anime capturing from vhs

[sepultribe]

hi. i have a question. what's more reliable? to capture video, encode it to realvideo9 (it has filters especially for anime) with full quality settings and then reencode it to divx/xvid? or to try and encode the whole thing to divx/xvid from the beggining... although with all the noise and stuff it gets crappy?

what should i do?

[jggimi]

MPEG-4 codecs are lossy. It is why the Capture Guide (http://www.doom9.org/capture/start.html) recommends capturing in a lossless codec.

Either leave them in RV9, or skip the RV9 step entirely. You will probably not be happy with the results re-encoding from RV9 to another MPEG-4 codec.

[2ZOD.COM]

I would guess you'r shooting for quality, so I would suggest capturing with a lossless codec such as Huffy UV, and then using a third party app to encode to MPEG-4 (divx) the right way, possibly with multiple passes.

[video]

i recommend not using any codec what encodes to not intra-only. This is, bacause you would like edit the video you captured. Depending on the computation power, you may not able to apply noise reduction/colour correction filters real time.
- If you have a large amount of free disc space, you may apply a lossless codec, such as huffuv or snow(with a specific setup)
- If you have less space, you may want to use a lossy codec with compression ratio no more than 1:5. Such as
-picvideo mjpeg
-DV
-mpeg2 with I frames only

If you prefer editing in a commercial suite such as adobe premiere or avid xpress I rcommend using a DV codec
If your capture card supports direct capture to mpeg2 then that's the best choice if it can be setup I frames only and above 10Mbps

The most important thing. If the material on the tape is interlaced, then you should use a codec what support interlaced material - it's, important.

[sepultribe]

no... u got me all wrong. my friend and i know that encoding realtime sucks. we want to capture the videos on the hdd first using no compression or a lossless codec like huffuv u said. the problem is afterwards... how should we encode the video afterwards? using real video or divx/xvid. He told me that he tried once to get it to divx and it was very crappy. he said that divx cant handle videos with noise (like vhs captures), it gets the picture full of pixels and artifacts... so he told me if i could find a way to get it nice. he tried realvideo that contains filters especially for anime and it worked really good. he used maximum quality settings... so now we want to convert that quite big video to divx super quality but small size avi's to redistribute them to other people also on the net...

anybody knows any good solutions?

thank u

[video]

Originally posted by sepultribe
no... u got me all wrong. my friend and i know that encoding realtime sucks. we want to capture the videos on the hdd first using no compression or a lossless codec like huffuv u said. the problem is afterwards... how should we encode the video afterwards? using real video or divx/xvid. He told me that he tried once to get it to divx and it was very crappy. he said that divx cant handle videos with noise (like vhs captures), it gets the picture full of pixels and artifacts... so he told me if i could find a way to get it nice. he tried realvideo that contains filters especially for anime and it worked really good. he used maximum quality settings... so now we want to convert that quite big video to divx super quality but small size avi's to redistribute them to other people also on the net...

anybody knows any good solutions?

thank u

if the source is VHS then try capturing with resolution 352x576(480). after that transcode to mpeg2 with 2-pass VBR. You can find here hank315's great mpeg encoder HC-12 within the other mpeg-1/2 encoders sub forum. 4MBps 2-pass vbr will okay after noise reduction and color correction. if you capture the audio with 48kHz, and do AC3 stereo transcoding, You can put everything you have captured into a DVD.

within the analogue capturing guide, you will find good tips and tricks how to remove noise, and other artefacts.

[Boulder]

Capture at a high resolution but resize to 352x576/480 for the final clip. If you capture at lowish resolution, the capture card will do the scaling which usually leads to bad quality. Better do this step yourself.

With half-D1 resolution, 2-3Mbps is usually more than enough with any decent encoder and two passes or OPV.

[video]

Originally posted by Boulder
Capture at a high resolution but resize to 352x576/480 for the final clip. If you capture at lowish resolution, the capture card will do the scaling which usually leads to bad quality. Better do this step yourself.

With half-D1 resolution, 2-3Mbps is usually more than enough with any decent encoder and two passes or OPV. That's okay if your material is not interlaced, if it is so, it's a pain in the ass, to do resize later even if the source IVTC-able. IVTC filters are often failing on VHS grabed materilas.

[Boulder]

What is so hard in treating interlaced material? It's no different from progressive if you know what you're doing;) In fact, if you resize only horizontally from 720x480 to 352x480, all you need to do is SeparateFields-Resize-Weave.

[video]

Originally posted by Boulder
What is so hard in treating interlaced material? It's no different from progressive if you know what you're doing;) In fact, if you resize only horizontally from 720x480 to 352x480, all you need to do is SeparateFields-Resize-Weave.
ehh, in case of perfect timing. vhs grabs comes with dithering scnaline offsets (even 3-4 pels), sadly. this can cause a lot of ugly artefacts if you do resize.

[sepultribe]

this is all chinese to me... ;) can somebody just say things more simpler than that?

anyway i really thank all the people who posted here till now.

[Boulder]

Smart bob-resize/process-reinterlace. That's all I need to say. Search the Avisynth usage forum, there's a lot of posts about treating interlaced material.

@video: did it occur to you that the capture card samples at a high resolution and then scales down to the desired one? It does not sample at the resolution you specify in your capture program.

[ronnylov]

Originally posted by Boulder
What is so hard in treating interlaced material? It's no different from progressive if you know what you're doing;) In fact, if you resize only horizontally from 720x480 to 352x480, all you need to do is SeparateFields-Resize-Weave.

I don't think it is necessary to separate the fields for a horizontal resize. In the capture guide you can see that some capture cards does not have crappy scalers and then you can capture 352 width directly.

[video]

Originally posted by Boulder
@video: did it occur to you that the capture card samples at a high resolution and then scales down to the desired one? It does not sample at the resolution you specify in your capture program.

i've a bt8x8 card, esp miro pctv pro. the chip can be programmed for scan line length, and how many samples you want out from that scan line. jittery lines are coming out in case of high resolution grabs from vhs, such as 720 or 768, so i've moved away and i'm using canopus advc 50 now. it has a full frame tbc, perfect a-v sync, and outputs to firewire in dv format.
Originally posted by ronnylov
I don't think it is necessary to separate the fields for a horizontal resize. In the capture guide you can see that some capture cards does not have crappy scalers and then you can capture 352 width directly.

and happily broketree 8x8 devices capable doing that :) however VHS is too jittery for it anyway, i got always 2-3 frame drops within a minute, so i moved to canopus