(no subject)
Aug. 12th, 2004 12:25 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
which_chickOn music videos. Now that I have basic capability in this (still some areas I'm foggy on, I totally lack the nifty tools of Adobe, and linear editing is going to kill me or drive me insane, but I can make clips and stick them together), I've been thinking about how music videos work. Music videos are like essays. They need structure. They need narrative (or whatever) progression. Could be variation on a theme, could be explaining/elaborating a concept. But they need progression, so that you GO somewhere with it. You can't just do "Explosions Are Cool" or whatever. Well, you can. But I'll not be watching that one, thanks.
One-joke AMVs are not very funny. A lot of funny or satiric efforts peter out (Try not to snicker.) after the first refrain. They've not got staying power. (No, it's not on purpose. I've been looking at fifty shots of that damn tower this morning and without sound, it's hard ... DIFFICULT to keep from noticing how it dominates... fuck. Fine. Snicker all you want.) And I think the lack of stamina is due to the fact that there's no progression beyond the first joke. If you've not got anything new to show me in the second half, I'm going to get bored. (Buffy voice: Bored now.) Do not spend the second half of the song reiterating the gags from the first half of the song. Go somewhere different with it.
On visual pacing. I've seen music videos by other people, and I like the ones where you're not sure you're seeing it all more than the ones where you are bored-as-hell looking at the same thing over and over. There is a lot of room for 'funny once' tricks in the AMV world, but do realize that they are funny *once*. Just because the song repeats does not mean that the video has to. In fact, when the song repeats, the video really shouldn't. You've got known text there in the music, so there's spare processing capacity for new video information. Extend the thesis, there. That's what your refrain thingie is for. (English major. I'm allowed to say things like "Extend the thesis" in informal livejournal postings about anime music videos. I have special dispensation.) Do not repeat the set of cuts you did the first time 'round even though it's easy to cut-n-paste. We've seen them already. Give us something new. Yes, even though it is six million times as much work. At least riff the one you did the first time. Variations on a theme worked for lots of dead white composers, it can work for you, too.
I don't like it when the images go so fast that I feel a seizure coming on, and I don't like it when there are a million cutesy effects if they don't help tell me get what you're telling me. Cutesy effects are like frosting roses. They're nice and I like a few, but if the cake sucks, they're not going to save it.
Lipsynch can be done with amazingly funny results... but it's going to be a right royal bitch to achieve in a linear editor. I am not laying out several hundred bucks for Adobe, which wouldn't run on my machine anyway, so linear editor it is going to be. Yippee.
Interestingly, experiments have shown that it takes me about four or five frames to get a good idea of what the hell I'm looking at. Three frames is not really enough. It doesn't leave the me enough time to process the information comfortably. These numbers were scientifically (yeah, right) determined by me sitting here and twinking with VirtualDub to find out what I am comfortable with in terms of getting the information vs. not-getting-the-information. I did look on the internet, but the internet was singularly unhelpful on this and I wanted to know *now* and not three hours from now. Let's hear it for trial and error...
Anyway, I need four or five frames (~.20 seconds is five frames) to process visual data well enough to integrate it into a coherent story. Three frames (~.125 s.) and I can *see* but there's no time to *process*. Stuff gets lost. I think I'll need to use the bigger exposure if I want people to get the jokes the first time around. Betcha that once the iconic stuff has been set up, though, that I can flip through 'em faster, like in a refrain sort of way, because it'll be review. Mmm-hmm. I can kind of see how that would work.
It's quite possible that young kids today will not need as many frames as I do. I'm old and slow and don't spend enough time watching video games... but I *still* read faster and process jokes faster than normal kids (judging from the crowd reactions at the fan parody track at *the con*), even when I'm dead tired and slightly drunk. So. I figure it'll all work out in the end.
I suspect that the exposure interval can be lessened for intercut scenes... scene A, scene B, scene A, scene B. That'd probably be (a) interesting plot-wise and (b) workable for shorter intervals with more visual noise. Yep. I'll have to try that next. Or, you know, intercutting actual scenery with black. I bet there's a way to generate plain black frames with the tools I have... people need plain black for leader spaces and fading-to-black and stuff. Then I'd have plain black to intercut with video. Too much of that would be not-good but it'd be interesting to play with in very small amounts. Might make a nice beginning, ending thing.
(time passes)
I am now able to make plain black frames that go nicely with my existing .avi files. This is a non-trivial undertaking because the documentation here is relatively minimal. The web site for AviSynth at SourceForge.net says:
# produces a black clip (3000 frames) with the remaining clip properties of the avi:
AviSource("E:\pdwork\DO-Heaven.AVI")
BlankClip(length=3000, color=$000000)
The web site is not exactly what I'd call correct. That does NOT do it. Extensive trials show that the script above produces a black clip of 3000 frames at 24 fps. Now, VirtualDub is not the world's most brilliant program but it's not stupid about fps's... It can damn well tell if you're trying to sneak two slightly-different clips past it and mash them together. Simply typing in 23.976 for the frame rate in the AviSynth script is not going to fool VirtualDub. You have to match your black frames to the original source, you poor sap, and the syntax in the example is not going to help you at all nor is manually setting the framerate. It has to be matched to the original, not typed in by your big, fat fingers. So, you know, blow a couple of hours one morning trying to figure out the problem... or read on for the solution.
