Mana Khemia 1 Chinese patch,video skiped during playback

[daniel229]

Never works before.
debug log (rename jpg to rar)

[unknownbrackets]

Skipped entirely, or does the video playback skip?

I see it getting the video data. I wonder if they patched the psmfplayer.prx file or anything...

-[Unknown]

[daniel229]

It plays a few seconds,and then skip the rest and going to the title.I think just patch the .pmf file,if replace the original one,it works fine.I compare these prx files with beyong compare,no difference.

[daniel229]

The other patch things are Japanese to Chinese.

[sum2012]

Update the log to v1.2.1-57-g3924d8b
https://gist.githubusercontent.com/sum2012/d671f413fbf098ea762f/raw/b112bf637f6115435ed840c3d7f4ed27b6fbc391/gistfile1.txt

@unknownbrackets
I try ffplay,it played video without sound,no missing video.
Can we use ffmpeg of HasReachedEnd function or HasReachedEnd has something wrong ?

[unknownbrackets]

Well, ultimately sceMpeg has some of the issues sceAtrac used to... although they're different.

sceMpeg just plays the packets you give it, really. Although it manages forward movement, you're welcome to loop a video by just feeding it packets from earlier in the stream. It will just play them.

So, "has reached end" is kind of a wrong concept to begin with. sceMpeg doesn't think about "end", it just knows when it has no more packets. If I gave it the first half of a video's packets, and then stopped giving it new packets... the video "ended". And if I loop the video, and keep giving it packets forever, it will never "end."

This issue may be similar to the Danball Senki issue - see #8464, e866d83, and #8437. Does it work if you put that change back?

-[Unknown]

[sum2012]

@unknownbrackets No,it is psmf

[sum2012]

Hmm so that mean that real psp don't read the header of length of the video file.

[sum2012]

@unknownbrackets
jpcsp can increase video time by this change
sum2012/jpcsp-1@48ca580
Maybe ppsspp also this change ?

[unknownbrackets]

So basically, you're saying that the length value in the PSMF header is ignored, and it should just use the filesize instead?

That sounds like something we should test.

-[Unknown]

[sum2012]

@unknownbrackets Yes,that is my meaning.
@daniel229 Use what software to patch the .pmf file ? I tried to use umdgen but crash in saving.

[daniel229]

Try the WQSG_UMD.

[sum2012]

@daniel229 Thanks,it work

[sum2012]

@unknownbrackets please also test "last timestamp value in the PSMF header is ignored"
JPCSP fix it in jpcsp/jpcsp@7e4a74d
and jpcsp/jpcsp@2c0d219

[unknownbrackets]

I used a short video with a header size (12 bytes into the file) of 0006A000. I tried 0000A000 and 0003A000. Both cutoff the video at different points and caused issues. It did not play the entire video.

I tried this with scePsmfPlayerSetPsmfCB, scePsmfPlayerSetPsmf, and scePsmfPlayerSetPsmfOffsetCB.

What header length does it have in this game, and what is the actual file size?

What versions of libpsmfplayer.prx and psmf.prx does it use? This is reported by PPSSPP in the log - you can also try giving me the crc of the files and I can compare with the ones I'm testing against. Maybe this only affects certain psmf versions.

-[Unknown]

[sum2012]

Accordering my earily log: https://gist.githubusercontent.com/sum2012/d671f413fbf098ea762f/raw/b112bf637f6115435ed840c3d7f4ed27b6fbc391/gistfile1.txt

scePsmf_library with version 0101, devkit 03070010
scePsmfP_library with version 0101, devkit 03070010
psmf.prx Size: 7,056 CRC-32: 089F0C2E
libpsmfplayer.prx Size: 35,520 CRC-32: BEE912BB

Chinese version header:

orignal version header(japanese)

[unknownbrackets]

Is the Japanese version 16863232 bytes (~16.08 MB)? How big is the Chinese version? I'm guessing it's larger than 5935104 bytes (~5.66 MB.)

-[Unknown]

[sum2012]

Japense Size: 16,863,232
Chinese Size: 17,039,360

[unknownbrackets]

Thanks - confirmed. That version of the library doesn't have scePsmfPlayerSetPsmfOffset or scePsmfPlayerSetPsmfCB, and also doesn't seem to respect the stream size. Let me check when it changed.

-[Unknown]

[unknownbrackets]

Okay, #8914 should give that behavior. I'm not sure about the last timestamp handling.

-[Unknown]

[daniel229]

Still skiped later with #8914 merged.

[sum2012]

@unknownbrackets we cannot port JPCSP of "last timestamp" fix
jpcsp/jpcsp@2c0d219 ?

[sum2012]

hmm , I has idea ,just not sure right
I am going to pull request in unknownbrackets:psmf-size branch