Zen Films

Render Jockey to MindFlesh Render Junkie

Good grief, these blog entries are getting fewer and further between – must have been busy these past months J I promise to do better in future. I have to confess to becoming a “render junkie” – sometimes waking up at

3am

to check on the damn After Effects render and starting again if it’s fallen over! It’s not always a fault of AE – on one occasion I managed to somehow get a corrupt Targa file that halted the render.

I’ve adopted a two-stage rendering process: first, render the composition to a Targa file sequence and the second render the TGA sequence to whatever format I need – Quicktime, MPEG2 etc. I think it’s a pretty good process because the rendering a 20min composition with all the layers and effects takes about 15 hours; rendering the Tga sequence to another format is about 5 hours. And that’s with the Nucleo plug-in. Without it render times are 3 to 4 times longer.

At this stage, all the effects and colour correction has been done to at least a “good” standard but now I plan to go back and refine further and elevate the work to “excellent” level. Why two passes? Well, the original clean footage is ok for the basic sound work of cleaning up production sound and adding regular scene atmos but for the extra-creative sound design work, the guys need to see what the picture will look like. Trying for excellent on the first pass would take too long. In this 2-stage process we can shorten the overall post-production time.

This week we added all the completed music and it’s awesome. I think (hope) audiences will be enthralled at the approach we’ve taken (he says with everything crossed!). The last reel (final 20 mins) is absolutely intense: we have only two lines of dialogue and the rest is a full-on orgy of adrenalin. Imagine Alien meets Viagra ;) Oh….. MindFlesh.

Sound Design - part 2 of 2

This is part 2 of the interview with MindFlesh sound designer, Jeff Darby. Here he talks through his kit and presents a simple example of combing unusual sounds.

DNA Computing & Parallel Worlds..oh, and an Update on MindFlesh

Wow - it's already the start of July. This is a bumper update because I've had my nose to the grindstone for weeks...er.. and I went on holiday last week to Rhodes (the Greek island)...and there's lots to say. Here's the headings:

• MindFlesh update
• Books I've read recently

MindFlesh update

Picture progress...

Three of the four reels have now passed stage one (of two) - which means they've been graded and special effects added. They haven't been given an overall "look" yet and some SFX shots need to be tweaked but all-in-all progress is pretty good. I hope that by the end of July all the picture grading and SFX will be complete... at least to a really good stage :) I'm not planning to screen the film until Jan 2008 at the earliest so that gives me time to (a) improve any SFX I fancy and (b) find out how I'm going to get the hi-def image files off my PC and onto a HDCAM tape for screening! On the last point, watch this space.

Sound progress...

Matt the editor has now transferred all the Final Cut Pro (FCP) reels to ProTools where he's made a start at organising everything into tracks - vocals, production sound, sound effects, music, atmos etc.
Meanwhile Arban the composer has completed about 95% of the music so all that remains is a final (music) mix and then we can slot it into Matt's ProTools projects.

Books I've Read Recently

Did I mention I spent last week basking in the sun and I'm now a bronze Adonis? Oh... then that's probably because I was hiding in the shade with the redheads and babies :) Forty degrees...luvvly.

Anyway, at the airport I picked up two books, Genesis Machines: The New Science of Biocomputing and Parallel Worlds: The Science of Alternative Universes and Our Future in the Cosmos. Is it me or are subtitles getting longer? :)

Believe it or not I wasn't on the look-out for a book on biocomputing but it just happened to be the store and I thought it looked interesting. There's an important lesson in distribution there. Anyhow, despite what you might think, it's actually quite a riveting read and I couldn't put it down. It's not at all about living computers, computers made of flesh, stolen pregnant embryonic fluids or anything in the slightest ethically disturbing. It's about using DNA, enzymes and other microbiological chemicals to perform discrete computing tasks...or maybe not because it's all still theorize-build-trial-and-error. What makes the book enjoyable is the author's tone although I could have done with a few more diagrams. The author also has this tendency to provide examples for the easiest concepts and then skip over the more challenging. It reminded me of my school chemistry teacher who used to dictate his lectures - example: "use a 'deflagrating spoon'... thats spelt  's' 'p' 'o' 'o' 'n'". Amazingly the book inspired me to refresh my programming skills and while sunning myself by the pool I knocked-up a little program in Excel that plays Rummy using fuzzy logic. Yeah, I know it's a bit nerdy but it's not much worse than playing su doku all holiday.

Image Sequence Utility

I recently paid for a small utility to be created by someone at RentaCoder and I'm very pleased to be able to offer it to Zen Blog readers :)

For some unknown reason it's called "Incognito" - probably because I didn't determine the name in the spec! Anyhow, here's what it does:

• it'll check an image sequence and tell you which files are missing
• it'll copy one image sequence (source) and paste it into another sequence (target), renaming and renumbering to match the target.

Why is this useful?

Scenario One: Well, let's say I have a composition that's about 32000 images long, all rendered in HiDef. If I need to make adjustments to a part of the composition, I'll just render the changed images. Problem: for some reason, the After Effects plug-in Nucleo "accidentally" deletes one image before the start of the newly rendered images - leaving me an image short. It's a pain in the arse but worse if I have to look through a heap of files to find it. Not any more with Incognito :)

Scenario Two: I've broken large compositions into smaller comps to make the projects easier to deal with. If I then want to stitch all the comps together, I can now use Incognito to copy and paste from various comp names into a new final image sequence without manually renaming (practically unworkable) or creating an Uber-composition and re-rendering.

