VCD Entry point accuracy problems.

[droopy928gts]

Hello all,

I've been searching your forum for a solution to a problem I've experienced with entrypoints, but could not find any solution. Me last chance is if any of you people have any idea where to look for the solution to this problem.

First let me tell you though what I use and think to be correct. Correct me if I'm wrong

-For entry points closed GOP's are needed??
-For entry points sequence header must be before every closed GOP??
-Entry points can only be established on an I-frame??
-Entry points can be set to EVERY I-frame of a closed GOP with sequence header??

TMPGEnc 2.54

The problem is following.

I've encoded in TMPGEnc a movie I've recorded from TV through a capture card. In GOP settings I selected "Create bitstream for editing" and the "Sequence header output interval" = 1. "Detect scene changes" is also activated. GOP sequence is default, being "IBBPBBPBBPBBPBBP". I use VCDImager to make my VCD images. I set the entrypoints to a certain I-frame in the encoded MPEG1 file.

The issue is that very often, the I-frame selected by me to be the entry point is not accepted and VCDimager then warns me and tells me it shifted to the nearest possible entry point. This can on occassion be about 3 seconds or more earlier or later. I checked the frames using VirtualDub, and GOP sequence before and after the selected I-frame, as well as the GOP from the selected I-frame as entry point are closed GOP's.

First I thought it to be my limited experience with XML, but when trying VCDEasy I noticed the same thing occurred when I entered either frames or time for chapters entry points.

Since I selected 1 sequence header before every GOP, this should work shouldn't it???

Any assistance would be greatly appreciated.

Thanks and regards, Leon

[droopy928gts]

Just trying, seems to have become snowed in!

Anybody out there that has any idea, Please?

Thanks, Leon.

[disturbed1]

Grab Mpeg Sequence Maker from the tools section, and run you mpv stream through.

TMPG doesn't put sequence headers (time codes) on every GOP, that'll fix it.

To view your possible entry points download Philip's Entry Point Assist (EPASSIST.ZIP)

[droopy928gts]

Thanks Disturbed1,

I had already tried that before. Used TMPGEnc to DeMultiplex, then MPEG Sequence Maker after which ReMultiplexed. When done so, my MPEG1 stream was terribly blocked. I just might retry using another De- / Re- multiplexer though.

Also thanks for the tip on the EPAssist program. Never heard of it, but am defintely gonna try it. I know it can't solve me problem, but at least I can check for the sequence headers with it.

Strange though that TMPGEnc doesn't put one before every GOP, while it says it does.

I'll post me findings later.

Regards, Leon

[droopy928gts]

Ah well, pitty

All to no availl. According to the EPAssist program there is a sequence header before every GOP in the original MPEG1 file. Don't know how reliable it is since it is a program from 1996 made with win3.1 or 95 in mind. It made a nice log though which stated all the entrypoints and their timecodes.

Anybody else have a torch which could light up this entrypoint accuracy problem of mine?

Regards, Leon

PS, using another muxer didn't make any difference. The blocks appeared after using MPEG Sequencemaker. So before re-muxing.