Metamorphoses

The cropping of this film continues to be all over the place…

The trailer on the Japanese DVD has some 2.35:1 shots squeezed into a 1.33:1 frame, while an English trailer on Youtube appears to be entirely widescreen. That English trailer further reveals how much even the French VHS is cropped, although it does also help me correct the French version’s frame distortion.

Changing gears a bit while my VCR is out of commission, I’ve started matching the Youtube version of the film to my own captures. Both versions were by the same distributor, and yet they are cropped a bit different. In some parts of the film, the frame can be expanded by as much as 20%!

It’s just a shame that the quality of the Youtube version is so low… I’m tempted to reach out the person who’d posted it, to ask if they still have that tape and would be willing to sell it to me.

Last night, I managed to get my VCR working again. Cleaning its mode switch with a cotton swab and some rubbing alcohol did the trick, After which I was able to get 3 more captures of Metamorphoses before turning in for the night.

I’d hoped to capture the tape a couple more times this morning, before work, but the tracking error was back. Back apart the VCR went, I bent the copper tabs back up, and cleaned the drum again for good measure. Once it was put back together, the tape’s picture was clear again. For about 3 minutes, anyway.

At this point the best fix might be to solder a wire between those tabs and the chassis.

Working on cutting together a full version of the luminance for Metamorphoses… I have 5 captures so far, but will need more. Lining them up with one of the full-color captures, at some points there are small gaps that will need to be filled in.

I’ve also found a lot of dropped frames in at least one of those captures, which complicates things further. Just one dropped frame will cause a capture to lose sync with another, and in some cases whole seconds were lost while recording. It got me wondering about just how complete my full-color captures are.

Fortunately, comparing the color captures didn’t take long, thanks to the “difference” blending mode in After Effects. If two frames are the same, they’ll appear black. Any differences will show in white. Out of 6 full-color captures, three appear to have no dropped frames

My VCR continues to be quite finicky… It seems like every time I go to use it, I have to take it apart to bend the grounding tabs back into place.

I wanted to get another capture of the tape this morning before work, but fixing the grounding would have been too much of a hassle. After work, I adjusted those tabs, but the VCR back together… and now it seems like the motor that tape’s reels isn’t working. This might have to be a weekend problem.

I wanted to do an experiment.. With 6 copies of this film stacked together, I’m running into diminishing returns in terms of noise reduction. The only way I might get better results, is if the color and the luminance can be separated from one another entirely. That would be really easy if this were captured with my Domesday Duplicator, but for now I need to find an alternative approach. And I might just have one, but it’ll take a bit of effort…

If I force my VCR to play back the tape as PAL, I get just the luminance, which looks quite a bit cleaner than the full-color image.

Subtracting the luma version from the full-color version isolates the chroma, so I can get a clearer view of just how noisey each channel is –

Blurring the color and then adding it back on top of the luma kind of works. The colors are more muted than before, even without the blurring. Adjusting the levels of the color layer increases the saturation, but nothing gets it quite as rich as the original…

Another example, using the faeries from the other day. One trick I can do once the colors are separated, is shift the individual colors around to try to minimize some of the bleed –

One downside to this whole thing is that I’ll still need to capture the movie multiple times, because in PAL mode the luma goes from clear to useless and back again roughly every 3 seconds. When it’s clear or not doesn’t seem to be tied to specific times on the tape, so I may be able to get the whole thing by capturing it several more times.

I swear, EVERY step towards digitizing the French version of Winds/Metamorphoses uncovers more problems.

I’d purchased yet another VCR, this one fully serviced/repaired, and with the ability to play back SECAM tapes. When it arrived, I tested it with a home movie recorded in NTSC… which the VCR detected as PAL. Since the tape was playing at the wrong speed, the audio was slow, and the video was scrambled. The input format can be changed manually, but doing so requires a remote, which was not included.

Testing the VCR with the French tapes went a little easier – they were detected as SECAM, but the tracking kept going in and out. A quick search online turned up a post by someone having a similar issue, with one of the responses claiming it was a grounding issue, and explaining how to fix it. I was nervous about opening up a $300 VCR that I could otherwise return for a refund, but I REALLY wanted to finally be able to capture these tapes.

After following those instructions, I put the VCR back together… and now it wouldn’t accept a tape. After taking it apart and putting it back together a few more times, it finally worked again. I’m not sure what I did to fix it, but at least it IS fixed – the tracking problem is gone, and the NTSC tape is now detected as NTSC. Even the duration counter – another thing that wasn’t working before – was now fixed!

Since VHS is such a lossy format, one of my goals is to capture each of these tapes several times, and then stack those captures to improve the image quality. Capturing the first tape went smoothly, the first time, but when I tried capturing it again, later, the VCR spat the tape out. On closer inspection, the tape had broken when it was rewound all the way to the beginning. I’m certain it can be repaired, but this was perhaps the most unexpected new problem that could pop up.

Fixing that tape may in fact be a non-issue. After capturing the second tape, and comparing the captures from both, the quality of the first tape is considerably lower. Not only that, but the cropping is the same between them.

The broken tape also being useless, quality-wise, and also kind of fitting. I’d purchased both of the tapes I could find specifically in case either one of them ended up being a dud. Now one of them is, twice over. But it also means I’d essentially spent $150 for a tape that is useless. Or $300 for the other tape. I’m probably better off just nothing doing the math on what this project has cost me so far…