Creating Fansubs - Questions and Answers

[jfs]

Even though you can put two audio streams into AVI I would strongly advise against it. Both because of the reasons already listed above, and then simply because it's outside people's expectations. It should be pretty well known that MKV is made to support multiple audio streams and people should know it can and know how to handle it -- and all MKV splitters should also handle it properly.

[Starks]

Originally Posted by mcpig

maceart-san: i was thinking about mkv files. but since others have suggested that avi might also work... perhaps i'll try both out -w-;

edogawaconan-san: thank you for the information! =)

sylf-san: thank you for the correction~ i'll change the topic now ^^
(edit... seems like u cant change the text in the actual topic... mah... anyway ^^

harukalover-san: thank you =) i'll pay attention to that when i make the file~

thefluff-san: we'll know when i try it out =w=

thank you all for the information! i will try them out after my finals... -w-;;;

What's up with all the wapanese and cute faces? It pisses us off.

[Harukalover]

Just some assurance in case I'm now thought to be an avi supporter and had recommended that as the best course to the topic creator.

can != should

[Zane]

I'm pretty new to all this but need some help:

I took a raw, seperated the audio and video, used AVSedit to insert subs into the video. I then loaded the AVS file onto VDM which was as follows:

avisource ("FILEPATH",audio=false).changeFPS(23.976)
converttoyv12()
textsub ("FILEPATH")

obviously the actual file was in the place where FILEPATH is.

upon loading the avs, I added the audio file, and then I used xvid and used the two pass. However, when I went to save as, first came a box which said:

Statsfile not found.

Then another box:

VirtualDub error

Cannot start video compression.
This operation is not supported
(error code -1)

Can anyone help me please?
Thankyou

[edogawaconan]

Have you done the first pass? The error means that it can't find statsfile required for doing 2nd pass.

[raziel666]

Sorry if this has been asked before, but is there some way to determine the compressibility of a given video file, so as to set appropriately the bitrate?

[edogawaconan]

Originally Posted by raziel666

Sorry if this has been asked before, but is there some way to determine the compressibility of a given video file, so as to set appropriately the bitrate?

I think I ever heard about it which involves virtualdubmod and some ancient tools
for me personally, I just try encoding it with constant quantizer and see the file size.

[jfs]

If you're using XviD you can tell it do not discard the first pass and optionally also have it do a full-quality first pass, if you want the result to be playable, that will effectively do a constant quant 2 encode of the video while collecting stats. You can use the size of the video file output by that first pass to gauge the compressability then.

[raziel666]

@jfs: Is it codec-dependent? I mean if I want to compress to x264, is it the same procedure?

[jfs]

All multi-pass things are dependent on the specific implementation, yes. I think that x264 also has something similar though, I believe it does a cq18 encode (approximately equivalent to XviD cq2) for its first pass.

[raziel666]

I did some tests with CQ18 the past few days, and I have some more questions, so please bare with me!
I tested it with three sources:
1) 720p WMV (bitrate @ 1600kbps)
2) a Blu-ray rip
3) a DVD-rip from R2 Jap DVD which had 2 episodes/DVD

I ran the CQ18 and the resulted bitrates were respectively like this:
1) 2151kbps
2) 15821kbps
3) 420kbps

I should mention here that all the tests were done without using any filters.

As for the first one, how is it possible that it gets a bitrate higher than the original? This one I cannot understand.

As for the second one, obviously the bitrate is too high. Probably it was around the original Blu-ray bitrate if I'm not mistaken from what I saw on DGIndex.

The third and the last one. I did an encode @ 1000kbps and checked various screenshots from both encodes and I couldn't find any difference. Then I remembered what TheFluff had done with Night Head Genesis from gg (that he had reduced the encode from 170MB to 80MB or something like this). Given that the encode was from a somewhat-action series, I still wonder if I didn't do something wrong.

As for the compressibility, given that this CQ18 encode gives a bitrate value of X. If X is too high (as in the first two cases), how much should it be decreased (or can it be decreased) to get a good result? Is there some way to determine this?
Moreover, this CQ18 pass produces some stats. If nothing changes in the profile and the avisynth file, can these stats be used to do the second pass of a "normal" encode or should I do a 2-pass from the beginning?

Thanks for anyone who answers and sorry if my questions sound a little silly / idiotic.

[TheFluff]

Originally Posted by raziel666

As for the first one, how is it possible that it gets a bitrate higher than the original? This one I cannot understand.

You're probably encoding some very nice compression artefacts at perfect quality. That, and the original encoder may have bitrate starved the original encode initially.

Originally Posted by raziel666

As for the second one, obviously the bitrate is too high. Probably it was around the original Blu-ray bitrate if I'm not mistaken from what I saw on DGIndex.

Well, HD does tend to take huge amounts of space.

