and so it begins... for real.

For insight onto the intentions for my thesis, I present the outline of my initial plan, created in May 2009. (click to enlarge it to readable scale.)

This outline details my interest in nostalgia, as well as how I specifically plan on attacking the visual and audio elements of the piece. Knowing that my methods would be dependent on what I obtained through my forthcoming interviews, as well as the results of my visual development, this outline was intended to be flexible. Nevertheless it's a solid indicator of my original goals.

vfx - final project

ARTifact - finished project

here is my final result of using compression artifacting intentionally. i didn't end up using found footage for this piece.* rather i morphed my present day self out of a still image from when i was young. as i was working out my ideas on nostalgia, i was repeatedly thinking about multiple times within one moment, as well as trying to prolong lost moments of the past. i love the idea of animating still photography for its application in this concept.

for the sound design, however, i did end up using found materials. although i'm getting a composer to do my music for me, i've found it fun and helpful to try to work these things out on my own.

*i've determined using this hd quality footage yields a different compression effect than found footage. since the pixels are smaller, the morph result is smoother and more blended-looking.

vfx - texturing and lighting

texturing is not as fun as modeling!! the plates and the tennis ball need work, but i don't think it matters for the shot i'm doing. as eric says, don't spend time on details that the camera will never see. i am pretty pleased with how the cup came out though.

the only previous lighting experience i'd had was on still objects in maya. it was a little more challenging lighting something well when it is moving through a space.

vfx - modeling

the design for my ship is the starship enterprise, made out of a plastic cup, a tennis ball, paper plates, pencils, paper clips, and toilet paper tubes. my original reason for picking this design was to motivate me to emulate reality. i am a firm believer that in order to abstract, it's important to first learn realism, to know what you're abstracting from. besides, i figure it would be much more valuable to learn how to purposefully replicate reality, since i'd be shooting for something very specific.

i wasn't thinking about it at the time, but this design reiterates my theme of appropriation. it just goes to show how much it's a part of my creative fiber!

ARTifact notes 3: got it! ...kind of.

after hunting around on the internet for a few days, i finally came across two programs i could use to intentionally ARTifact. both are free downloads, and essentially, they're just too simple to figure out the errors i'm purposefully setting up. the first program is ffmpeg, where i can import a video and set up an export so that it only sets an i-frame once (at the very beginning.) then i take it into avidemux - a simple editing program that also displays whether or not a given frame is an i-frame or a p-frame. then it's just a matter of finding those i-frames and deleting them.

after weeks of frustration, my first test - though crude - felt like a huge success:

ARTifact notes 2

conveniently, as i was getting more and more frustrated by trying to intentionally replicate the look of compression artifacting, a few music videos came out within a couple weeks of eachother. one was a kanye west video, but i prefer this one:


using these videos as search cues, i was able to compile research on how it the effect is achieved.

the effect is colloquially referred to as "datamoshing," but that name is kinda dumb, so i opt to call it what it really is: compression artifacting. then i cleverly capitalize the ART part to emphasize the intentionality. heh heh.

first, i found an interview with the animation house that made the kanye video. the information wasn't detailed enough to be terribly useful, but it at least got me started:

1. no compression QT.
2. take into a program that upped the data rate significantly. then compress into an avi codec.
-enunciates each cut as an I-frame.
I-frames hold info for color and structure for a series of other frames called delta frames. I-frames happen at the beginning of a cut. delta frames move the key frame through time.
3. take compressed avi and bring it into another program that would illustrate all of the keyframes. then delete keys. tada!

unfortunately, since this didn't explain exactly what programs were used, or how to actually achieve the effect, more research was required. i was also pretty foggy on the concept of i-frames and p-frames, so i also made the effort to learn more about those:

useful links:
http://popmodernism.org/appropirate/delta.html
http://download-finished.com/

questions i have: can i do this using youtube vids, since they're already so compressed. what do i do to find keys (i-frames)?

my notes on video compression:

interframe coding - in video compression, the coding of differences between frames.
intraframe coding - compressing redundant areas within a single video frame.
A video sequence is made up of keyframes of the entire image. Between these are delta frames, which are encoded with incremental differences. depending on the compression method, a new key frame is generated based on a set number of frames or when a certain percentage of pixels in the material has changed.
delta frame types:
p-frame: predictive/predicted frame
b-frame: bi-directional frame or bi-directional predictive frame.

in a motion sequence, individual frames of pictures are grouped and played so viewer registers videos in spatial motion.
b-frames rely on frames before and after, contain only the data that have changed from preceding frame or are different from following frames.
p-frames follow I-frame - contain only data changed from I (color or content)

all frames needed for predictions are contained within one GOP (group of pictures).
- can be as small as one i-frame, not usually larger than 15 frames.
every video frame is broken into blocks of 8x8 pixels or y, r-y, and b-y.
-these are divided into macro-blocks 16x16.

b-frames require less data than p, but more encoding/decoding

mpeg2 GOP order: IPBBPBBPBBIP....
always 1 I, 1 P, 2 B's
Data is stored this way, though it is out of viewing order. GOP's can be open or closed - determines if that GOP is independent and can be cut. really it's: IBBPBBPBBP....

with my youtube stuff, if i export a jpeg sequence, these essentially become all I-frames?

bitrate - number of bits processed per unit of time. bit = binary digit (0 or 1). 1 = on 0 = off.