yt

This is more for my own personal historical archive than anything else, but i thought i’d post it on my blog in the event that anyone else was interested.

The idea to make a video of me playing chain factor came as a result of me not finding any online videos of gameplay that could help guide my own play to being better, so i decided to make my own video of one of my better runs. I got lucky – the run that i ended up recording was the first run that i did, and while it’s not my best score, i felt it was good enough for me to use.

the run actually took about 22 minutes to complete, so the first step was to speed up the video so that it would meet youtube’s 10 minute specification limit (although i do realize that yt’s limitation has more to do with filesize rather than length). Doing that meant that i couldn’t use the original music/soundtrack without it sounding ridiculous, so the next step was to find music to go with the run. Ten minutes is longer than songs typically are, but i immediately rejected the idea of using more than one song because i didn’t want the video to be broken in half by two songs. The only piece of music that I had in my iTunes library that was close to ten minutes was Cheating, Lying, Stealing by David Lang. so i sped it up slightly to get it to the needed length, and planned to just stick it in the background of the video.

Once i had chosen the tune, it didn’t feel right to just have the piece sit in the background while the video did nothing but show a static gameplay field. So i decided some basic manipulation would be easy to do. So the “tremor effect” for all of the opening kick drum segments was born. At the time, i was to just going to do that in appropriate places and call it good, but once i started to put in the effect and thought about what was happening in the rest of the music, it wasn’t enough. I felt like the music deserved more – it’s a fantastic piece with a lot of immediate appeal as well as a lot of analytical depth. To have the video manipulation not reflect that depth goes against my general artistic principles. So i started brainstorming in my head ideas for what should happen in each section of the piece.

And it kept growing. and growing. and, um.

here’s a basic rundown of each section: the effects, the motivation behind them, the evolution of them, and some of the technical construction of them:

Section A (0’00”-0’34”) – Tremor Effect: I went to the web to figure out how to do this in FCP since i don’t have a copy of After Effects or a similar program. Basically it involved creating a copy of the snippet of video in question, and then doing a right and left reposition multiple times every two frames. I decided that the only thing that i wanted to actually tremor was the playfield, so i had to create cropped copies of the right “score” side, the left “Back To Menu” side, and the bottom “Level Up” side that would run independently of the playfield. This would be key to later sections.

Section B (0’34”-1’06”) – Echo layers: Originally, the idea i had was to create a “ghost layer” every time the cello changed notes. Each layer was supposed to clearly come from the spot that it just got left off, and all of the layers were supposed to be slower. I tried this at first and decided after i watched a few layers that it moved too slowly and was too boring, so i changed the concept to instead make the layers a mix of slower and quicker and have them start in a spot where at the very end of the section they would all converge to the same moment.

This was very early in my FCP video editing chops – if i had done the middle/late sections first, i would have done these sections differently. Probably a little cleaner, and also more interesting.

Sections A’, B’, A” (1’06”-2’41”) – Recap and Ripple: The ripple is the only thing that i did differently for the section recaps. That was a basic FCP video effect; nothing too special there.

Section C part 1 (2’41”-3’39”) – Moving Menu/Score: Originally i had an idea of having either the score or the menu jitter around for every piano hit, but since i lost my score to the piece from when i analyzed/performed it in college, it ended up being too daunting and impractical. I still wanted the menu and score to move, so i simplified the criteria.

i took the screen and replicated it six times: one for the cropped version of the playfield, one for the “Level” indicator on the bottom, one for a white bar on the left side along with the sound toggles, one for the “Back to Menu” that was on top of it, one for the white bar on the right side, and one for the score that was on top of it. The white bars served as a backdrop for the moving menu and scores, and i’m guessing that i probably did this in the most inefficient way possible – i didn’t create a .tga of a static white backgorund, i just cropped a white portion of the playfield and then zoomed it by 1000 percent. I’m betting that this took extra processing power because even though the video was “invisible” since i only picked a portion of it, i imagine that the video was still running in the background, which would have caused for more cpu needed and more time to render. but oh well.

getting the menu and score to move was a fairly simple matter of finding the frames with the audio that i wanted to line the move with and then creating two adjacent keyframes: one to hold the previous position, and one to immediately move it to the new position. i also added some motion blur to give the move some more “depth”.

Section C part 2 (2’56”-3’39”) – Number Fill: Conceptually the gradual number fill turned out exactly how i wanted it – start with a basic number fill, gradually hit a point where the entire board is filled with numbers by the end. In some of its execution i’m also pretty happy with what i did; it was a deliberate choice to start with a predictable pattern before finding new ways to break it – start with all 7s, then 6s, then 5s, then break that by doing something different, then break that by doing a more random pattern, then break that by turning the numbers upside down, &c. Even so, i’m not *completely* satisfied with that section because at some point it loses its sense of direction because i didn’t pace it properly and think enough ahead.