The following does it:
BlankClip(AviSource("c:\d2vfile\sku2.avi"), length=30, color=$000000)
One-joke AMVs are not very funny. A lot of funny or satiric efforts peter out (Try not to snicker.) after the first refrain. They've not got staying power. (No, it's not on purpose. I've been looking at fifty shots of that damn tower this morning and without sound, it's hard ... DIFFICULT to keep from noticing how it dominates... fuck. Fine. Snicker all you want.) And I think the lack of stamina is due to the fact that there's no progression beyond the first joke. If you've not got anything new to show me in the second half, I'm going to get bored. (Buffy voice: Bored now.) Do not spend the second half of the song reiterating the gags from the first half of the song. Go somewhere different with it.
On visual pacing. I've seen music videos by other people, and I like the ones where you're not sure you're seeing it all more than the ones where you are bored-as-hell looking at the same thing over and over. There is a lot of room for 'funny once' tricks in the AMV world, but do realize that they are funny *once*. Just because the song repeats does not mean that the video has to. In fact, when the song repeats, the video really shouldn't. You've got known text there in the music, so there's spare processing capacity for new video information. Extend the thesis, there. That's what your refrain thingie is for. (English major. I'm allowed to say things like "Extend the thesis" in informal livejournal postings about anime music videos. I have special dispensation.) Do not repeat the set of cuts you did the first time 'round even though it's easy to cut-n-paste. We've seen them already. Give us something new. Yes, even though it is six million times as much work. At least riff the one you did the first time. Variations on a theme worked for lots of dead white composers, it can work for you, too.
I don't like it when the images go so fast that I feel a seizure coming on, and I don't like it when there are a million cutesy effects if they don't help tell me get what you're telling me. Cutesy effects are like frosting roses. They're nice and I like a few, but if the cake sucks, they're not going to save it.
Lipsynch can be done with amazingly funny results... but it's going to be a right royal bitch to achieve in a linear editor. I am not laying out several hundred bucks for Adobe, which wouldn't run on my machine anyway, so linear editor it is going to be. Yippee.
Interestingly, experiments have shown that it takes me about four or five frames to get a good idea of what the hell I'm looking at. Three frames is not really enough. It doesn't leave the me enough time to process the information comfortably. These numbers were scientifically (yeah, right) determined by me sitting here and twinking with VirtualDub to find out what I am comfortable with in terms of getting the information vs. not-getting-the-information. I did look on the internet, but the internet was singularly unhelpful on this and I wanted to know *now* and not three hours from now. Let's hear it for trial and error...
Anyway, I need four or five frames (~.20 seconds is five frames) to process visual data well enough to integrate it into a coherent story. Three frames (~.125 s.) and I can *see* but there's no time to *process*. Stuff gets lost. I think I'll need to use the bigger exposure if I want people to get the jokes the first time around. Betcha that once the iconic stuff has been set up, though, that I can flip through 'em faster, like in a refrain sort of way, because it'll be review. Mmm-hmm. I can kind of see how that would work.
It's quite possible that young kids today will not need as many frames as I do. I'm old and slow and don't spend enough time watching video games... but I *still* read faster and process jokes faster than normal kids (judging from the crowd reactions at the fan parody track at *the con*), even when I'm dead tired and slightly drunk. So. I figure it'll all work out in the end.
I suspect that the exposure interval can be lessened for intercut scenes... scene A, scene B, scene A, scene B. That'd probably be (a) interesting plot-wise and (b) workable for shorter intervals with more visual noise. Yep. I'll have to try that next. Or, you know, intercutting actual scenery with black. I bet there's a way to generate plain black frames with the tools I have... people need plain black for leader spaces and fading-to-black and stuff. Then I'd have plain black to intercut with video. Too much of that would be not-good but it'd be interesting to play with in very small amounts. Might make a nice beginning, ending thing.
(time passes)
I am now able to make plain black frames that go nicely with my existing .avi files. This is a non-trivial undertaking because the documentation here is relatively minimal. The web site for AviSynth at SourceForge.net says:
# produces a black clip (3000 frames) with the remaining clip properties of the avi:
AviSource("E:\pdwork\DO-Heaven.AVI")
BlankClip(length=3000, color=$000000)
The web site is not exactly what I'd call correct. That does NOT do it. Extensive trials show that the script above produces a black clip of 3000 frames at 24 fps. Now, VirtualDub is not the world's most brilliant program but it's not stupid about fps's... It can damn well tell if you're trying to sneak two slightly-different clips past it and mash them together. Simply typing in 23.976 for the frame rate in the AviSynth script is not going to fool VirtualDub. You have to match your black frames to the original source, you poor sap, and the syntax in the example is not going to help you at all nor is manually setting the framerate. It has to be matched to the original, not typed in by your big, fat fingers. So, you know, blow a couple of hours one morning trying to figure out the problem... or read on for the solution.
The following does it:
BlankClip(AviSource("c:\d2vfile\sku2.avi"), length=30, color=$000000)
