what filters were used in .hack by elite-fansubs?

[d_dave]

I've just seen an encoding of .hack (anime) by www.elite-fansubs.com and I have to say that they did a superb job. I've never seen any encoding job so clean. Does anyone know what filters were used in the encoding process??? An AVS script would be nice.

[mf]

Fat chance :p. Anime encoders rarely give away their techniques.

[OUTPinged_]

Most probably no filters were used at all.

Maybe a temporalsoften or nomosmooth, but that is unlikely - they would use c3d or sshq instead.

Best encodes come from best sources.

[digitize]

Nah jedi didn't use any hybrid filters. I know him semi well, and during that time he used 2d cleaner (vdub version). But as outpinged said, best encodes come from the best sources, and e-f was able to get good raws for that. Also to consider is the fact that that show is widescreen, and much easier to get a much better encode at a file size of around 170mb.

[MasterYoshidino]

it also depends on the anime's amount of motion, some anime series like aoshi compress way better since they lack as much high motion scenes like evangelion or other mecha style anime,
and since the more void space there is (i.e. less detail in drawings) then the more that can be smoothed and compressed by spatial smoothers.

[OUTPinged_]

and during that time he used 2d cleaner (vdub version).

--;; You would never overestimate their choice of filters...

[digitize]

@outpinged
not too sure what you mean....
anyway, i know the guy, meaning i talk to him on a day to day basis, i know what he used.

[JediNight]

*blinks* I recently started visiting the Doom9 boards again to brush up on what's new and noticed this topic :) I was the encoder of .hack//Sign for Elite-Fansubs as well as Vandread 2nd Stage and various other episodes.

For .hack I would have used either optimized 2d cleaner or SmartSmoother HiQ. I didn't find a significant difference in quality between the two so I used 2d cleaner the most for the speed. Then of course Warpsharp was used as well. That's pretty much the standard filters most people used I think?
Don't underestimate finding a good raw though. I spent many hours scouring around for good raws to use. The better the raw the smaller you can make a file without the quality going to heck. Really noisy sources start falling apart badly under 150-160mb or so.
Also your Nandub settings can have a big impact over the compressability and size of your ending file. A well "tuned" Nandub should be within +-2mb of the predicted filesize otherwise something isn't quite right with your settings. As digitize also mentioned, hack was from widescreen (16:9) raws so you are only using 75% the bits as a 640x480 encode. (I wish there were more WS raws, I prefer them because of the bit-saving and they are nicer to watch imho)

The new Avisynth 2.5 is the next wave in encoding improvement to be sure. I'm seeing some impressive results that I believe are from it. (Live-Evil Wolf's Rain and TW's Naruto) I'm not sure how AVS/Nandub compares to AVS/Xvid, but once we see an RC1 release version I think Xvid will be the way to go -- if not better quality, than the fact that it is truly freeware. (The speed gains from YV12 encoding are impressive, YV12 + Xvid should be even better)

[sungey]

yeah YV12->xvid encoding is truly fast ... back in the days of nandub, i had to use Huffyuv to speed up encoding and reencoding process, it was like 1 hours for each pass (huffyuv->nandub)... with yv12->xvid i can even resize with bicubic/lanczos and apply filters i normally use and still only need 70 min for each pass ....and no need for reencoding (no subblocks etc) and no colorspace conversion too.