the masters are masters and i am just confused

[sean madison]

Ok, i gotta be doin somethin wrong or somethin cuz i have been workin on this for 6 hrs straight and cant figure it out!

I am using tmpgenc 2.51 and i just ripped a special feature scene of gone in 60 seconds.

It was only a 40 sec clip that had all fast motion constantly so i thought u know, if i can encode in svcd and not have any blocks on this scene, the actual movie will be flawless; boy am i stuck big time!

Ok this is what i have done and results-

I used smart ripper to rip the vob onto hd.
Then i used dvd2avi just as described in sefy's newbow dvd ripping guide and made the d2v file.

I then import the svcd ntsc template and do NOT modify it ok, i just import it and encode.
The results are bad! I do not have interlacing problems but i do get blocks!

I know vcd is better at "blurring the image" so it isnt so crisp to be potential to produce blocks so i imported the vcd ntsc template in and also did not modify it, i just encoded.

It too produced blocks!

I even went as far as importing the dvd 352x240 1850 cbr kbps temp and THAT EVEN made blocks!
Very little blocks but still yet, i am talking about quality.
I'm not aiming as identical quality as the vob since that is impossible but i am just aiming at encoding w\o blocks, thats all!

My little "headache" about all this is as follows-
Is it making blocks just because the picture is so clear and crisp?

Also, i dl movies off the net too ok, including dvd-rip avis and ts scr and so on
I encode all that in res of 480x480 svcd ok at bitrates as low as 1000 max cq_vbr

Even on avi dvd rips it does NOT produce blocks at high motion scenes! Why is that? why does it not mess up when someone else made it but for me, i'm "screwed."

Also, lets take screeners and telesyncs for example since i know many of u do encode these.
As same here, at 480x480 mpeg2 stream, it never makes blocks. Is that just cuz the quality on them are so far away from the crispness of dvd, the blocks would be unnoticable?

Also, i even tried makin me an avi in dvd2avi and even that was bad!
I tried microsofts v1 at 75% and 1900 kbps and divx 4.12 with setting of 1 pass quality based and divx 3.11 fast motion-0% and 2000 kbps. ALL 3 avis turned out bad cuz they produced blocks!

I got avis that got some pretty fast action in em and they dont produce blocks and they use divx as well!

What is goin on here?!?!

PS-
Please dont make this into a battle zone, this is a question, not a war.
if u wanna debate, agree with this advice-
Debating is arguing on the topic, not over the person who believes in the topic.

[alucard2050]

i'm not familiare with SVCD at all yet so i'll help with VCD. sounds like you've got everything correct for what you want to do but here's my suggestions. try encodeing at 480x480 1650cbr. set buffer at 122 and motion at highest settings. audio at 128. use filter to reduce noise blocks and select high quality mode. aspect ratio at 4:3 and make video arrange method full screen... see if that helps.

[sean madison]

i have done all that but even then, that shouldnt be necessary anyway.
i have done EVERYTHING u just mentioned.

Im goin by this-
Movie groups such as the one who released LOTR svcd dvd rip, it has no blocks and its quality is excellent and not just that movie either; i am talkin bout ALL svcd dvd rips.

Why why why?
If i can, i will take a snap shot of the blocks and let u see.

The biggest thing i cant understand is i know the smaller the res, the less likely blocks will appear so at 352x240 why would blocks appear even if using 1850 kbps cbr?

[Truman]

Simply the bitrate is too low for that procedure. TMPGEncoder is not that good at bitrates lower than 2000. You should use the new 2pass mode.
Also keep in mind, a DVD is quite heavy compressed alreaddy (~1:30).
There are visible blocks sometimes and invisible distortions as well. So it is usefull to use a soft denoise filter.
Actually which SVCD template did you use. I have tons of svcd templates here.
It seems to be a very high motion clip if blocks even appear at 352x240 @ 1850. Motion search precision should be set to "high quality (slow)".

[sean madison]

yes, it is a super fast motion clip.
Thats why i was sayin to myself, if i can master this part, the whole movie would look awesome!

I just used the one thats been around forever-
SuperVideocd {NTSC}.mcf

Now i had been doin some tests, not with 2 pass new type but with old type since i didnt dl that version yet and the quality never increased.

