with something around 1920x1040 resolution.
That’ll safely cover all bases - from DVD to Film - for possible festival screenings as well.

Even though thats a pretty big frame size, it really shouldn’t take too long to render since the geometry/textures are very simple. If the Lighting TD’s do a good job optimizing the scene lighting/rendering wise, and we divide up the render jobs across as many team members computers as possible, I think it can be an ideal / realistic goal.

#11

I’m cool with that, the bigger the better. Render time would be the only issue then.

Help me out with that res. What does that compute/equate to? Film res, PAL x3 :D.

#12

Whyatt’s suggestions sound good to me. I think that working from a non-compressed rendered file such as .tiff would be better (space permitted of course). I have found that saving any compression until the final export from the editing software yeilds the highest quality final product.

Another thing to think about is:

Are we going to be doing render layer passes?
If so then an image format that includes alpha is required such as Tiff.

I like the 1.85:1 aspect ratio. I feel that using a widescreen format is much cooler and more professional.

As far as final output:

I believe that by having the most options for viewing of this film would yeild the most amount of people that view it.

DVD to be ordered by people (Free to people that worked on it).
downloadable .mov (sorenson 3 codec)
downloadable .avi (divx codec)

My setup consists of:

After Effects 6
Final Cut Pro
DVD studio Pro
DVD Burner,

my dad has a printer that can print directly on the CD/DVD instead of on a sticker. Not sure if we want to go that far, but its cool non the less.

Its good to talk all this through and get a solid plan in place before production actually begins. Keep it up all

-Brad

#13

As far as rendering machines go,

I have access to up to 9 machines for rendering. Including 3 dual 2GHZ G5’s.

6 of the machines are farmed together at my work and the other 3 are my personal computers. All running Maya 6.0.1 Just thought I would through that out there.

So I am available for assistance in rendering as well as compositing and editing.

-Brad

#14

I agree.

Good point. Any postprocessing we do will require that. We could include a background plate. One layer for orange and then a separate layer for all the greys. That way we’ll be able to do easy depth of field effects and such. The scenery is 2 square rooms. I can’t imagine we’d need anything fancy for that and the storyboard doesn’t call for any special effects that I see.

I think it’s official then barring anything unforseen.

Kirt mentioned that Ballistic will get us anything we need. We’ll have to see what that means in detail.

I’ve only edited stuff alone. Working in a team like this should be interesting. We’ll just have to be very structured as to who does what task and coordinate with the director.

Everything sounds good so far.

#15

yeah, I havent ever edited in a group before either, Its sure gonna be interesting, but good experience non-the-less.

I like your idea of seperating out the greys and oranges for use in emphasis and depth of field.

sounds good everyone.

-Brad

#16

As for layers required for editing I imagine we will need:

Backround difuse
background shadow
background specular
backgrond 3 point light
background ambient occlusion or gi shadow pass.

orange diffuse
orange difuse
orange shadow
orange specular
orange 3 point light

greycubes diffuse
greycubes difuse
greycubes shadow
greycubes specular
greycubes 3 point light

overall zdepth

Correct me if im wrong but I think that will give the most amount of control during editing.

#17

I’d recommend composites flowing into editing as Quicktime files using the Pixlet codec (which Pixar devleoped for this very task, and is built-in to Quicktime 6 and higher). Should be single frame RGBA tiffs up until that point though.

I’ve edited quite a lot in groups (tons of commercials, animatics, shorts) and I’d suggest setting up a pipeline like the following:

  • Raw gray/orange/background plates composited together with depth of field and color correction.
  • Raw composites from multiple compositors are approved (or kicked back) by a single individual responsible for consistency.
  • Composites go downstream to get added to the “work print” – replacing shot by shot in the animatic. A timecode window should be added to each work print preview that is posted so that we’re all speaking the same language when it comes to direction – ie “cut more quickly from scene 22 at 4:45.24” etc.
  • Work print gets “tightened up” by a single editor with director’s feedback to get final timing.
  • Work print gets sent to music/foley artists as final.
  • Rework on specific shots get ordered on this point. May go all the way back to animation stage.
  • Begin work on titles and credits.
  • A single editor, with the director, adds reworked scenes, music, foley and titles to the work print.
  • Rework ordered, if necessary.
  • Credits and titles finalized, timecode window removed from work print.

I’d suggest that the final proper edit remain as Final Cut Pro files, but compositors and people working on titles use After Effects or Motion.

