Lim: Notes
How I Made My Vid
June '06
This is How it Works
I'm writing all this down while it's fresh in my mind and I can just sort of spew it (and also I have all the files open right now).
I have no idea about Final Cut or that sort of thing. I just made this up as I went along.
I spose I should also mention that I do have a bit of musical skill so it'd be disingenuous to suggest I am a total n00b at all this stuff.
I hear music in my head. All. The. Time. When I think about certain concepts, especially strong emotions and memories, I often experience them aurally or rhythmically. I move to a beat. I also have a slight hearing problem (it's not my ears, it's my brain) that means I can't truly pick out speech from other sounds; friends from RL will know I have a sometimes unnerving tendency to stare at people's mouths or abruptly turn off music/telly and slam shut windows to try to isolate different sounds. Mainly in broader social settings I just smile and nod and try to look like I know what's going on (I rarely do; I can never hear anyone)...and listen to music in my head! :D
So all this is to explain why for me everything is about the music and movement. I am not a storyteller, naturally. In fic I rarely manage to achieve plot, or message; I'm more like an interpretive dancer, only precisely ninety thousand times more pretentious. Oh, god, kill me now.
The Source.
I listened to the song. I played it on my computer. I played it on the stereo. I learnt the words. I sang it to myself in the shower, in the swimming baths, in the garden, as I boxed. I played it on repeat in my head and on my headphones. I danced to it in my sitting room. I wrote out the lyrics and stuck them on my fridge. I played it to myself on my dining room table, I played it to other people. I played that bleeding song until I walked to its rhythm everywhere.
I neither reread the fic nor rewatched the show in this time. I knew it was for Freedom though — it sounded like Freedom.
I set out the lyrics and wrote the basic timings down the margin. I came to think there were about 23 distinct parts to the song, so I marked those out too, then I assigned a number to each line within each part and on a separate sheet of paper I listened to the song again and wrote out immediate visual associations I had (like mallet's mallet), which gave me somewhere to start with gathering the video source. Some of these immediate associations were pretty literal (our hearse -> puddle jumper) (ancient bees -> those swirly Ancients lights), but I knew that once sequenced they should build together because the song is so well constructed...it has a narrative.
[Cathexys asked me why I chose that toast scene on the line dying breath
(0;01;50) and not the one that fit with the story with him flying a jumper and I thought not because we know that John did not die in the footage we have of him doing that and so it might be weird and confusing. I dunno. But the Auroran crew, in the end, really did breathe (exist with purpose) until their dying breath - which is awful and wonderful at the same time and I know that John taught Rodney that kind of hope and about duty and oh my god. *flails*]
I manipulated that blend by hand, too. See? Fun with motion blur, vivid light, and liquify!
About here I also went to sxc and ran some keyword searches on things like "students" classroom" "space" "pilots" etc. When I write fic I do a similar thing with the visual thesaurus. I wanted to get some visual associations from other people, to sort of anchor myself, because I do tend to spin off madly into incomprehensible riffs left to my own devices.
I took a load of downloaded avis that I had on our home network from The Binge and I roughly cut out a load of clips with Quicktime Pro. I pasted them into iMovie and pasted the audio at the playhead. This was my storyboard. In iMovie I moved stuff around, cut out bits, threw some parts away, and noted what needed filling in. I only spent a couple of hours on this - very rough story-boarding in front of the telly one evening. I just wanted to see the music with the movement, to get a feel for it.
Then I cut out the clips I wanted a bit more neatly and opened them in Adobe ImageReady CS. I sampled every 3rd frame (I guessed) and saved each resulting document as a psd.
Then I opened up the files and started cutting away the chaff. I cut out any talking bits and wonky bits and then I set about making each part kind of beautiful. I like mucking about in Photoshop much better than editing stuff in iMovie and fun is key for me. I find iMovie really frustrating and unfun. In fannish terms I basically made icons out of the frames, colouring them, blurring, luma-keying...The chalk and charcoal clip right at the beginning was actually really really long and was originally going to appear throughout the vid but in the end that tiny part was all I kept because, well, I don't know; I thought I was labouring the point. Although there are a couple of thousand still frames I didn't actually go through each one by hand, colouring and editing. I grouped the frames in sets according to scenes/shots and then worked on the topmost layer. I then recorded the processes as an action and applied the action to all the other layers in the set.
