Monday, March 5, 2007

Lowlights of Making a Highlight Video

Wow, I'm about to pop my blogging cherry. And you're all here to witness it. That's actually kind of disturbing.

This won't be nearly as cool and well thought out as
"the boy's" recent scientic breakdown of the ramifications of Thames' injury, because I wouldn't be capable of that kind of thought even if you spotted me the topic and gave me 10 years to complete it. I'd eventually drool on the keyboard enough to short the whole thing out and have to start over.

Instead, I'm simply going to talk about the only real contribution I've ever actually made to the Mizzou community at large, and that's the 2006 highlight videos. It started out as a "let's see if I can actually do this" act one night in early September because I couldn't sleep and needed something to do, and it has become the thing that gets me 40+ TBIM's in a week either praising my skills or bashing my choice of music, and each one requesting I put the game on DVD and send it to them. I only actually did that one time, btw, and since it took more than 10 minutes so I decided it wasn't worth doing ever again (note: it took a lot more than 10 minutes, actually. Getting the entire game onto DVD, with editing out all commercials and even editing out halftime, eventually took almost 10 hours of actual computer time. 8 of that was just the computer shrinking the video down to a quality that could fit on a DVD and then writing the files, but still.)

These videos were a labor of love all season, but in the end it was more labor than love. The only one that came together perfectly and nearly effortlessly was the Texas Tech game, which ironically enough is basically how you can describe Mizzou's results that day. But the Ole Miss game, the first one I did this year, took me almost a week because I was still learning what I was doing. The
New Mexico game became proof to me that even though Limp Bizkit is a multi-platinum artist, nobody actually likes their work. Oh, and that one also was what finally proved that I have an intent to harm children on the internet by using a song that has profanity bleeped out in it. The Colorado game reminded me that people still cannot accept rap music, despite it being 2006. A lot of people questioned why I even bothered to make a highlight reel of the aTm game seeing as how we lost (Aggie fans were plentiful to point that out). The KSU game wasn't televised live so I only had the replay show to go from, and when it was broadcast in KC they didn't show the 1st quarter and instead showed the 4th quarter twice. And the ku game was the day my DVR decided to only record from 11:00 to 11:30 that day so I had to beg, borrow and steal to get MU46 to bring me a copy of the game on VHS. And people wondered why it took me 2 months to get the bowl game video out?

Let me describe the process for you, if you care at all. First and foremost, the game has to be recorded. The DVR performed well for me all year other than the ku game, so we'll give it a 9-1 record in the 10 games it recorded (I recorded the games against Oklahoma and Nebraska, but I can't polish those turds). So I watch the game live while it's recording (about 3 hours for a normal game), then I have to hook up the DVR to my PC through my USB Capture device and watch the game again while recording it with the video capture software. I record it quarter by quarter so I have smaller files to work with, and I fast forward through the commercials to speed up the game. Still, this is another 2 hours minimum and usually ends up being closer to the original 3 because after each quarter I have to go over to my PC and stop the capture, then let it go through and detect scenes of the file I just captured. What that means is it will try to detect where the on-screen action drastically changes to create markers within the video so I have one massive file of the quarter, but I have about 300 different segments of that quarter so when I want just a certain 10 seconds of video I don't have to clip 25 minutes out to get it. Honestly, without that I don't see how any of this would ever be possible. This scene detection takes about 3-5 minutes per quarter. So then the game is over, and while I'm watching it again as it records I'm usually making mental notes about what plays will and will not be worth viewing. I'm also trying to figure out a general tone for the game to give me an idea what kind of music will be appropriate, but I still don't necessarily know what song I'll be using because I have to make sure there's enough video to cover the songs and enough music to cover the video.

Now that it's in the PC, and I've spent at least 6 hours already on this without even starting the editing process. I typically try to come up with something for the intro before I even look at the 1st quarter, and it's usually something simple like showing the chick from FSN talking about how Mizzou has a chance to go 5-0 for the first time since 1981 or maybe it's the announcers talking about how Ole Miss' offense is going to be so powerful this year because of the ties to the Miami Hurricanes, but I had such grand plans for the ku game with including clips from "Ride With the Devil".

