Post
by rose4emily » Thu Nov 25, 2004 3:22 am
The narratives video is now about halfway done, and the hard part is all behind me.
I wrote a Java program that reads an audio file and, based on the average amplitude over each set of samples corresponding to a frame, selects which narrator image best fits that amplitude and writes it to a shell script that I then use to compile the finished narrative.
I then composited together the remaining image sets needed for the 4:3 narratives, so I'd have plenty of material to work with.
I also created a secondary set of images that allow "Miazawa" to blink, giving her a somewhat more lifelike performance.
The amazing part is that, after I run each narrative through all this process, it actually works pretty well - and produces a perfectly watchable talking narrator that selectively blinks and is accompanied by a slide show neatly framed by her television display. This means we don't have to use any cheesy tricks like moving the camera into just the slides durring the narrative to make producing the narratives possible - something which really didn't look all that great for some narritives, due to their having a relatively small number of "slides" to show.
So, I spent six hours on a lipsynching program (about four and a half of which were spent on trying to get it to read a "wav" file properly - something which fell together pretty quickly after I remembered that x86 processors use a "little-endian" byte order), and am now able to finish off a rather nicely animated narrative for each segment in about 15 minutes (most of which is spent on preparing the images and deciding where the "blinks" and slide transitions should go).
Tomorrow I'll put together all of the widescreen narratives, using my new narrative recordings and the widescreen "set" I assembled the other day, and try to post a couple from each set on my Grace account (on the school-administered server) back at RIT so you can see an example of what I've been up to.
Hopefully you'll agree that the lip-synch, while not perfect (there are certain subtleties that can be percieved but would require a lot more code for me to deal with them), is at least as close as is seen most anime, and far better than that seen in most dubs. It'd be nice if I could also get some head movement or something like that into the narratives, but I haven't had much success with trying to do so, and don't think it's too big a deal for the 30-second length most of these narratives fit into.
---
Song - I don't yet have your narrative for "Simplicity". I also only have an older version for "Warrior's Dance", which is noticably noisier than the new narratives. My personal FTP is either offline or inaccessable from my current location, so I'll PM you with info on how you can drop those on my Grace account, so I'll be able to complete the fullscreen narrative set.
---
The end credits script is just about done. I'm still down a couple images, but I'll be ready to assemble a "proof of concept", at least, by the time I'm back in Rochester using placeholder images. I can then substitute in others if I recieve them after-the-fact.
I've created "video image" substitutes for every video I don't have an end-credits image for using a screenshot selected from each video. I also did a little tweaking to these images to help them look good in the context of the end credits.
---
The intro is being done on the same idea as the end credits, with text (in the frilly font used for the video titles) appearing and dissappearing over the course of time. Unlike the end credits, the text is generally just being faded in and out of view. I am working on a little idea to give the main title itself a little life with a moving highlight (I don't have any good works for describing what I'm going for, but a lot of film studios do it with their logos).
---
I've also taken care to shave off anything the video segments themselves that really doesn't belong - like the repeated audio at the end of my "13'37"", or the pre-existing title bumpers in "A Boy I Knew". No actual content had been removed, just a couple of redundancies. Actually, I think those were the only two things I had to trim off, but it's late, so I might be forgetting something.
---
Finally, the phrase "the end is near" is more probable with relation to this project than it is with respect to the existance of humanity.
may seeds of dreams fall from my hands -
and by yours be pressed into the ground.