Originally Posted by raziel666

The third and the last one. I did an encode @ 1000kbps and checked various screenshots from both encodes and I couldn't find any difference. Then I remembered what TheFluff had done with Night Head Genesis from gg (that he had reduced the encode from 170MB to 80MB or something like this). Given that the encode was from a somewhat-action series, I still wonder if I didn't do something wrong.

It's pretty random. Some shows compress insanely good for no easily discernible reason (not easily discernible that doesn't think like an MPEG compression algorithm and does DCT's in his head, that is), others don't. H.264 can be insanely efficient sometimes.

Originally Posted by raziel666

As for the compressibility, given that this CQ18 encode gives a bitrate value of X. If X is too high (as in the first two cases), how much should it be decreased (or can it be decreased) to get a good result? Is there some way to determine this?

Trial and error, pretty much. That and experience. It's not a fixed ratio, and additionally it's pretty subjective; some sources can be squeezed to higher quants than others and still look decent. Remember that quants are not an absolute perceived quality measure, nothing is.

Originally Posted by raziel666

Moreover, this CQ18 pass produces some stats. If nothing changes in the profile and the avisynth file, can these stats be used to do the second pass of a "normal" encode or should I do a 2-pass from the beginning?

You should be able to use the stats for a second pass with the same settings, yes.

[adrianmak]

I'm testing clip comment to create animate effect on karaoke
I want to use several clip comment to clip a karaoke and the clips will move outside the screen into the movie stage to build up the whole text

below is one of the clip. however I can't see any clipped text moving into the movie stage, another i did wrong ?

Dialogue: 0,0:01:43.35,0:01:50.64,*Default,,0000,0000,0000,,{\clip(0,439,640,449)}{\move(-300,200,200,200,0,150)}This is a demo text

[edogawaconan]

Originally Posted by adrianmak

I'm testing clip comment to create animate effect on karaoke
I want to use several clip comment to clip a karaoke and the clips will move outside the screen into the movie stage to build up the whole text

below is one of the clip. however I can't see any clipped text moving into the movie stage, another i did wrong ?

Dialogue: 0,0:01:43.35,0:01:50.64,*Default,,0000,0000,0000,,{\clip(0,439,640,449)}{\move(-300,200,200,200,0,150)}This is a demo text

because the clip starts at (0,439) and ends at (640,449) when the text moves from (-300,200) to (200,200) (AKA out of range). Try \clip(0,0,640,449) instead

[adrianmak]

Originally Posted by edogawaconan

because the clip starts at (0,439) and ends at (640,449) when the text moves from (-300,200) to (200,200) (AKA out of range). Try \clip(0,0,640,449) instead

This is the original sub


This is the clipped sub

Dialogue: 0,0:01:43.35,0:01:50.64,*Default,,0000,0000,0000,,{\clip(154,433,492,448)}紅牌侵略者Urere慘遇災難

Then I added the \move command

Dialogue: 0,0:01:43.35,0:01:50.64,*Default,,0000,0000,0000,,{\move(-300,433,154,433,0,200)}{\clip(154,433,492,448)}紅牌侵略者Urere慘遇災難

But the result is , I can saw some pixels is moving across the movie as shown below. The clipped sub isn't the same as the first image of this reply



Just my feeling of ass, the \clip , rotation and drawing command is the hardest part.

[edogawaconan]

are you sure the position is correct? (ie tried it without \clip)

[adrianmak]

Originally Posted by edogawaconan

are you sure the position is correct? (ie tried it without \clip)

Thanks giving this hint. The position is incorrect.

[adrianmak]

Every time I create the simplest karaoke effect like

{\k150}This{\k200}is {\k75}a{\k210}demo{\k250}

When karaoke started, the first syllabus can't show completely. How do I show the karaoke a little bit earlier than the vocal start ? Let say original vocal start and end time is
00:00:45.10 - 00:00 51.32 {\k150}This{\k200}is {\k75}a{\k210}demo{\k250}

Then a adjust the start time a bit earlier like
00:00:44.80 - 00:00 51.32 {\k150}This{\k200}is {\k75}a{\k210}demo{\k250}

The karaoke effect may not sync with the vocal.

[edogawaconan]

add {\k30} before {\k150}This.....
Like this
00:00:44.80 - 00:00 51.32 {\k30}{\k150}This{\k200}is {\k75}a{\k210}demo{\k250}

or change the k value to how cs you moved

[adrianmak]

Originally Posted by edogawaconan

add {\k30} before {\k150}This.....
Like this
00:00:44.80 - 00:00 51.32 {\k30}{\k150}This{\k200}is {\k75}a{\k210}demo{\k250}

or change the k value to how cs you moved

Normally, how much should I put the karaoke show up earlier than vocal start ? 0.5s ? 0.4s ? or 0.2s , etc