TANGENT: Do you have any idea how difficult that actually was? You'd think it would be easy to just record video off of a DVD, but you'd be wrong. I figured I could play the DVD on my TV, just like I do the football game, and record it onto my PC through the exact same capture procedure. No, turns out you can't do that. The DVD or the player or something is smart enough to know you shouldn't be able to do that, so it safeguards it from being done. The video and audio continuously fade in and out, so it's worthless. So I find a program that extracts the .vob files from the DVD and that works great, but the video editing software doesn't understand what those files are so they're useless to me. I finally find this program that is $15 or something that will extract the video and audio off of a DVD into a .avi file and decide it's worth it so I buy that and have it extract all of the chapters off of RWTD. This isn't as easy as I thought, because there's more than just the movie on the disc so I have to know which "Title" I'm wanting to pull off of there otherwise I may get the Director's Commentary or the previews or the extras or some other rubbish. I finally get that thing going, and I start going over the .avi files to find the scenes I want and I notice that the audio and video are not synchronized. They're actually about a full 15 frames off, which is about half a second. And it's quite noticeable. So to correct for this, I had to drop the video I wanted into the video portion of the editor and mute the audio, then drop the same video clip onto the sound effects layer of the editor and adjust it just right so that the audio actually synched up with the video and then link them together so that when I clipped them to put shots of the crowd or the team into the middle it would take out the video but leave the audio there. All of that for about 65 seconds of video at the beginning of the highlights. But, that's not what I'm here to bitch about. END TANGENT

Back on task...I've figured out what I'm doing for the intro and I've got the Marching Mizzou CD to choose music from so this part is easy and done. I watch the preview back in full screen, it looks good, so I move on to the game. I load up the game's 1st Quarter video file and it gives me the 18 pages of scenes it has detected. There's 18 scenes on each page, and normally each play is contained within a single scene and then if there was an instant replay of that play it will be in another scene, and sometimes 2 scenes. Random crowd shots or cheerleader shots also get their own scenes. Again, I can't tell you how important it is that the software does this for me because otherwise I'd have to watch the game again and just mark the tape, so to speak, for every play that I might want to use. If it wasn't for this, I could only do this as a full-time job. As it is, I spend enough time on it. So I pretty much watch the game at 4x speed, another necessity this software has built-in, and then as I get to a play I set it to normal speed and watch it. If it's on no consequence, I just hit the 4x button again and move on to the next play. If it is a good play, I drop that scene from the source file into the timeline of the movie. From here is where the real "editing" gets done. There's usually a good 2-6 seconds before the snap of the ball and then another 2-6 seconds after the tackle is made that don't need to be shown, so I have to trim the start and finish times of that clip to be where I want them to be. So you have to set the zoom level on the timeline to the right settings so that you can see multiple clips at once to keep an idea of where you're going but you also want to have frame-by-frame control over the clip so I've got it zoomed in so that the timeline basically looks like a ruler with every 10 seconds marked by a number and then every second in between has a tick mark. I adjust the clip to about where I want it to start and end and then go back to the source file to keep watching. In the case where it's a good play and got an instant replay or two from another camera angle, I have to decide whether or not I want to show the live play, the replay, both, or combine them. I didn't really get the idea to combine them until about halfway through the season, so it's pretty noticeable if you watch all of my videos in order where I actually started getting better at this. It makes for a really cool effect sometimes when you start off showing the play from the snap, then right as Daniel releases the ball you cut to the receiver he's throwing to so you can watch him catch it and make his move. Or maybe it's a nice run from Tony Temple so I show the play from the start with him getting the handoff and heading through the line, then as he makes his cut at the 2nd level it goes to the replay camera and shows him in a closeup from the front as he stiffarms a defender and moves to the sideline. There are even times where I'll start off showing the play live, then cut to the replay angle, then cut back to live. This is done by taking the scene, playing it through to the point where I'm going to cut to another angle and instead of trimming the shot to that point and starting again I actually split the clip there so it creates 2 scenes out of that one. Then I put the 2nd scene in between them, trim it to where it needs to go, then put the 2nd half of the first scene at the end.