And to tell u the truth, VBR old type does not look any better or worse than CQ_VBR!

I dont know if tryin just this clip is throwin off my quality judgment altogether or what but all i know is is this-

We can dl dvd rips online in avi form.
Why can i take those, even though they look VERY good too and encode in 480x480 and it NEVER make blocks, why is that?

If i have to make an avi, then i want it to do that too but damn, last time i tried it, it made blocks, even usin 2300 kbps usin both divx and microsoft v1

[stanwebber]

if the source is that demanding you need to inverse telecine(or forced film in dvd2avi if applicable) regardless of whether you go svcd or vcd. you simply cant afford to squander 20% of the bitrate on the ntsc shortcut if things are flying around the screen

[sean madison]

ok i'll try inverse telecine but what settings in there should i use since there are about 20!

Also, is the only reason why im havin all these blocks just cuz its a pure dvd rip?
I can capture the dvd movie with my capture card and encode it fine with no blocks at low bitrates; its just that i dont want to have to do that cuz bright areas are white and dark areas are pure black, know what i mean?

[sean madison]

ok, i just tried inverse telecine inside and out and it helped some indeed but not 100%

I could cut the bitrate down to 1500 cbr w\o it being too bad, hardly noticable unless u were right up on tv.

Now, here is the thing though-

Either im testing a clip thats so fast action(this is not part of the movie), that it is invalid to do testing on cuz no matter what, it will produce blocks. I hope thats 1 reason and the main one, but, if it isnt, here are a couple of things i want to say and ask-

When i go to use dvd2avi, i finally was able to figure out how to make an avi from dvd rip w\o blocks ok.
Well, not testing this yet, but all avi dvd rips i have dl before and encoded in svcd res of 480x480 at bitrates usin CQ_VBR of max as low as 1100, it can produce either no blocks or so minimal blocks, its never noticable ok.

I was just wonderin if u think its better to just go from vob to avi then to svcd when using lower bitrates?
NOTE- i want svcd ONLY(mpeg2 stream only), not vcd or xvcd(not mpeg1 stream).

Also, one more thing-
All of the avi dvd rips i get, the res on em is NEVER 720x480, its like 640x272 or 576x344 and stuff like that.
Would it be better for me to resize the avi if i plan to go vob-avi-mpg and if thats a good idea, how do u resize and with what prog cuz i tried dvd2avi but it wont let me resize res, just crop it.

See, al the avi dvd rips i dl are resized in usually those 2 above res's and it converts to 480x480 perfectly.

[stanwebber]

is the movie 16:9 anamorphic? if not(or 2.35:1) you can crop & replace the original letterboxing as a considerable amount of noise can be generated here. if you are really diehard you can crop/resize to align the letterboxing on 16x16 macroblock boundaries. check your colorspace as well. 2 tv scale conversions(0,0,0>16,16,16>31,31,31) will cost you bits

[mijman]

Have you tried encoding it to 720x480 at your bitrates to see how much it blocks up? I realise that you want to end up at 480x480, but maybe even as an intermediate step. In this special scene there was a lot of fast action, is there an equivalent amount of action somewhere in the main movie that you can conpare with?
I've never mad a divx, so I don't know how their bitrates compare. Perhaps, because of their high compression, and chopping off the black latterboxing bars, they save enough space to have higher bitrates. (just an idea)

[sean madison]

I got it now, i tell u what i did, but thanks everyone anyway.

I chose to encode in 23.976 fps and use 3:2 pulldown. Just by doing that made the quality increase remarkably but still blocks and some interlace issues, so i then did this-
I went into advanced tab, went to de-interlace and chose even-odd field.

I swear to you all, when i did that and re-encoded it, it was a fresh encoding moment. I used min 700 max 1200 quality 59 on CQ_VBR and the quality was remarkable.

NOTE- there was very little, and i mean very little noise when very fast action occured but i cant really help that, but it doesnt matter because it looks so crisp.

Anyways, that test could hold up to 84 min of video in svcd onto 1 cd.
I know i said that u can hold more with nice quality and this is still no lie.
I have not yet tested and lower bitrates but when i do, i will let everyone know!

