Behind The Scenes: Resurrection

As you may have seen from other blogs here, my current project is an animated series called Resurrection. This post is a sort of behind-the-scenes look at how each episode is constructed, with a mostly Adobe workflow. All of these screenshots come from the creation of Episode 2 of the series.


The process starts off with a script, which I write using Adobe Story. Story has a fairly nice TV show template, and also tracks characters across scenes and such. The thing you’ll want to pick up here is the page and line numbers – we’re looking at 2-9-1 in this screenshot.

audio_script_pageThis script is printed out for recording by all the cast. We’ve been recording in my school’s Drama department’s wardrobe room, which provides a decent acoustic space – no humming lights (when we’ve set it up), and the walls lined with costumes prevent reverb problems during recording. Recording is done in multiple passes, one per character, with the other actors doing their parts off-mic to make reactions more realistic. I use a Zoom H4n recorder with an attached Azden SMX-10 microphone, compositing a recording out of the internal and external microphone.

This means, of course, that the audio for a given episode comes in for processing in very long wav files. That’s where Adobe Audition comes in.

audio_octavian_dialogueBasically I run through the file as a multitrack, splicing out each individual line. I then render out each selection as an individual wav.

audio_individual_fileI do a little processing on each – I sample the noise floor of each take, and use the hiss reduction tool to eliminate what ambient noise there is in the recording (our space is good, but not perfect). I also normalize all the dialogue to -3dB, so that if I need volume dynamics later I can start from a uniform volume base. This file gets saved in two formats – the original 24bit 48000 kHz, and a second 16bit 44100 kHz file. This lower-quality file is what goes into Flash (which can’t handle higher quality files). These files are named based on page number and line number, so that they can be easily referenced when pulling them into Flash and Premiere.


Scenes are constructed from two fundamental units – character models and environments.


Environments are large vector graphics sets in Flash that comprise a large area. The one above is Octavian’s office, which is built in multiple layers – foreground and background elements of varying depths. This can be adapted  for multiple shots, and character models can be placed almost arbitrarily within it.

character_model_octavianCharacter models are articulated “puppets,” which I can place into scenes and pull apart as needed. They’re a set of nested movie clips, so articulation is only as detailed as it needs to be for an individual scene. Each character model has different views – Octavian’s (in this picture) can face forward, to the side, or away from the camera. This is facilitated by the fact that the character models are symmetrical – I can just flip one side view to make Octavian face either way. As I need more views, I add them into the model, making them available for subsequent shots.


So we assemble a scene out of (in this case) two models and an environment. To try and fake a little depth of field, the wall behind the Captain is blurred out. Both Octavian’s and the Captain’s mouths get put on separate layers in preparation for the next step, inserting the dialogue. This is where the lower quality render of each clip comes in – we load it into Flash, put it in its own layer, and then we create empty frames for all layers along the length of the audio file.


This leads to the use of possibly the most useful plugin I’ve ever found for flash, SmartMouth from Ajar Productions. SmartMouth, after being pointed at the right audio and mouth layers, synchronizes the mouth’s set of shapes (visemes) to the sounds in the dialogue. This saves an absolutely colossal amount of time manually doing lip sync, and in my experience it works pretty well. I manually go over the sync later to make sure it’s not totally off, but on the whole it’s fairly accurate -more so with male voices than with female, though (it sometimes gets a bit lost on Amanda’s dialogue).


Once the shot’s all animated, I render it out to an individual QuickTime Animation encoded mov file.



This mov file gets pulled into Premiere Pro, which is where I do all my editing. Because the audio coming out of Flash is lower quality (lower even than the quality going in – I haven’t quite figured out why), I need to re-sync the high-quality audio files back in.

sync_in_better_audioThere are tools for automating this process, but I haven’t particularly looked into them – visually matching the waveforms is not too challenging. This is how the rest of the episode is constructed, shot by shot.


Audio Mixing

Once the editing’s done, I remove the synchronization audio track from the timeline, leaving just the high-quality dialogue audio behind. This is not quite done – blank audio works in some scenes, but not often. Mostly I use it in the interior of Amanda’s house. To add some environmental noise, I pull the timeline into Audition again.

add_in_foley_enviroI pull in sounds from Adobe Resource central both for audio ambiance, and for foley sounds – any object or action that should make a noise when something happens to it.

acid_special_musicThis particular episode also called for some special music, during the evidence-gathering montage. I put the music together in Sony Acid Music Studio 7.0, render it back out as a wav, and plunk it down at the appropriate spot in Audition.

sound_mix_completeThe completed audio track looks something like this. I export this back into Premiere, then render out to a 720p mp4 file for upload to Youtube.


And that’s all there is to it! If you have any questions, feel free to ask in the comment section.

This entry was posted in Animation, Artwork, Development, Serial and tagged , , , , , , , . Bookmark the permalink. Post a comment or leave a trackback: Trackback URL.

Post a Comment

Your email is never published nor shared. Required fields are marked *

You may use these HTML tags and attributes <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>