FLCL DVD Ripping

[O Sliggity Slice]

I'm using some footage and what I have is really lossy, pixelated, and sloppy looking for what should be some of the cleanest looking anime around. I'm using huffyuv, but not futzing with any of the configuration. Also, it looks fine on the WMM preview screen, but like shit when I export it. The unfinished product is always saved in wmv format to save space. Will saving in DV-AVI make any difference? If not, what will?

[Qyot27]

I'm using some footage and what I have is really lossy, pixelated, and sloppy looking for what should be some of the cleanest looking anime around. I'm using huffyuv, but not futzing with any of the configuration. Also, it looks fine on the WMM preview screen, but like shit when I export it. The unfinished product is always saved in wmv format to save space. Will saving in DV-AVI make any difference? If not, what will?

That's why it looks that way. Check your WMV settings; it's likely the problem lies there. And what do you mean by 'unfinished product'?

DV-AVI will make a ton of difference in quality, and will allow you to make a higher-quality XviD encode of the final video than if you were to try to convert from WMV to XviD. The drawback is that DV-AVI takes up nearly as much space as Uncompressed video, IIRC. But to retain the quality, that's the only thing I can think of that you can do.

[Scintilla]

The drawback is that DV-AVI takes up nearly as much space as Uncompressed video, IIRC.

IIRC, DV doesn't even take up as much space as HuffYUV -- it's big, yes, but not as bad as all that.

[rose4emily]

The DV to HuffYUV disk usage ratio depends a lot on the footage.

Think of it as something like the filesize ratios of encoding an image in JPEG vs. PNG.

Usually the JPEG is smaller. Much smaller. This is because most images, especially those from "natural" sources such as photographs, contain a lot of detail and variation of sample values within the picture. There is also a significant correlation between the value of one pixel and the next in most "natural" pictures.

PNG, on the other hand, sometimes produces the smaller file. This is usually the case for diagrams, screenshots, and other such "synthetic" graphics. These graphics have nearly all of their pixel values on a few tight bands of their histograms and often have very hard edges that don't compress very gracefully through the DCT.

DV, like JPEG, uses the DCT to decorrolate image data before pushing it through a lossy quantization step and a final round of entrophy encodng. HuffYUV just does the entrophy encoding - much like PNG, but without PNG's more advanced (read "computationally expensive") value prediction model. Storing a video losslessly as a PNG stream, by the way, often offers almost a twofold improvement over storing the same video as a HuffYUV file. Unfortunately, you can kiss realtime playback goodbye if you do that.

Live-action footage unquestionably compresses better with DV than HuffYUV. Like photographs, live-action films contain a lot of spatially correlated but highly varied data. Super-low-budget animation, on the other hand sometimes compresses even better in HuffYUV than DV, thanks to it's being composed entirely of large blocks of identical colors. This is, of course, assuming a pristine source, which you rarely have (even a DVD is pretty lossy by editing standards).

Anime is in a fuzzy area where DV usually offers a somewhat lower bitrate than HuffYUV, but not the kind of savings you'd see with live-action footage. Still, with DV you know how much space a file is going to take up, and can usually edit it in real-time due to the simplicity of the codec and efficiency of most of its implementations. That, alone, is a pretty big advantage, even if there were no real space savings.

As to the issue of loss, encoding to DV is lossy, but most DV editing processes are lossless and the degree of loss is, once again, much less than that incurred by encoding the video stream to MPEG2 for distribution on DVD.

[Qyot27]

The drawback is that DV-AVI takes up nearly as much space as Uncompressed video, IIRC.

IIRC, DV doesn't even take up as much space as HuffYUV -- it's big, yes, but not as bad as all that.

I've never actually used DV for my own videos, I probably should have said that as a disclaimer. What I was basing it on was the bitrate of the video that my school recorded using its mini-DV camcorders. For 720x480, it was about 25 or 26 MB/s, whereas Uncompressed is something like 30 MB/s, I believe. HuffYUV, even on RGB mode, rarely went over 10 MB/s on the same footage. Maybe it works differently if it's not being recorded via camcorder.