This was the first time that i deliberately decided to take a snapshot of all of the numbers indivdually into still .tga’s as opposed to grabbing small clips of video. It did me a lot of good in the long run i think – it would have been a headache both cropping-wise, timing-wise, and rendering-wise if all of those numbers were film instead of snapshots.

Section D (3’39” – 5’28”) – Rotating Playfield and Number Trails: The slowly rotating board felt appropriate for the mood of this section; since everything prior to this part was primarily percussive, the more legato sense of this section needed a more legato visual effect. The white-faded rotation that lines up with the piano cluster hits is meant to be a variation of the original “Batman” rotating segue, and although you can’t tell, it’s a copy of whatever the current playfield is at the time. Originally i had it in negative colors, but it was too distracting from the main playfield action, so i decided to change my approach.

The number trails were fairly straightforward to do, but is also one of my favorite effects in the whole video. It recycled the .tga snapshots of the previous section, just placed in strategic spots with the piano cluster hits as well. The thing that i wrestled with a little here was how the growing number of “stuck” numbers obscured the playfield, problematic because despite all of the video manipulations i was doing, the main premise behind the video was still to demonstrate gameplay. Ultimately i decided that i liked the effect too much for the lack of complete clarity to matter enough, and i’m glad i kept it in.

Section E (5’28” – 6’13”) – Moving Playfield: Another ‘gradually evolving’ section where i tried to establish the basis for the section by zooming in place, then breaking that expectation by zooming to different spots, then breaking that by adding x-axis rotation, then breaking that by adding z-axis rotation. Standard fcp functionality, but i think it’s fairly effective. i’m annoyed that by doing the z rotation, the “crop” changed so that you could visibly see the score as it rotated, but i was too lazy to try to create a moving crop to match the rotation. too much work for too little return.

Section F (6’15” – 9’06”) – Pendulum Playfield/Zoom Echo Playfields/Snare Drum Flashes: The original concept i had for the Pendulum Playfield was instead to have the hits be “mirror polarity”, as in for every hit it would flip between a mirror playfield and the regular playfield. I nixed that idea for the same reason i was wary about the “sticky” numbers in that i felt that it would obscure the actual gameplay too much. When i first did the pendulum swinging, it was an extreme and unchanging swing the whole way, and the result was pretty dissatisfying because after establishing the swing, it didn’t go anywhere and got boring too quickly. The gradual increase of the swing gave it direction but a subtle one; hopefully it’s something that you can easily not notice because it’s gradual enough and there’s too much other stuff going on, and before you realize it, the swing is at its peak.

The zoom echo playfields effect was a fairly straightforward execution at this point since i had done a different version of that earlier in the piece. I systematically created two ‘echo playfields’ that would zoom out to 1000 percent centered on a random spot, then two ‘echo playfields’ that would zoom in to zero percent centered on a random spot. This repeated for every moving note in the violin part. I toyed around with trying to make the playfields change opacity over time, but having multiple layers on top of each other achieved the effect well enough and any more lessened the impact of the swinging pendulum which i still wanted to be main focus. i did put the opacity of all of the layers back to 100% when they all came back in a collapse to try to create more visual tension. That particular moment i tried about 10 times and i’m still not completely happy with it. I had this idea of playfields zooming suddenly in in rapid succession and in rapid velocity, but i couldn’t get the effect to work the right way, so i settled for the final effect here because at this point i was also impatient to get the whole project done. I think i have a better idea of what to do if i ever tried something like that again.

The snare drum flashes came from taking a few snapshots from the background combo flashes, photoshopping out the gridlines, and then putting them all in frame by frickin’ frame. Granted, once i got the main repeating pattern, i could copy/paste the repeating pattern and place it when i needed to, but for each one i also had to make sure that where it hit didn’t potentially collide with new objects in the playfield, so it involved looking at each one fairly carefully, and when the pattern was interrupted, i’d have to shift the whole pattern around.

the snare drum hits in the music contribute greatly to the tension of the climax, and although i think i conveyed that okay in my visualization of it, it gets completely lost because of the echo playfields zooming in. I’m not completely happy with how that turned out, but again, after so many failed attempts and just wanting the whole thing to be done, i decided to call it good.

As for the final recaps of the opening sections, i put some consideration into doing something different with it to give it a better bookend but decided against it because doing anything different felt like it would have been completely out of context.

The whole project took me roughly six or so weeks to complete. crazy considering that originally i was going to make it a one-session video edit and call it finished, but i’m glad that it turned out the way that it did, because i’m happy with how it turned out, and it’s expanded my vocabulary and conceptualizations of what i can do with video manipulation which will hopefully help me with my Green Lantern project.

go me.