I love dvd-ripping!! (keep in mind i just started, hehe)

[stanwebber]

deinterlacing a progressive image will blur the hell out of it. if i had known how completely unrealistic & naive your goal was i would not have wasted my time on you. i am not amused

[sean madison]

i do not agree.

The crispness of the clip is astonishing, believe me.
My assumption in your part of blu, it ma blur it at fast motion which in turn would be good.
Of course i dont know if it is significant blurring at fast motion cuz of course, is fast and i dont feel like studying the movie, i'd rather watch it, know what i mean, hehe

I will put the clip on my website as soon as i get my new HD in, ok

Anyone who disagrees can watch it and then reply back to that.

I dont see, im serious too, any blurring in it.
It might be might not.

[KennyC]

I chose to encode in 23.976 fps and use 3:2 pulldown.

Hi sean,

What do you mean by 3:2 pulldown, sorry for my ignorance.



Cheerz

[kinneera]

He means telling the DVD-player to telecine on the fly (23.976->29.976 fps).

I agree, though, that de-interlacing a progressive image is idiotic. The reason it would help, if it all, is precisely because it blurs the hell out of the image, and the fewer the sharp edges the better MPEG works in general. There are far better ways of doing this than inappropriate deinterlacing however.

[sean madison]

ok, if there are better ways, how?

This is everything i had tried and some things listed may not correspond to quality issue but this is all i did try though-

(this is for svcd)

I tried in video tab-
non interlace, interlace, 3:2 pulldown and inverse 3:2 pulldown

In advanced tab, i tried-
Inverse telecine Inverse telecine worked kinda but still produced some blocks.

I see if more than one person agrees bout deinterlace blurs, if u or anyone else is interested, i'll send the clip to ya so u can see the quality.
IM me on AIM- shizzzon

If u think its blurred tell me cuz to me, it looks fine(in terms of not blurring on still parts and fast motion looks fine)

[kinneera]

I tried in video tab-
non interlace, interlace, 3:2 pulldown and inverse 3:2 pulldown

In advanced tab, i tried-
Inverse telecine Inverse telecine worked kinda but still produced some blocks.

Do you even understand what these settings do and why you would use them? It appears not, in which case I would suggest you take the "video performance master" off of your profile, and then search the forums (or hell, the internet) and find out - it's certainly been discussed before plenty of times.

If u think its blurred tell me cuz to me, it looks fine(in terms of not blurring on still parts and fast motion looks fine)

I don't have to see it to know that it's not going to look right, simply because you're deinterlacing video that isn't interlaced. If you have a film source, the better method I refer to is to apply a mild spatial smoother - this will assist the MPEG compression at the expense of softening the video. However, it will perform that softening much more appropriately, and to a degree that you specify. This technique is sometimes even applied very mildly to DVDs prior to encoding.

[vitualis]

sean, people here are blasting you not because they are nasty, but because we are exasperated by your ignorance in what is really quite simple conversion... Perhaps you really should take off the "video performance master" and replace it with something a little bit less aggravating.

Deinterlacing a progressive picture as others have stated is completely inappropriate. Deinterlacing is to (obviously), remove interlacing. To deinterlace a progressive picture will do nothing other than blur it. If you simply want to blur a picture, just use some sort of blur filter.

However, as others have said, to do so, you will be irreversibly losing detail and thus quality. YOU may think that the end MPEG may look better (because the blurrier image is easier to compress), but to some people, this unnecessary and completely arbitrary loss of detail is mindless.

If you do want to blur your video by a mild degree to help with compression, do it the way kinneera just posted.

3:2 pulldown basically bumps up the frame rate in a particular way (replicated fields and such) by 25%. The ONLY time you should use this is if you for some reason need to convert a 24 fps (23.976 fps) progressive clip to 30 fps (29.97 fps).

Inverse 3:2 pulldown does the opposite to the above and the only time you will use this the opposite to above too.

Regards.

[sean madison]

ok this might sound stupid but im not trying to be so bare with me, ok

If i stop using de-interlace on progressive which this clip is progressive, and use like sharpen filter but put it in the negatives to blur it, will it provide better quality than de-interlace?

Of course we all strive to have perfect quality and crispness and the only way to get rid of blocks at low bitrates is by blurring as far as i can tell.

Do u think using the sharpen filter in the negative numbers will make it better vs de-interlace?

