xlogo question, plus some audio questions

[spiritgumm]

I've used xlogo successfully, but noticed that in action scenes, the mask area uses incorrect backgrounds. Can the command line be tweaked to be more aggressive in getting a more accurate background? I do a 2-pass with Vdub (and also for dvd video with Hcenc). Here's a command line for the sample I attached:
XLogo("Logo7_x_610_y_470_2.bmp", X=618, Y=478, alpha=0,SCENE=0,BLEND=TRUE)
sample.avi

Speaking of Vdub, when feeding an AVS script into it, I wanted to keep the original audio format, but selecting "Direct Copy" in the audio menu outputted wave. Is there a way to have the audio pass thru untouched? It's loaded in the script separately (demuxed with DGindex).
Of course I can manually mux the original audio (using Avidemux for xvid, or Muxman for m2v). In that case, would I add in the audio delay reported by DGindex?

I also noticed that when I set the "Audiodelay(.066)" in my script (the audio delay was 66ms), the resulting audio wave audio wasnt synced. I had to use 0.66 instead. I thought the delay was suppose to be 1/1000 of the delay reported in DGIndex. Did I need to change it one decimal place because I'm IVTCing the video?

[AlanHK]

I've used xlogo successfully, but noticed that in action scenes, the mask area uses incorrect backgrounds. Can the command line be tweaked to be more aggressive in getting a more accurate background?

Xlogo works from the area outside the mask and extrapolates it to fill the mask area; it doesn't know what the real background is.
It can never restore the area perfectly.
Your sample if that's the result of xlogo filtering, is about as good as I think it can be.
No doubt it bothers you, but I had to look twice to find where it was.

If you want better, then look into InPaint, which is rather more complicated.
http://avisynth.org/mediawiki/InpaintFunc
I don't want to spend the time to tweak it myself, but you may if it's important to you.

ISpeaking of Vdub, when feeding an AVS script into it, I wanted to keep the original audio format, but selecting "Direct Copy" in the audio menu outputted wave. Is there a way to have the audio pass thru untouched? It's loaded in the script separately (demuxed with DGindex).
Of course I can manually mux the original audio (using Avidemux for xvid, or Muxman for m2v).

When you load an AVS script, it passes the decoded and uncompressed video and audio to VDub.
So the audio is Wave as far as VDub knows.
If you want the original audio, you have to open the native file, not the AVS.
VDub can do that for XVid at least.

In that case, would I add in the audio delay reported by DGindex?

I also noticed that when I set the "Audiodelay(.066)" in my script (the audio delay was 66ms), the resulting audio wave audio wasnt synced. I had to use 0.66 instead. I thought the delay was suppose to be 1/1000 of the delay reported in DGIndex. Did I need to change it one decimal place because I'm IVTCing the video?

The Avisynth function is "DelayAudio()".

DGIndex names its demuxed audio with the delay in ms.
But I don't do much with DGIndex, so can't help you there.

However, IVTC does not affect audio in any way, so that's not it.

[spiritgumm]

thanks Alan, and I'm sorry I forgot to mention the logo location. It's mainly noticeable when the red coats run by. I assumed I wasnt using the command to it's fullest potential. I've seen some amazing work by member Poisondeathray with Removedirt in another thread, so again assumed I was probably not doing enough.

Thanks for the explanation about Vdub.

Yes, sorry, DelayAudio(). The milliseconds from DGIndex should be converted to seconds for Delayaudio, but it's off a decimal place. At least, that's how it comes out when decompressed to wave in Vdub.
I guess I'll just add the delay into whatever mux program I use and see if it works.

If I decide to process the audio in avisynth with Audiodub(a,v), Delayaudio(whatever), and Soundout, would I still need to use the delay number when muxing the audio/video?

[manono]

I guess I'll just add the delay into whatever mux program I use and see if it works.

I remove them using DelayCut. After that you don't have to worry about them.

If I decide to process the audio in avisynth with Audiodub(a,v), Delayaudio(whatever), and Soundout, would I still need to use the delay number when muxing the audio/video?

No, you shouldn't have to. But if I wanted the original AC3 audio, I wouldn't stick audio in the script to be decompressed and later reencoded.

[spiritgumm]

I tend to forget about Delaycut because the GUI always confuses me at first. I still wish the DGIndex delay (in decimal places) was consistent.

Have one more question: Is there a way to take a big Xvid, cut it up into separate shorter files without causing errors, and still see the initial bit-per-pixel value?
The situation is I have several TVshow dvds, and the VOBs do not correspond to individual episodes. Using DGIndex, I made one big Dtv for a dvd with the intent of making one big Xvid, then cutting it up into the episodes later with Avidemux. Of course, the episodes werent separted by keyframes, so Avidemux wont make clean cuts. I also noticed the bit-per-pixel rate went down steadily for each episode. I would have thought each episode would be consistent.

[manono]

I still wish the DGIndex delay (in decimal places) was consistent.

It is consistent. What do you think is wrong with it?

Is there a way to take a big Xvid, cut it up into separate shorter files without causing errors, and still see the initial bit-per-pixel value?

Since BPF is a nearly worthless gauge of quality, I have no idea why you'd care. You could open the original D2V file in GKnot, fill in your final file size and the audio bitrate, and get the average BPF that way.

I also noticed the bit-per-pixel rate went down steadily for each episode. I would have thought each episode would be consistent.

Perhaps the later episodes compressed more easily. Which is why BPF is next-to-useless.

[AlanHK]

I tend to forget about Delaycut because the GUI always confuses me at first. I still wish the DGIndex delay (in decimal places) was consistent.

Post about that on Doom9, the author (Donald Graft) is a regular there (and occasionally here).