Some parts needed to be done on a frame by frame basis. The last section with the text and the cat (00;02;19 to 00;02;32), I erased the background by hand, but I didn't do it neatly or anything so it didn't really take that long...I simply zoomed out to about 25% view, took the block eraser and made two or three broad strokes, then zoomed back in and with an airbrush set to 100% drew an outline around the characters. We're talking 5-10 seconds per frame. Like, 10 minutes. And it looks great. The four frames with the wraith on 90% opacity (0;01;03) took me forever and it looks shit. It's a mystery, da?
Then I needed to rename all the layers so they were consecutively numbered (I didn't realise this at the time and had to go back and do it later). I downloaded a script from here and then I renamed 0-9 to fit 00-09 by hand in each one. (I actually asked the dude to fix this and he did, but only for CS2, which I don't have)
I now have CS2 and the redone script, which didn't work as it happens, but my mate fixed it instead and I can easily email to anyone who wants it.
...I had a whole part with stop-motion from those pictures I downloaded, which I jettisoned for this vid. It just looked like it was getting too much, you know? However, I did spend one very entertaining evening manipping Rodney's face onto a still image of a lecture hall full of students for morphing but I had to cut it. It was too silly, however happy it made me personally, and it emphasised Rodney's ambivalence about teaching over his love for John, which is all that matters! That and his manpain! Oh, no, it's not all that matters but...oh, sshh. In the end all that is left of that section is the whiteboard in the titles. The perspective is sloppy in the text but this is okay. OK? We are telling ourselves that sloppy perspective is OK. In small doses. Sanity is more important than perspective. Oh, wait...
So I sort of mucked around with this for a bit, not really knowing what to do with myself and then a few days ago I got hold of a copy of Adobe After Effects. You can download a 30 day trial from Adobe. I'd never used anything like it before but it looked really ace so I read the manual, some tutorials, a website or two and watched a few podcasts. The help files are completely clear and useful, which I thought was against the law but there you go.
So then in AE I imported my mp3 and set up a main composition, the inspiringly named "Comp_1". I set its frame rate to 12fps. I converted the audio to keyframes and generated some random numbers; I related those numbers to the audio waveform using a simple expression. Expressions create relationships between layer properties and use one property’s keyframes to dynamically animate another layer.
This was the basis of my vid - it's the music being expressed through numbers... I find this really hard to explain because I sort of think I already said what I wanted to say... I just said it in my vid. I don't know a better way to express it and if I did I would have expressed it in that way instead and not bothered making the vid in the first place. I can point to little six-frame parts and say, look! Memory! (0;01;50) or Things Learnt the Hard Way (0;01;40) and stuff but I...
Anyway, I then composited the generated animation on a white solid and precomposed the layers to make a new composition for my project file, which I called numberPulse. I find these sorts of briefly descriptive filenames useful.
Then I imported some more stuff for my project file, a load of psds, some stills, some movie files. I divided them up into folders because with so many assets the project file was this huge list of data and I have a 12in iBook and I couldn't physically see what I was doing.
I started making compositions. The frame sequences I made by pulling the folders from my project file onto my composition timeline, then sequencing the selected parts (just click sequence if you have consecutively numbered files). Then I precomposed each sequence and nested it back into Comp_1. I actually fiddled with these sequences fairly significantly and made them much jerkier (0;00;20) and sped up parts and slowed parts down (0;02;12) and cut out even MORE frames (0;02;33) and flipped some round (0;01;00) and all sorts of nonsense. This, I think, is the difference between image manipulation and vidding, because in vidding the aesthetic seems to be "the clearer and smoother the better" whereas I kind of think I want to make pretty graphics that speak to me on some emotional/nonverbal level.
Anyway, my nested comps. I made loads of these wrong and had to start again a fair few times, but that's OK because at each stage (avi source, .mov clips , psds, project compositions) I kept a version so I never had to track too far back.
