I've got the ffvfw video encoder DS filter under Video Compressors, not in DirectShow filters.
...but, I cannot create the fake.avi window, as output of ffvfw. How can I create this window with fake.avi file, and how do I make then run it?.
Sorry if little off thread.

Me stupid Thx a lot. There she is
Under DS filters, search for "File writer" and name it blahblah.avi.

Read my signature

__________________
Go for SECAM =)

OK, all tests I've made have been with MPEG-2 at DVD resolutions. I really haven't tested VCD resolution, and I didn't really tried 352x240 MPEG-1 and looked at the VBV biffer size.
Now, at VCD resolution, the smaller VBV buffer is not much of a problem, even if it exists. But that is refering to quality. As far as playability on standalones, a small VBV buffes will be an issue on the mayority of players. So you are right. We're talking about two different things
So more tests have to be done at VCD resolutions, to see if in reality, the bitrate settings do create a correct average bitrate on smaller resolutions. You have an excelent DVD player Oh I agree 100% with you
It's far more advanced that CCE ($$$$$$$$$), MCE($$$) and TMPEG($)
I still can't believe some people who insist that MCE is the best encoder. But that's another issue I have on another forum, whuch is giving me headaches trying to explain, because they insist on the superiority of MCE I've trying to trace him, and I've had zero luck
I've been looking for him over at Sourceforge.net, and nothing. E-Mails sent, no reply. I guess we just sit and wait Because the totall average bitrate of a full encode is completely different to the settings on the "Bitrate(Kbps)" slider. But the VBV buffer value is fixed, afer I found out the constant for the correct value required. I guess I do something similar
My wife records almost every episode of "Stargate SG-1", and guess who's job is it to fit them on one disk It is
And when you add a little noise, it looks extremely pleasant Did you compare MPEG-1 and MPEG-2 at VCD resolution, with ffvfw
Because the tests we made, ffvfw looked better on MPEG-2. But that was on higher resilutions. I still haven't really made any serious tests at 352x240 MPEG-1 with ffvfw. What we really need is the MIN and MAX bitrate constraints in this encoder. That will really be helpful, to keep the boundary limits on the standalone players. Once this issue is resolved, I think ffvfw can be used for just about every encoode Hey, that's exactly why I keep playing with it, and tweaking the hell out of if
It's so full of options, that once we get them operating correctly (we're very close already. I'm positive of that ) we won't look back at other encoder ( for a while ) No need to apologize. We're all here to optimize, experiment, and learn from each other. The only people that have to apologize, are the ones that continuously insist that something is green, when in reality, it's really red ( As my case on another forum ) Excuse what

-kwag

Thanks, already running. I'll do some speed tests...

You stupid!!??, don't think so. Follow your posts with interest.

I've made some test (a while ago, but no one confirm that) with MPEG1 PAL VCD res, on cartoons, and results war beter than with MPEG2 even with notch matrixes. Yet this function is 50% slower than MPEG2 (I think that realy goes to the matrixes).

EDIT: @marcellus
I'll repeat docs question: why graphedit instead of VD ? I know some benefits of this, but I want to know yourse.

__________________
Go for SECAM =)

I use graphedit since I'm not sure if Vdub makes or not a colorspace translation to rgb, that way I am sure that all remains within yuy2. And moreover I can do the whole thing in one step-video and audio track. And because I think that way uses few memory. I use to set it from task manager to the lowest priority and use in the mean time the computer for some other stuff.

@Kwag
Thank you for paying attention to what I wrote. Did you tried H.263 quantizing method with your big resolution encodings? Because for me gave the best results. Hydeus pointed out a fluctuating quality problem and I encountered sometime. All I can say about it is that is related to non-intra q matrix (the standard mpeg one) thas has all values set to 16. I think that if it could be possible to make a H.263 encoding with notch matrix that would be a hit in the balls

Marcellus

Yes I did, with identical results. There was no change in file size or graphic pattern, as viewed on Bitrate Viewer. I guess internally it uses the current selected matrix, and always encodes as MPEG. But H.263 is a complete different monster
There's no way we're encoding H.263 with ffvfw with the settings selected, becaue if we were, we couldn't possible play it back

H.263 is:

One I frame. Just like mpeg1/2.
One P frame. Just like mpeg1/2.
One PB frame. Represents 2 frames, and based on a previous frame.
One B frame. Just like mpeg1/2.
One EI frame. A frame having a temporally simultaneous frame which has either the same or smaller frame size.
One EP frame. A frame that has 2 references, one previous and one simultaneous.