[sean madison]

oh and one more thing, since its gonna be blurring, u think i should have that thing checked in quantize matrix to do no motion search at still picture parts?

[videoguy79]

sean--

Anything is going to look better than using a de-interlace filter on progressive source! I would suggest reading about the interlacing process--once you do, you will realize why doing such a thing is completely ludicrous. Furthermore, I personally would not recommend using a negative value in the sharpen filter. Even if you try to kill some of the sharpness in the video, your low bitrate will still result in some blocks, and a even blurrier picture to boot. I would recommend doing what Kinneera said--use a mild spatial smoother. MPEG has a hard time with sharp edges--it takes more bits to resolve a sharp line than it does a smooth one. I am not trying to put you down, but please realize that just playing around with pre-encode settings is not going to drastically improve quality on your final encode. Read my earlier post about data entropy and compression and you'll understand why video cannot be of high quality below a certain bitrate, especially using a codec as old as mpeg-1. You would be better off putting Vaseline on your television screen to "smooth" the video than expecting Tmpgenc to do well with such drastic filtering at such low bitrates (your talking to somebody who uses 4500-5000kbps max. on some of my encodes).

[vitualis]

Why don't you just do a few experiments yourself with DIFFERENT video clips and see what tends to work better?

In general though, if you want to prevent block noise in TMPGEnc, it has an option under "Quantize Matrix" to "Soften Block Noise". This is probably a better option than blurring your video willy nilly if you push for lower bitrates.

As for "no motion search at still picture parts", I've never tried it but TMPGEnc says that it can do something to "slight motion" as a side-effect. A bit of blurring on a still part of the picture I think is preferable to a motion artifact but that depends on you.

Regards.

[sean madison]

ok, what exactly is a "mild spatial smoother"????

That is either something i never heard of before or worded differently fo somethin else.

What is it and what prog can do it?

[videoguy79]

Sean--

"Spatial" and/or "Temporal" smoothing is something that I belive Virtual Dub can perfom, though I am going to guess that Tmpgenc's "noise reduction" filter should work on the same theoretical process. Temporal smoothing uses a low-pass filter to cut everything off below a video noise floor, and spatial smoothing uses morphological filtering. I have tried Tmpgencs' noise filter and I get the impression that it uses these processes. Try using it. I have never used such settings myself, so good luck.

[sean madison]

PROBLEM on both of those answers-

For one, i use d2v and not converting to avi cuz i dont wanna lose quality before it gets to tmpgenc. Therefore, if used vdub, i would have to convert to avi, do it, then resave and encode AGIAN into avi then encode AGAIN in tmpgenc.

2nd problem-
I have used noise reduciton in tmpgenc before, its totally lame and sucks.

It smooths too many things out as being all one color rather than many.

I will NOT use noise reduction, sorry

[videoguy79]

If the "noise reduction" filter is not to your liking, I don't know what else to suggest? BTW, Can your burner burn 99 minute CD's? If so, you can gain about 190MB extra space, space which you could use to up your bitrate a little bit, thus helping quality in a tangible and real way.

[sean madison]

yes it can but the cds cant hold it

[videoguy79]

yes it can but the cds cant hold it

Huh? What is meant by the "cd's can't hold it?" The statement I made is one of simple mathematics. If I have an extra 100MB laying around, and my movie is 100 minutes long, I can apply 1 megabyte extra to each minute. Depending on what bitrate you use, this could give you a decent increase in picture quality, though it certainly won't be earth-shattering. I have some 99 minute blanks coming in the mail--they will let me put a few extra minutes of 3500kbps (ABR) video on a CD compared to what I am getting now. If I can eeek out extra bits for my 3500kbps encode, why can't your @700kbps encode benefit also?? Please clarify.

[sean madison]

o u mean 99 min cds

U made it sound like burn 80 min cds to hold up to 99 min worth of data.

I've been too afraid to try 99 min cds due to the fact that i dont know if my dvd player could read up that far

[videoguy79]

What's there to fear? The worst that could happen would be that the CD wouldn't work. Big deal. If anything, many DVD players I have run across have no problems with those discs. My GE player (a clone of the Apex 703) plays them flawlessly. If it does work, you will have a better chance at fitting watchable-quality video on 1 CD, which is what you are aiming to do, correct?