Apart from the original avis, which I shoved back onto our network pretty sharpish, the whole project took up only 2.8gb on my hard drive, which is pretty decent, considering the final render was 700mb. Also, because I imported each psd as a folder, and then made each section into its own composition, and then nested everything inside Comp_1, it was pretty simple to open up each part and edit any individual frame I wanted.
**[ I didn't layer and precompose every sequence in this way. Some I imported into their own master composition and then nested from there. I did this because when you precompose a selection of layers inside a master composition, the resulting comp will be the length of the master composition, which can be annoying if you want to then have several instances of the same composition along your timeline and overlap or blend them, just from an ease-of-manipulation POV - with a composition made independently, with its own settings, it's easy to see where the beginning of the visible sequence is, as opposed to the whisker of 0;00;00. You can manipulate any comp setting retrospectively, of course, but just an FYI. ]
I layered up these compositions just like you would layer still images in Photoshop only with the added dimension of time...Basically AE IS Photoshop, but with a timeline. For my purposes it was, anyway. So things look pretty different in their own comp and in the master comp - according to the blend modes and their position in the layer order and such.
And then I played. I am just learning so I used hardly any of the preset effects, except I did apply DataPacket and TypeWriter animations to the text, which you control a bit like kerning in PS. I had my bit of paper with my timings and my major changes scribbled on it; I had my various receipts and backs of envelopes with all my important thematic elements recorded (I find envelopes inspirational! Don't judge me.) so I just kept referring to that and building up the timeline.
I set opacity keyframes to blend the layers and I cut everything by ear. I did initially mark up the waveform but the window is so tiny on my screen it was only useful for one or two layers:
which is kind of funny because one of the main reasons I hated iMovie was that you couldn't a) edit clips by entering numbers and b) generate a template from the waveform. So then I get this program that can do both of those things and I ignore it! I'm an awkward bugger sometimes. I still hate drag and drop interfaces though.
I used colour solids on different blend modes (Colour Dodge, Exclusion etc) to tie together some of the more disjointed footage. I used a luma key to silhouette the lads in the first section (0;01;00- 0;01;03), not hand-keying like in the last section (00;02;19 to 00;02;32). I used the same five or six colours throughout the whole process, though I think there is one part (0;01;57 - 0;02;03) where this really breaks down and it could have been much better handled but I'm not obsessing over it now omg! On to the next one, I say.
So then when I was happy with it (not really happy but fucking done with it) I rendered the whole file to a Quicktime Movie, which took about half an hour? Fifteen minutes, maybe. And then I was right back again in Quicktime Pro7, exporting to divX, avi, m4v and .mov , for different players and bandwidth and such.
So there it is. Made on my iBook, G4 (1.33) with 512 mb RAM, running OSX Tiger, though I think you can run AE on almost anything - I know Ubuntu Linux supports it. Hardcore.
That's all I can think of right now. Sorry if it's incoherent. This vid has a companion vid in my head to "Fidelity" from the same album but that's a project for another day.
And here is my plan, painstakingly compiled from my bits of paper:
ATLANTIS/ATLANTEANS AS PRESENT GHOSTS JOHN PARALLELS WITH ATLANTEANS NOW?? WHAT? ROD->JOHN->ATLANTIS (via students) = DIRECTION OF GAZE/SPEECH that solo's really long, 10,000 YEARS LONG HAH aurora shots virtual reality INSIDE ROD'S HEAD?? (1)lecture theatre (2)rod look down and blink (3)puddle jumper (4)stargate RADIO melting ice running from wraith??? Explantis! BlueShep omg notes on the cage in Aurora can AE do that talking, playing, teaching, classified, secret, silence, music, saying, not saying, hearing, leanring, knowing, teaching, talking pos[rod lecturing faces all turn into him, then all disappear leading just one on one, take apart one, morph to machine- pieces become people (Shep?) parts remove heart brain mechanical oh don't be so ridiculous you idiot] rod empty cup ford turns his back shep raises his glass and DIES OMG DEATH but also hope yes atlantis glowy panels
Er, yeah. Sorry about that. Notes are bollocks, eh?
