@cb4960  How come you rarely, if ever update your software?  j/k I saw the "v22" and thought I had misread the link. 22 updates... my hat is off to you good sir!

Interesting! Is there any way to encode to iphone friendly video format? I would love to see that in a future release!

cb4960 wrote:

Wow! It might be time to retire that Apple II . I hope it at least worked OK when it eventually finished.

Even at 1080 resolution, it shouldn't take more than 30 minutes on a somewhat modern computer.

Is the video file on a DVD or external hard drive or a network? If so, move the video file to your internal hard drive first.

If that's not the case, what codec does the video file use? What are your system specs?

The computer might be an older one (Athlon 64 X2 3800+ 2 GHz, 2 GB RAM), but with other files it was working fine, so...^^ When it finished, it was alright

The file is on my internal hard drive. Like I said, with Clannad it did run okay.

The thing is, I really don't understand anything about codecs and such, so I'll just go ahead and post the information MediaInfo shows me:
Video
Format                           : AVC
Format/Info                      : Advanced Video Codec
Format profile                   : High@L5.1
Format settings, CABAC           : Yes
Format settings, ReFrames        : 3 frames
Codec ID                         : V_MPEG4/ISO/AVC
Duration                         : 24mn 9s
Nominal bit rate                 : 193 Kbps
Width                            : 640 pixels
Height                           : 352 pixels
Display aspect ratio             : 16:9
Frame rate                       : 23.976 fps
Color space                      : YUV
Chroma subsampling               : 4:2:0
Bit depth                        : 8 bits
Scan type                        : Progressive
Bits/(Pixel*Frame)               : 0.036
Writing library                  : x264 core 60 r900 a9af942
Encoding settings                : cabac=1 / ref=3 / deblock=1:1:1 / analyse=0x3:0x133 / me=esa / subme=7 / brdo=1 / mixed_ref=1 / me_range=32 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / chroma_qp_offset=0 / threads=3 / nr=0 / decimate=0 / mbaff=0 / bframes=2 / b_pyramid=1 / b_adapt=1 / b_bias=0 / direct=3 / wpredb=1 / bime=1 / keyint=250 / keyint_min=25 / scenecut=40(pre) / rc=2pass / bitrate=193 / ratetol=1.0 / rceq='blurCplx^(1-qComp)' / qcomp=0.60 / qpmin=10 / qpmax=51 / qpstep=4 / cplxblur=20.0 / qblur=0.5 / ip_ratio=1.40 / pb_ratio=1.30 / aq=0

Hope that helps?

@Oniichan,
If you include .1 releases, it's more like 43 updates (but who's counting? ).

Cranks wrote:

Interesting! Is there any way to encode to iphone friendly video format? I would love to see that in a future release!

What are considered iPhone friendly video/audio/image formats? Thanks.



@rigol,
Thanks for the info. Probably nothing that I can do, but I'll look into it.

nest0r wrote:

Just noticed the speaker name doesn't matter, at least for Kage Shibari. Although perhaps in the future I can imagine using the filters to select lines by particular speakers. I like the idea of having ‘turns’, as well, for conversations. Seems like there are production/discursive possibilities I hadn't thought of...

Currently, subs2srs ignores the speaker. Shouldn't be hard to add it to the actors list though. I'll put it on the TODO list.

Last edited by cb4960 (2011 March 05, 9:37 am)

Probably Mpeg4\MP4. I tried .avi and had no luck. I'm not sure if you can play videos on the iphone anyway (in Anki, I mean).

I made a test build that (hopefully) outputs iPod/AnkiMobile compatible .mp4 video files.

Download subs2srs iPod Test 1 via MediaFire

Since I don't have an iPod myself, I'm asking for a volunteer to help test with AnkiMobile.

Thanks.

Last edited by cb4960 (2011 March 06, 9:46 pm)

nest0r wrote:

Can you write a program that makes walking several feet and clicking stuff easier?

No, but I can do the next best thing. Here is a 3 card sample iPod video test deck:

Download the iPod Video Test Deck 1 via MediaFire

Alright, 0.5 way there! (iPods sure seem picky though).

