JOURNAL: Kai Stromler (Kai Stromler)

  • with no one by my side 2015-05-23 17:57:42
    SH127:
    - Source rip: complete
    - Music: complete
    - Precleaning: complete
    - Storyboard/planning: done, more or less
    - Clipping: complete
    - Edit: 5:03/5:28
    - Postproc: none

    Once dinner finishes baking, I'll be able to get back to it; a full maximum effort last night got me through the last wall hit, and I was able to pick up this morning and grind on for another 25 seconds. There's technically another 25 left, but realistically only about 12 that are going to need close editing before the song starts to trail out. In the process I bled over onto a third cut-tracking sheet, making this the most paper-intensive video since SH111. There's currently 370 cuts in, and I'm projecting to finish around 400 -- not a record, but still a pain and a half to keep track of and battle through. I've already spent more time in edit than in cutting (a record, at least lately), and am really, really ready for this video to be done. It's looking pretty good, but it's been looking pretty good for a month; I want it to look *finished*, and *released*, and *cleared off*, so I can work on things that will be slightly less demanding.

    --Kai out

     
  • winter's wet, summer's hot 2015-05-22 10:03:01
    SH127:
    - Source rip: complete
    - Music: complete
    - Precleaning: complete
    - Storyboard/planning: done, as far as it goes
    - Clipping: complete
    - Edit: 3:45/5:28
    - Postproc: none

    Amazingly, I got through all of my errands, an intense vocab lesson, the traditionally biggest day of the week in terms of new shows dropping, and catching up with the server upgrade discussion here in time to grind through another fifteen seconds...which albeit ended up needing a little tweaking after I watched the nightly build this morning. Tonight will be tough again on all of these axes, but I should be able to get a little more work in and then hit into the long weekend to power through the remainder.

    Process-wise, the big lesson coming out of this video is that check builds are essential; in-editor, you're dialed in on a frame of ten seconds at a time, with limited visibility to where those ten seconds came from and only the most nebulous idea of where those ten seconds are going. I *have* to do them here because I can't really see what I'm doing otherwise, but less-complex and/or standard-definition videos will benefit from more checks as well. The smaller lesson is that I have really reached the useful limits of "spc" as a quality metric.

    Ever since the beginning, fourteen years ago, I've tracked stats on the length of video covered by the average component cut and though about spc (seconds per clip) as an indicator of pace, attention, source quality, whatever. Through the Flag of Hate, it became clear that there were edge cases that could make the metric look weird; on SH127, where I'm being guided by these check builds into cutting less hard and leaving more frames in per cut added, it's becoming clearer that the spc I'm capable of cutting at has dropped below that required by the music. On the one-measure-per-cut basis, I should be coming in about 1.1 to 1.2 spc (tempo is kind of like 110, maybe as slow as 100); between layering and effects that go in as timeline cuts, I'm on pace for about 0.9.

    What this means is that spc is a deeper and more complex metric than it appears on the surface. Spc is about source, and passes in information about what compromises need to be made with the source pool to bring a video from idea to instantiation. If the source needs to be smashed around with triggers, razored small to make up for lack of in-frame motion, or overdriven with effects, spc will drop towards zero. If the source pool is infinitely rich, with an oversupply of suitable source for any point in the video to the point that it can be assembled exactly as the music requires, without the need for bridging or kludging, spc will converge on the smallest practical musical phrase in the source...usually a four-beat measure. This makes it less useful, immediately, as a measure of improvement, but more useful *within* the process: if you have a watch with a sweep second hand and know how to count, you can determine while timing out the song what your spc *ought* to be -- on a per-section basis in case of any tempo changes -- and then use that in the editing process as a fairly *objective* measure of whether you need to cut tighter or let the video breathe more. And if you *can't* push in the direction that the metric says you need to, the problem goes back to the source pool, and should force a re-examination of that part of the process. If the footage is there, you need to cut better in the collection process/do a recut; if it isn't, the idea probably isn't so good and probably ought to be junked or significantly reworked.

    I'm still going to track and report this metric despite the change; by luck more than anything else, all of the numbers I decided to track on process when I started in this hobby almost a decade and a half ago are still varying degrees of useful, and I fully expect that I'm going to need the relatively raw data to build new advanced statistics out of, or to interpret in different ways, for as long as I continue doing this -- and with nearly thirty ideas on the board, not all of which will be cancelled before termination, that's not going to be for a while yet, unless I happen to get killed doing something stupid.