+From comments and discussions on This is How it Works
some excerpts which may be useful
July '06
Okay, ummm, let's see what I've got open here. So I made you this example to kind of hopefully illustrate what I mean.
This example is long lost, I'm afraid
See here is the titles composition with different elements all on individual layers. And there are blocks of colour and there are lines of text and all that jazz and they possess a whole load of properties and I've assigned values to those properties, like, for example, to the text layer "by lim" duration property I've assigned a length of about 4 seconds. And actually I did that by dragging the little handles together but I could have done it by simply inputting the appropriate variables.
So, it's a bit like any other code. Like CSS. If you think of your element, you need to describe it like a little biography:
Jane{height:5'6"}
Element{property:value}
But wait! There's more! We can change the values on all those properties, of course, and really maybe it's better if you think that each element has an enormous number of properties, most of which have a value of zero - the default variable is zero (and an element where all the properties have a value of zero is No Thing). And so if you can think of a state: a property, and then you can think of a progression, then you've made an animation.
[Please note that that is in fact linear interpolation. I wasn't paying attention when I was scribbling away.]
So here on the element "Rodney"'s scale property I've set a keyframe (a state of being in time) in one place and then another keyframe a little while later with a smaller value. And I've set the interpolation method to linear, so it follows a smooth line from one state to the other (joins the dots).
And see here I've applied an effect to the text and animated it that way. An effect is a shortcut, it's a phrase. Instead of painstakingly describing an echo every time, you can simply describe the type of echo you want - assign a value to the property "echo", which in itself posesses properties. It's a language.
And then I've nested that composition — I've folded it up so it's like a layer — into my main composition, see:
So just like all those opacity and position values were wrapped up into the effect "echo", now all those elements and all their behaviours are wrapped up into the layer "titles", which you can then start right back assigning property values to again just like we did in the first place.
And then you can start blending. *g*
[Some more to and fro-ing]
Yeah, sure. Well, I had the tune and a project file open anyway because I'm vidding that song just now so I just wrote in some random nonsense and it was the work of a moment! ::makes complicated flourishing gestures::
To the text at the beginning I just applied an effect. I opened up the animation presets menu, picked the one I wanted and dragged it on to the stage. Then I changed some of the settings a bit, I think.
If you look at it going backwards slowly, frame by frame (just press your left arrow) it's easier to take apart what it's actually doing, because all of the letters are actually doing the same thing, it's just that they're staggered.
It's moving through the alphabet as it approaches its destination. That's its path.
- it's going from here to there in a straight line (dropping down)
- the letters are moving incrementally (an increment of -1) through the alphabet from Z until they get to their destination (of the text I originally set)
- as those letters travel they are moving incrementally through opacities from 0% opacity to 100% opacity (fading in)
But, you know, I didn't know that it was going to do that. I just set its end state (just the text written out as normal), dragged a preset that seemed appropriate to the sound onto the stage and had a look! :) So there are different ways you can engage with it.
This conversation eventually resolved itself into a screencast, which you can find here
I sampled every third frame from a 29.97fps mov because you can get smooth motion at a frame rate of 12 frames per second. That's about how the human eye works. I wanted some parts to be jerky and some to be smoother and some to run faster than the original so I sampled it (to translate to approx 10fps) so that I could choose between six and twelve frames to put in each second of the timeline. I'm sure there's a better way. As I said, I'm making this all up as I go along really.
I think it's rare to animate frame by frame. In general, animation is based on a process of keyframing. You set a start point and an end point and then tween (fill inbetween) rest. You can do this by hand or, on a computer, you can interpolate. It makes sense to do it with maths on a computer. If one can describe motion one can programme motion paths.
So, like in AE you can use linear or bezier keyframes to animate an object along a path or curve. In fact, you could do that with any computer, I could probably programme my calculator to do that if I were really REALLY bored, but we're using AE the tool.
Of course, you don't have to just move things around. A path isn't just about velocity; it's a progression of being; it's a story. If you think of a position or point as a state then it's easy to see how you can animate any state of being. Like the state of being blue along the path to the state of being red.