OK, let's try that again, this time with AAC:

Download the iPod Video Test Deck 2 via MediaFire

For my reference:

About AAC:
I couldn't find any pre-compiled Windows versions of ffmpeg that included libfaac. Some kind of legal issues. I used ffmpeg's internal AAC encoder. It's supposed to be in it's "experimental" stage, but it sounds fine to me (though I'm not one of those sick audiophiles).

This is what I'm passing to ffmpeg:
-y -i "INPUT_VIDEO.mkv" -ac 2 -map 0:0 -map 0:1 -vcodec libx264 -fpre "/path/to/ffmpeg/presets/libx264-superfast.ffpreset" -fpre "/path/to/ffmpeg/presets/libx264-ipod640.ffpreset" -s 368x240 -b 600k -strict experimental -acodec aac -ab 128k -ss 00:01:52.270 -t 00:24:15.920 -vf crop=368:240:0:0 -threads 0 "OUTPUT_VIDEO.mp4"

Last edited by cb4960 (2011 March 07, 10:48 pm)

Hooray! I'll try to finalize the new iPod-friendly version of subs2srs this weekend.

Lol, I started it, but I totally missed all this. Thanks Cb4960. +dude, you gotta get a iPod. It's seriously worth it!

nest0r wrote:

By the way, have you thought any more about integrating OCR with the vobsubs here? It could be a kind of secondary area (since there'd be occasional errors, users could have the usual vobsub expression field + the OCR'd one to compare). Seems like the possibility is there since there's a high level of consistency in terms of size and whatnot.

Anyway, just something to keep in mind. ;p

I was going to perform some more testing with Tesseract but never got around to it. I'll play with it a little more one of these days.

Edit: For now you can always use Capture2Text to help look up the occasional word that you don't know. I've been using it mostly for manga and visual novels. It's not perfect, but after a while you get a feel for what it will OCR correctly and what it won't. For example, for vertical text, it hates 言.

Last edited by cb4960 (2011 March 09, 7:44 pm)

As you've guessed, vobsub2text was the program I wrote to experiment with OCR'ing vobsubs.

nest0r wrote:

So do you feel there's a superior optimization? Do the colours of the original impact the new colours?

Right, the color selected by the user substitutes for the original vobsub color. That's one of the things I was experimenting with. I got sidetracked though and never established anything conclusive.

Another thing I could experiment with is thickness/boldness of the text (without doing a little research, I'm not exactly sure how I would do this easily - maybe ImageMagick has some filter I could use). The OCR tools seem to prefer thinner text over thicker text/more bold text.

nest0r wrote:

Also, how does the Scale option function? Is it resizing the images and the option lets you keep it within limits to prevent distortion?

The Scale option resizes the vobsub image before sending it to the OCR tool. A scale of 1.0 means use the original size. A scale of 2.0 means double the vobsub image size. For Capture2Text, I found that a scale of 4.0 worked best for normal-sized text. Vobsubs are usually fairly large, so a smaller scale might work better.

nest0r wrote:

Edit 3: Oh right, seeing Nukemarine's post in the above link, I see you can just use the Subs 2 slot for the secondary Japanese .srt. Although then if we have English subtitles they're still left out. But with text glossing that shouldn't be a huge deal. Or perhaps one can merge the secondary Japanese .srt and the English .srt as dueling subtitles, then use that as Subs 2.

If/when I ever integrate this into subs2srs, I would output both the images and the OCR'd text allowing the user to fallback to the image on occasions when the OCR messed up.

Currently, vobsub2text uses nhOCR as the OCR tool. As mentioned in my above post, I wanted to experiment with the Tesseract OCR tool. Both have their advantages as disadvantages. When using Capture2Text, I almost always use nhOCR because Tesseract won't even attempt to OCR unless the capture contain enough characters. Most lines from vobsubs are probably long enough though. If I find Tesseract to be generally superior, I can always fallback to nhOCR if the line is too short.

Also, I wonder how well Tesseract would work for English subtitles. If it can't even handle those with a great amount of accuracy, Japanese might be out of the question (though, for non-vobsub usage, I've found Tesseract's English OCR to be quite good - nhOCR is quite the opposite). OK enough rambling...

Last edited by cb4960 (2011 March 09, 10:35 pm)