Also, I’ve got a ton of G5s to use for rendering as well.

#18

How much time do you recommend I assign for the post production/editing period?

#19

Trey, sounds like you’ve got a pretty good grasp on how we should work this. What is final cut compatitble with, Premier? That would be my only problem.

Theresa, depends on our communication. The faster we can get input from Kirt and each other will determine it. Anywhere from a week and up. When is the actual drop dead date again?

BTW, my system drive just died. I’m fixing it now but I’m down until furthur notice. I’ll likely have it up after the weekend.

#20

Theresa,

I don’t think it would take more than a week or two to get the plates composited and a rough cut put together.

The people doing music and foley can’t really do their thing completely until the cut has been tightened up and locked down completely in terms of timing, though – and the people doing titles and credits can’t finish until the very very last stage before final editing polish and mastering.

The big gotcha for the keepers of the schedules on this is rework. It’s inevitable that some pieces will need a little more punch, or in context the lighting or timing will need to be tweaked, etc. Rework will have to be done, and it may have to go all the way back through the animation, lighting and rendering crews again and checked again for consistency before being reinserted into the work print.

Don’t forget rendering time, either. The piece is fairly simple, but at high def res with global illumination, it’ll still take a non-trivial amount of time (we can get a pretty good estimate on rendering time once lighting tests happen). Not to mention that if we’re cranking out a minimum of 15 plates & passes at high def uncompressed resolution to a server, just uploading the renders might take a few days.

We’re still shooting to have a finished product before siggraph, right? In that case, I’d work backwards from July 20 in setting dates for major post milestones and give as much room as you can for rework and polish.

#21

Wyatt,

Final Cut is compatible with Premiere via either the Automatic Duck plug-ins or just by using edit decision lists (EDLs).

Not sure it’s that big of a deal though, as long as each editor and/or compositor delivers their “chunk” as the exact same format/size/codec Quicktime via whatever software they use. Need one person to review each submitted chunk for consistency, no compression artifacts, etc before the “master” editor gets their hands on them.

This means that reworked sections will have to go back to their specific editor.

I have the feeling that after compositing (with whatever software) and then dividing up scene chunks and editing them collaboratively, one person (using whatever software) will take the final edited chunks and put them together for polish, tightening, addition of titles, sound and music, etc. From that point, it’ll be that one person and the director working on the final edit simply because a) the files will be too damned massive to move around easily at that point and b) we’re shooting for a single voice and vision, which are being distilled through the final editor and director at this point.

#22

By the way, the final piece should be uncompressed, Animation codec Quicktime at full resolution, 100% quality with 48k sound. We can retarget to any source we’d like from here without compression artifacts.

It will be a HUGE file, but uncompressed is the only way to go for the master.

If we’re thinking of doing festival circuits, we might consider creating a second master with a surround mix (environment echoes n stuff) after Siggraph is over with.

#23

Hi Tharrel

I agree with all of your comments it seems like an ideal way to go but I think there will have to be some compromises for this short due mainly to file size. ( and how much server space we can get hold of…)

We might have to except that the frames have some compression on them just to ease the upload need.

But I guess we will need to get a consensus from the group and do a couple of test shots and work out the math…

#24

I think that simply zipping the chunks of rendered frames would have a large effect on reducing the amount of time & space we’re talking about.

After compositing, the pixlet codec is pretty good at lossless compression of large, fairly flat shaded stuff.

Don’t think there’s any way around the uncompressed bit at the end, though – not if we want to maintain enough quality to retarget to HD or festivals without artifacting.

I’ve heard great stuff about the H.264 codec – which is built for HD – in some cases close to 75:1 compression ratios. It’s not scrubbable, though, and requires serious horsepower for playback, so I don’t think it would be very usable for composites.

Anyhow, you’re right about doing the math once we start we’ve got some approved lighting test passes. The collaborative stuff I work on, we simply drive a few CDs or files on an iPod across town if they’re too unwieldy to FTP… come to think of it, shipping a couple of CDs wouldn’t be terribly expensive, and it’s an option for those without a broadband connection…

–T

#25

No worries I’m alive again. New 80GB drive, back in business.

I don’t think we’ll able to finalize this discussion until we hear more from Leonard about what Ballistic is offering. If we do need to compress, how should we use. I’ll check out the HD compression you were mentioning Trey. The scrubbing issue is likely a big one.

#26

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.