Have one more question: Is there a way to take a big Xvid, cut it up into separate shorter files without causing errors, and still see the initial bit-per-pixel value?
The situation is I have several TVshow dvds, and the VOBs do not correspond to individual episodes. Using DGIndex, I made one big Dtv for a dvd with the intent of making one big Xvid, then cutting it up into the episodes later with Avidemux. Of course, the episodes werent separted by keyframes, so Avidemux wont make clean cuts. I also noticed the bit-per-pixel rate went down steadily for each episode. I would have thought each episode would be consistent.

Try demuxing to MPEG with PGCDemux first it.
If you demux by "Titles" you should get a separate MPEG for each episode.

thanks Alan, and I'm sorry I forgot to mention the logo location. It's mainly noticeable when the red coats run by. I assumed I wasnt using the command to it's fullest potential. I've seen some amazing work by member Poisondeathray with Removedirt in another thread, so again assumed I was probably not doing enough.

Yeah, I could see it. When the background has abrupt large changes of coulour it can't be seamless.

I'm content with removing 98% of the logo, which is pretty easy. You can spend a lot more time and maybe get 99%, but I can't help you there much.

Yes, sorry, DelayAudio(). The milliseconds from DGIndex should be converted to seconds for Delayaudio, but it's off a decimal place. At least, that's how it comes out when decompressed to wave in Vdub.
I guess I'll just add the delay into whatever mux program I use and see if it works.

If I decide to process the audio in avisynth with Audiodub(a,v), Delayaudio(whatever), and Soundout, would I still need to use the delay number when muxing the audio/video?

Yes.
For final tweaking of sync in an AVS project I use VirtualDub and its Audio display which shows the waveform, so I can see exactly which frame a sharp sound occurs in and adjust AudioDelay accordngly.

I use the version of VDub from here: https://forum.videohelp.com/threads/288922-Interesting-Build-of-VirtualDub-Mod which has the audio display and also the AVS script editor, so I can edit the script delay and refresh with F5 to confirm it's right.

[spiritgumm]

Alan, thanks for the tip about PGdemux; the only problem is it doesnt produce a d2v file, but I guess that isnt necessary to load mpeg and whatever audio format in Avisynth.

Looks like you guys have conflicting answers about using the Delay number after being processed thru Avisynth for later muxing.

manono, you can read about the audio delay problem in top post.

Some filesharing sites use BPP as a quality benchmark, so I need that value to be consistent. If that's not an accurate quality guide for Xvid conversion, is there a simplified guide for making good xvids?

[AlanHK]

Alan, thanks for the tip about PGdemux; the only problem is it doesnt produce a d2v file, but I guess that isnt necessary to load mpeg and whatever audio format in Avisynth.

After you get the MPEG for each feature with PGCdemux, then you can use DGindex on the MPEG to make the D2V to load in Avisynth.

Looks like you guys have conflicting answers about using the Delay number after being processed thru Avisynth for later muxing.

No conflict, I just don't know what your problem is. A good idea to check sync anyway, it may not be correct in the original file.

Note that PGCdemux will directly tell you the delay if you click "Check A/V delay".

[spiritgumm]

Alan, before your last 2 posts, I decided to load the TV episode VOBs in DGIndex and edit (separate the shows) in the DGIndex window. Is that an okay option, or does that cause some unexpected issues?

Inre the conflicting responses about audio delay after avisynth processing, manono thought I shouldnt have to deal with it, but you indicated I would.

[AlanHK]

Alan, before your last 2 posts, I decided to load the TV episode VOBs in DGIndex and edit (separate the shows) in the DGIndex window. Is that an okay option, or does that cause some unexpected issues?

As a general rule it's better to extract MPEGs from VOBs before doing anything to them.
VOBs are meant to play as part of a DVD fileset, they may work in other contexts or they may not. An MPEG is a stand alone video file.

Inre the conflicting responses about audio delay after avisynth processing, manono thought I shouldnt have to deal with it, but you indicated I would.

Well, if you just play it in VDub you'll know if sync is correct. I would always check.

[manono]

Some filesharing sites use BPP as a quality benchmark

No one with any knowledge of video encoding uses BPF as a quality measurement. They may wish that you include the figure, sure, but I wouldn't worry about whether or not it reaches some arbitrary guideline for quality.

...so I need that value to be consistent.

No you don't. Just do a few encodes all for the same quality and then measure the BPF and see how much it (and the filesize) varies from encode to encode. If it were really a useful measurement, don't you think the figures would at least be very close? Use different kinds of sources - dark and slow, and bright and action-filled. Also, you can artificially inflate the BPF by lowering the resolution. Do you think that's a good idea?

manono, you can read about the audio delay problem in top post.

There's not near enough information there to draw any conclusions. The decryption itself can often be the cause of the problem. And sometimes it's a good idea to cut off a few frames at the very beginning using the '[' button in DGIndex before creating the D2V.

[spiritgumm]

Some filesharing sites use BPP as a quality benchmark

No one with any knowledge of video encoding uses BPF as a quality measurement.

So is there a reputable guide for better conversions?
I didnt mean to misrepresent the site's quality criteria which also suggests keeping the resize mod16, IVTCing, etc. Here's the line about BPP:
"The file size should be in correlation with the duration of the movie. This means that for most rips the video bitrate should be over 1000kbps and bits-per-pixel ratio at least 0.18."

[manono]

So is there a reputable guide for better conversions?

You don't need a guide. Just do quality-based conversions. Then you'll get results consistent with the quality you desire. But the file sizes may be all over the place (and is further proof of the uselessness of the BPF as a strict guideline).

Of course, if you have to make them for a specific size (CDs for AVI or DVD5 for DVDs), then you have problems with that suggestion. For DVDs I get around that by adjusting the quantization matrix I use.