Interlace artifacts in Ciao

[silver_cpu]

Hi, just made an encoding of the first Serial Experiments Lain DVD. It, like many new anime DVDs, contains interlacing. When I disabled all field operations in GK and enabled the interlacing option in Xvid (and made absolutely sure that both passes were set to the same settings), I encoded the video. However, it still contains all of the old interlacing artifacts. Does the "encode with interlacing" feature work yet? I'm using the "cartoon mode" option, as well since this series contains a rather unusual amount of still frames and large single-color areas (rather like Chobits, if you've ever seen it). If anyone would like more detailed info about what I'm doing, just let me know. In the mean time, I'm going to continue to experiment and see if I can find what's going wrong.

[scharfis_brain]

interlaced XVid is just a special encoding mode, that preserves the interlaced structure of your video.
This means all interlacing and combing remains in the video!

[silver_cpu]

Pfbbbtttt!!!!!

I could retain the artifacts without a special feature, lol.

But in all seriousness, what is the use in this feature? Is it so that you can preserve the interlacing, and then use realtime deinterlacing with a supporting decoder? Is it for playback on TVs?

[aaar9800]

compression is better that way, since the codec doesn't have all these stairs on the edges to code. It separates the 24fps in 48fps with half vertical resolution. The motion is much easier on the codec that way. Then all these fields are weaved together on playback.

Correct if I am wrong, or if this explanation is too simplistic

[silver_cpu]

So in other words, it's best to enable this option in order to not only acclerate the encoder but also to preserve the original video stream (rather than de-interlacing and possibly blending)?

[Koepi]

you could put it that way. But for pc monitors this might be not desired, this is certainly useful for TV output.

Regards
Koepi

[Stigma]

Yep, in short, this switch is only for letting the encoder know what methods it should be using in order to get maximum compression. It wont change your video in any way, just compress it differently to accomodate the interlacing-lines.

I dont recommend you use this unless you encode the whole thing as interlaced (for example if you intend to use it for TV output). For example, enabeling it for an anime that has a ot of interlacing artifats after you used decomb(post=0) on it, then it would do more harm than good for compression optimisation since _most_ of the video would be progressive.

Hope that helps
-Stigma

[manono]

Why not just IVTC Lain and be done with it?

[Stigma]

Because many many anime cant be inverse-telecined properly, they leave artifcts due to mad mastering. THe poster thought he could ix that by just encodig it as interlaced, but the feature dosnt work the way he thought.

-Stigma

[Mango Madness]

there are some interesting developments from Bilu and MF in the avisynth forums concerning hard anime deinterlacing that you should probably check out.

[silver_cpu]

Really? Maybe I'll have to give those threads a check, I've been looking for a suitable way to encode anime & other mixed media for some time (as have many others).

Btw, is there a software decoder for .avi files that will de-interlace in real time? In other words, if I enable interlacing on the encoding, will I preserve all of the original fields, even if the source material contains both progressive and interlaced material? Take for instance STNG (well known problem). If I encoded it and preserved the interlacing, would I be able to play it back normally, or would this cause problems with the progressive portions?

[Stigma]

FDDshow can deinterlace on the fly, so that would work fine. However, you still face exactly the same problems of course, because FFDshow dosnt have any more "magic" methods to deinterlace than Vdub or avisynth. It does fieldblends, interpolation, and some simple bobs.

Basicly, its all a question about WHEN you want to deinterlace it i suppose. The result will be much the same, exept you have more options when deinterlacing before encoding, you have better tools for the job, and you wont require lots of CPU power to let FFDshow deinterlace it on the fly.

-Stigma