And then it keeps on going. Total video time of a regular quarter of football, minus commercials, is about 35-45 minutes depending on how many scores there are or clock-stopping plays. Though I watch most of the game at 4x speed, the act of editing even 10 plays into the highlight reel from a certain quarter makes up for that. Each quarter, I'd guess, takes about 30 minutes or more to cut down because I want to get everything just right. So now we're looking at 2 more hours of work spent, and it doesn't have music yet. But during the re-watching of the game, and then watching it again as I cut plays down into highlights, I start to get a general feeling about the game; a tone of some kind that was set. For the Ole Miss game, it was sort of a coming out party for the whole team because we dominated from start to finish on both sides of the ball and the team just looked fast (in every direction). Talking to people at the tailgate post game, it seemed most people were saying things along the lines of "It's time to get on board with this team." So the first song I chose was "Last Train to Transcentral" by the KLF, as it was a very fast and upbeat song that talks about getting on board. Well, that song got us to about halftime. I needed something else that would fit with The KLF, but I wasn't about to use the song generally known as "Moo Moo Land". The music from the lobby scene in The Matrix fit well and kept the tempo nicely so I added it in. And after adding it in, I was about 45 seconds or more long on music with no video. I really didn't want to cut the song off if I could avoid it, so that's when I thought it might be a good idea to go back and put shots of the crowd and cheerleaders into the game to give it more of a "game" feel. After finding appropriate shots from the different quarters, putting them in the right place, and adding back in a second or two on certain plays pre-snap or post-tackle I got it to the point where the video and music were ending at around the same time. This process takes another half hour or more, because I'm going back over film I've already gone through to find crowd shots and they're not as obvious. As I got more practice at this, I started putting the crowd shots and cheerleader shots into the timeline as I went along because I knew I was going to need them and I could always cut or delete them to shorten the video.

So now I have rough draft. I'm 9 hours or so into the project (including watching the original game, so 6 hours post-game work here). I watch the video a couple of times to see where video and audio need to synch up better and then go back in and make minor adjustments to start and end times for certain clips. Most clips it doesn't matter at all if they match in any way, but there are certain times of the song where the video just works better if the action on the screen matches the music. Sometimes, it's absolutely perfect (like the Texas Tech game where big hits accompany "BOOM!" and that one was all planned out ahead of time) or like in the KSU game where during one of the choruses to "Jump Around" the players were jumping in unison to the words "Jump, jump, jump...". At the end of the ku video, the players' hands are pointing up to the beat of "Hey, hey, hey...goodbye". Was all of that necessary? No. But I think it's the added little subtlety like that kind of stuff that makes people appreciate it that much more. It's more than just turning on the camera and pressing play on a CD, so to speak. So depending on the length of the song, the number of audio cues that need to match up with video, etc., I'm looking at another 30+ minutes to get it where I want it. I save the project file so I can go back to it again, and now it's time to make the file.

For a video on YouTube, it has to be under 10 minutes in length and it has to be under 100MB in size. That puts a pretty big limit on the quality you can create the file to. I always made mine into .mp4 format, which is not only what YouTube suggests but it seems to get the best results for retained quality in audio and video while sacrificing file size. I always had to play around with the audio quality and the bps on the video to get it as close to 100MB without going over. I'm the Bob Barker of video editing, apparently. I've got a pretty decent computer as far as power, so compiling the video into this .mp4 would take another 15 minutes or so. Most of the time I walked away from my computer at this point because it doesn't need me and I could really use a break anyway. Afterall, we're over 10 hours I've spent on this one game. Once the file is done, I watch it again to make sure it's what I'm looking for and when I'm happy with it I upload it to YouTube. This takes awhile, because upload speeds on the internet are about 1/6 that of download speeds. Uploading the file normally takes 30-45 minutes, but again I can walk away from the PC while it's doing this. Once it's uploaded onto YouTube, a URL is generated for the movie but it still takes anywhere from 10 minutes to 1 hour for it to become active. I never knew how long it was going to take, I just basically had to keep clicking on the URL until it actually appeared. Once it appeared, I'd watch it again on YouTube to make sure it looks okay. Then, after all of that, roughly 12 hours of effort later, I could post the link on Tigerboard, PowerMizzou, etc.

All of that effort....for this:

One more to go, and that will be the end of season film. I can't even begin to fathom how long this one is going to take...because I'm going to have 10000+ scenes to pour over from 32 different files. I know what music I'm going to use already, but I basically have to go watch each game again almost in its entirety to get the clips I want and then put them in the right order since it won't necessarily be chronological or anything.

Labor of love?