    --Kai out

     
  • no star in the east 2015-05-20 10:21:05
    SH127:
    - Source rip: complete
    - Music: complete
    - Precleaning: complete
    - Storyboard/planning: done, as far as it goes
    - Clipping: complete
    - Edit: 3:30/5:28
    - Postproc: none

    As anticipated, I got through the second wall hit, then finished up the backwards sections and got set up for what's probably going to be the most intense stretch in the video, just in time to not work on it at all for several days. I might get some progress in tonight, but between the housekeeping stuff I have to do and the final unit test of the intermediate language course, I'm not holding out a lot of hope. Edit will probably creep to completion over the weekend, video to drop early next week.

    The delays involved in getting this out, which are still not over yet, have gotten to the point where it's not realistic to look at doing two projects, or even one involving a 30-episode cut of dirty DIY VHS->DVD conversions that will need cleaned up, before that hard stop in four weeks. This has put kind of a wrench in the prod schedule; with three weeks, I can do one thing that has a short cut and a medium to long song, and then maybe the video idea I kicked around with the unused scraps of the first Kujaku-oh that were surplus to SH126's requirements. There are about five ideas on the slate that fit these requirements, but most of them are kind of weak and/or will need some major work that doesn't precisely come into scope. Fortunately, I'll have the time around burning the source materials for this one off to do the checks, but if I had better-shaped ideas for this limited amount of time, I could be doing lead-in work on a video in that time rather than fumbling around trying to decide what video to start working on.

    Sure, I could just start cutting for the VHS->DVD project, and then leave it open while I'm overseas, but the last thing I want to do is to come back from Japan with a pack full of niche boxsets and Macabre Mementos rarities, raring to attack new ideas from forgotten OVAs and drugged-out techslam, and have to get back to work on something that I'd left mostly-cut for two weeks of dodging bears and sleeping in bus shelters on the other side of the planet. That's not good for current or future projects: finish what you start, start something else if and when you come back alive.

    --Kai out

     
  • come and make me holy again 2015-05-19 09:54:57
    SH127:
    - Source rip: complete
    - Music: complete
    - Precleaning: complete
    - Storyboard/planning: done, as far as it goes
    - Clipping: complete
    - Edit: 3:11/5:28
    - Postproc: none

    I'm still not up to the wall hit, but it's lurking; hopefully, I can tie off the backwards section before I hit into it, but managing that transition over is going to take an uncomfortably large chunk of whatever time I happen to have for editing tonight. Being more than half done by song time is fine and all, but I've done only one out of the likely three cases of rip-out-and-rebuild that I'll need to do on this video; with that in mind, most of the hard work is still decidedly in front of me, and that's not even counting the fact that I'm largely out of lyrics at this point and the concept is going to have to hold the video up on its own the rest of the way.

    The good news is that doing check builds is going to make that hit a lot less painful when it comes; I've found and fixed at least one major issue per build since I started doing them, either technical or stylistic, so the added value is worth the fifteen or so minutes of doing nothing while I render an ever-increasing extent out and then sit through two to three minutes of something I've checked six times already. The bad news is that despite ten years working with this editor, I still do not have a concrete reason for *why* it happens, or what the "wall hit" behavior actually *is*. The exported uncompressed AVI, when it occurs and I have to rework it to restart, tends to be within spitting distance of a whole-gigabyte figure, so address space may be an issue. But more complex videos see them faster than simpler ones, so the number of cuts on the timeline, not just the uncompressed duration of video behind them, is a factor as well (and well that it is, or I wouldn't be able to restart after one) -- probably the dominant factor, since the hits come at the same place time-wise, generally, whether I'm working with SD or HD inputs. As long as I can recover from these hits when they happen, it's not worth it to take the time to spade out exactly what's going on with them; it's enough to know that they're there, about where they're likely to occur, and how to identify as soon as I get stuck and get kicking on recovery.

    --Kai out

     
  • beg the truth from strangers 2015-05-18 11:18:37
    SH127:
    - Source rip: complete
    - Music: complete
    - Precleaning: complete
    - Storyboard/planning: done, as far as it goes
    - Clipping: complete
    - Edit: 2:04/5:28
    - Postproc: none

    Keeping the hammer down, I pushed on past the wall hit later last night and should be able to get across the halfway mark, if not all the way to the next hit for tonight. Introducing no little difficulty into this is the fact that I can't accurately see what I'm doing in the editor, and need to cut check builds rather often to confirm that what I'm doing is actually working, and actually connects back to what went before. This is a pain, but I'm actually doing it, and it's had a good effect so far in terms of keeping the effects where I need them to be, and reducing the amount of cleanup mistakes that I'd otherwise need to to go back and fix on wall hits. It's going to take a while, but the work is its own reward; at about 40 seconds per hour, I won't finish tonight, but the couple hours that I'll be able to put in will show good results.

    --Kai out

     
Current server time: May 24, 2015 07:20:37