How to Use

Download the executable here. Your browser will complain that it's an executable but I've not had any problems. The file size is exactly 212,992 bytes (or 208k in summary). 

I think it's pretty clear what you need to do but one thing you need to know is that it expects files in the format "name_#####" where "name_" is your filename and ##### is a sequence number 00000 to 99999.

It's provided free of charge to use and freely distribute but it's provided without support or any apologies. I had it developed for me and I'm giving it away out of the kindness of my heart :) It'll be nice if you mention ZenFilms if you add it to your site for download. Enjoy.

Noise Reduction on HiDef images

I made a great purchase this week - Neat Image noise reduction software. It's intended for use with digital still cameras and scanners but I thought I'd give it a go with moving pictures :)

Now, the noise I'm reducing is not digital artifacts and aliasing: it's sometimes grain from using high gain on night shots, sometimes when I've scaled up a shot (and hence magnified the grain) and sometimes I've sharpened specific areas of the shot where I felt text could be a little clearer. I say "sometimes" because it really is on a case-by-case basis. I've done all of the above many times on this production and only had to resort to Neat Image on less than 1% of the cases.

To use Neat Image, I first output a clip from After Effects as a sequence of TIF images. Then, in Neat Image (NI) I load in a test frame - usually one I know it'll find a good noise profile from - and practice getting a good output image using various noise filter settings. I've found the noise filter presets often to be the best solution or at least a really good starting point. Once I'm satisfied, I save the noise profile and filter setting. Now the batch processing can begin: point NI at a source directory (the TIF sequence I output from AE), tell it the noise profile and filter to use, direct it to output to a target directory and then press "Queue" (which is the "do it" button).

On my 4-processor Alienware beast (boasting again) NI is pretty fast. It'll eat through 500 images faster than I can make a cup of Orange and Lotus Flower tea, never mind drink it.

Finally, holding the shift key while dragging the directory folder to AE will effectively import the sequence as a clip. Now in AE I can chose to either replace the original "noisey" clip or overlay it - adding it to the composition and tweaking back the opacity.

Nice. :) 

Data Overload!

So... the MindFlesh EDL has been exported from FCP and imported into Premiere on the PC - not without several problems though: seems that the "reverse" instruction to run clips backwards doesn't translate correctly between platforms and neither does "change speed". Oh, it comes across in the EDL but seems that Premiere interprets it differently to FCP.

Still, with these hickups out of the way, we've now conformed all the footage from the original tapes. Each reel has its own Premiere project file which makes loading much quicker (and provides a little protection against corruption spreading across the whole film!)

Due to the huge amount of data that's created, the project is spread across several disks:

• drive C (500Gb has the project files
• drive J (1 Terabyte) has the base HDV files
• drive F (500Gb) has the raw uncompressed files. So far only Reel One plus the B-roll and additional uncompressed clips for reel 1
• drive L (500Gb) has the graded/processed output files (for reel one, 20[min] x 60[secs] x 25[fps] Targa files each about 6Mb). Reel one now occupies about 170Gb

So, it's almost tidy at the moment while I work on reel 1 but as the story unfolds, data will spread to J, F and L to become one unholy mess!

Sound design, San Francisco & San Diego

So, last week Helen and I were in San Francisco on business. We managed to meet with several writers, producers and distributors primarily to renew old acquaintances and let them know that MindFlesh is on its way! :)

It was also a good opportunity to fly down to San Diego where sound design maestro, Jeff Darby, lives and talk about his ideas for the film, timescales, practicalities and so on. Here's the first of two video podcasts. In this first vidcast, Jeff talks briefly about how he's approaching MindFlesh. In the second vidcast we'll get to see inside his home studio.

Picture Locked!

Hooo yeah! Our first post-production milestone reached!

We have a 75min film that's now ready for sound design, music, special effects and grading. Some SFX have been done along the way and Arban has started composing - we have about 3 or 4 absolutely cracking sections of music - but there's still a lot to go.

Technicalities...

So what now? In Final Cut Pro (FCP) we've cut the film into four 20min sequences each with its own academy leader (the countdown) with a sound blip on the 2 and a short tail - a countdown from 3 and the blip at 2. With each sequence top and tailed by a blip we can easily check if the sound ever goes out of sync with the picture.

Each sequence is now being output to a Quicktime file with burnt in timecode (BITC) that we'll use as reference for all music discussions and SFX/grading.

Why 4 x 20mins? Well 20mins corresponds to a reel of 35mm film - should we ever print back to film we're good to go. In fact we'll now refer to the project in terms of Reel 1, Reel 2 etc. Breaking the film up like this also means the project files are faster to deal with - especially in Premiere which previously seemed to take forever to check all the clips and load up.

Particle Illusion

My Particle Illusion demo finally expired and I had to buy a copy :)

Arban Severin: Composer

I spent yesterday in Edinburgh with Arban Severin our fantastic composer. I shipped a DVD last week of the current cut of MindFlesh so that we could discuss when, where and how the music would play its role. Even though the picture hasn't been locked yet, it's good to start testing ideas as soon as possible.

Here's the March podcast where she discusses starting with a key scene to establish a musical theme and then work out from there. At the end of the podcast she discusses using cubase vs proTools.