And that's just scratching the surface.
So as you can see, no matter if you encode with that setting set to H.263, you're still encoding IBBP structure

-kwag

Nope, I read your post sometime ago about this and took it for granted
I think that purpose (MIN and MAX bitrates) in the encoder is out of reach the way this codec is built right now. Since you can not set bitrate constraints but only Q constraints, the final MAX and MIN value is strictly dependent only on source that you encode. And that means that you have to tweak it on every file you encode till you are blue in the face. The only place you can set a Max value is in 2nd step of encoding and this brings us again to two steps encoding. And even there is no MIN value. I think that the internal logic of this encoder is much different than the ones we tried before.

I think that the H.263 thing is misleading (me or you). I didnt believe the encoder makes a H.263 stream (I would be very gratefull if you could explain me what exactly is H.263 because I have no clue -I know it's something in relation with mpeg 4 but no furher) but uses some "esoteric" method to quantize. And, unless I have to be taken to an eye doctor, it seems to me better looking than mpeg method. The obtained stream is an mpeg2. So, if you didn't try but just thought that could not be played, please give it a try! You might be surprised. As I said, it played on my standalone.

@Kwag
To be more clear: I am talking about quantizing method and you about GOP structure...

I'm going to bed now, here it's 3.45 am...

That's because the stream it made, it's a standard IBBP mpeg stream
There's no standalone that will play an H.263 stream.
Here are a couple of link to what H.263 is, and btw, H.263 is not related to MPEG-4, although some MPEG-4 CODECS do include H.263. But it's more related and used on low bitrate video conferencing protocols.

http://www-mobile.ecs.soton.ac.uk/peter/h263/h263.html
http://www.ece.cmu.edu/~ee899/lecture5.pdf

-kwag

Thanks Kwag for those links. I can't say that I understood much. I have no time right now to dig deeper, it's monday morning already.
But again, DID YOU tried that option? Because all I can understand from your answer is:
-you didn't try it because you don' believe will give anything useful and think you would just waste your time OR
-you did try it but it's simply encoding the "old" mpeg way with standard mpeg matrix, so that option there is basically doing nothing, just taking up space, so you wasted your time already

Please, give me a yes or no answer.
Sorry if I become annoying
Marcellus

Hey, it's our "job" to be annoying We all are here for search of good, great, and even beter solutions in MPEG encoding.
At the start, Kwag don't wanted to hear of ffvfw because of all this bugs and shortcomings, but few of as insisted, and finaly he started tests. After few days we gained powerful encoder And we keep going. I'll do some test on H.263 mode, but now I have a litle to litle time for this

I also noticed that VBV problem appear in mux part, and good (or mabe newest) standalones have no problem with that. I do my test on Panasonic NV-VP30, and this baby eats everything

__________________
Go for SECAM =)

Hello: What kind of correction is that. When must we make that correction?

Here as already posted, so watch the thread

BUT: As I see he encoded using the std. matrix, IF you use NOTCH there also should be the opportunity to still re-treat that stream in restream using the same matrix while this stream is encoded with.

So if possible EVERTIME use in the matrix settings "same as source" if possible.

I really don't like when people are tlaking behind my back so I have to fix some point there :
I am the only one that declared to drop the codec for a while so I guess you are talking about me there.
I'm personally very hurt by the words "shame". I shouldn't have to justify my act but I quit the codec for lack of time ! Do someone have a problem with that ?

I use to be on the forum from 9am to 4am since weeks, helping people, answering their questions and waiting for 30 min more to see if they have other question before to close the thread and I won't talk about the dozens of hours spent on script testing to provide the one that are in the avi->kvcd forum for instance.

I bought a DVD writer since 2 month and did have to time to burn only 2 KDVD. I have a TV capture board on my desk since 3 weeks and didn't even had the time to plug it into the PC.

I spent 2 nights (yes, two NIGHTS) on ffvfw, and let continue to see defaults than nobody seems to see. I just can afford more to it.

I'm sorry to take it personnaly, but I do.

Sorry, but I would never use WinDVR for capturing as it gives a worse quality than just a good mjpeg YUY2 capture codec! Or another better high bitrate capturing codec like the one from MCE (yes for capturing MCE at high bitrates about 8000-10000kbit is good and fast).

Second: At least do capture in 352x576! so you capture fieldbased as you never know if a broadcasted Hollywood movie is just speeded up to 25 and fieldshifted ("looks" like interlaced but its telecined) or even a very bad conversation broadcasted from 29.976 to 25 fps! Yep I got that last week when watching the capture from Man in Black!
Because when capturing in 288 height, the second filed will just be discarded which ends up with "stairs" in fine circles etc.

I'm sorry if you took it that way. All the posts are floating in my head I didn't really know how many people said that they quit, much less who they are when I wrote that post. Futhermore, I might use an improper tone since I'm not a native english speaker.
So I apologize.

Things being clear now, you're welcome to the board