Intresting huffman stuff

[SatStorm]

Hi Baker,
CCE do have an inbuild resizer and I know it. I also know that it is much better to use frameserver with this encoder for many - many reasons. I wrote exaclty: "With CCE you always use frameservers to resize", not "With CCE you always use frameservers to resize because you can't do it with that program direct". How you concluded that I don't know about CCE's inbuild resizer?
The way I put it that way, is because I don't know a single person using CCE without frameservers! It is a must with this encoder, as it used to be with TMPGenc a few versions before. But I do know many users do NOW direct this proccess with TMPGenc, including me. Ofcourse, when problem occurs with avis, I do frameserve them with virtualdub, plus I always have to frameserve DVB source with DVD2AVI. But when I grabb a TV show, I don't have to frameserve it anymore to TMPGenc, I just load it, cut the commercials (with this excellent cut scene fuction it has), I load my template and hit encode. That's it!

Now, about capturing....
Inside the digital reciever (Digi boxes are dedicated-to-one-service-only digital recievers), the picture proccess part, is convert the any-digital-frame-size to analogue PAL. That is 704 X 576 (broadcast PAL).
So, when you grabb from the SCART you have always D1 CCIR picture. This picture may come from an anamorphic 352 or 480 or 522 or 544 or 704 or any horizont size X 576/480, but is de-anamorph to CCIR specifications and then it is converted to analogue PAL. It is called D-A (digital - analogue) proccess and it must be done, because TVs only understand analogue Video In systems!
So, the rule is "grabb higher for better picture", as for all analogue situations. You do know now, how the analogue grabbing goes... It is like taking pictures of the screen many times each second. The PC card don't understand nothing, just photograph endless times in a raw.
The analogue grabbing have limitations. Grabbing analogue from a satellite reciever, is like grabbing from an ideal SVHS tape (without the analogue noise and interlace swaps between fields, so the picture is stable). SVHS/Satellite (analogue output) is the higher quality you can go grabbing analogue way with the use of typical cables and PC grabbing cards. Now again we go back to an old subject, which the opinions are not the same for all of as.
For me, any analogue grabbing beyond ~ 388 X 520 (both pal/ntsc), has no meaning. I said than, knowing that for DVD/DVB/DV material, there are other ways to get the stream info to your PC than grabbing the analogue way. Unfortunatelly, Sky Digital is an exception. Currently there is no way to grabb direct the DVB transmission because there are no modules of the system Videoguard, which is used by Sky Digital, so to receive those channels to those special DVB receivers, able to grabb those transmission (nokia 9600 with dvb 2000 firmware) or PCI DVB/S cards with C.I. The direct DVB grabbing is posible only on some FTA sky digital channels like Boomerang, TCM, Sky News, Channel healts, etc.
Back to the analogue grabbing
The only exception in the analogue grabbing, is grabbing from LD with the use of S-Video. Then you can Grabb 544 X 480/576 and get what S-Video can give you for sure!

Now, because the standard is 576 for PAL and 480 for NTSC, it isn't suprise that NTSC's analogue grabs from TV/VCR/etc, are always BETTER PAL grabbs, because with PAL you boost (wanted or not) to 576 vertical lines (from the ~520) so you add fake lines or noise to your capture.

The limitation/faults of the current encoders we use and the codec we use not let us have usefull grabs beyond ~388 X 520 in PRAXIS. There are at least 250 noisy group horizontal dots per line plus ~60 lost/noisy vertical lines per frame using eather composite or s-video. It is unoticable on your eyes, but is very noticable to the encoders. That's why I say grabb @ higher 388 X 576/480 and you are sure that the grabb have included for sure ONLY usefull and not noisy info. (noise I mean from lost horizontal grouped dots)
This is my personal opinion and I have tested it a lot to conclute it.

Other users, many ones, much more than those who believes what I state here, goes with the book specifications and state that grabbing 720 X 576/480 have benefits. Well, in the book yes, but as I said in praxis I don't see this. I don't see this. Others might see it. Well test yourself and follow what is better for you.

Other example: On books, S-Video is far better composite. In praxis, you can grabb with composite @ 720 X 576 any analogue source and the results are identical than grabbing with S-Video at the same resolution. How this is possible when the book specifications limit composite?
But even if you go with the book, then because for the reasons I posted, (I support as higher grabbing the resolution of ~ 388 X 520), the composite (officially 640 X 480) is more than enough for any analogue grabb.

Again, because I see this results and I believe and have those conclusions, it doesn't mean that YOU (all) have to adapt my opinion. I simply post it. You are testing yourselfs and conclute to YOUR opinions. Then you post them here and ALL TOGETHER we help newcomers to our wonderfull hobby.

The secret for a successfull forum is to post all the opinions and alternatives a way anyone can test them. By forcing others accept one opinion, destroys anything. I always post all the alternatives and propositions in my replies, even if I don't support/believe it myself. That is the right way, I think!

About tests: I done 18 months of tests, wasting time to crazy GOPs, resolutions, encoding tests, etc. So, my conclusions are those I posted various times, and nothing from the current situation can change them. I don't expect people adapt them, or blind follow them. I don't even say that they are right or wrong. For me (and some others) are correct. For others, maybe much more many, are faults. I respect that. But also, respect my opinion! (this is not for you Baker, but for other well known users). I don't ask something impossible I think!

[bangoman]

v interesting thread!

I do a lot of capturing off Sky digital, in an attempt to capture every Simpsons episode to SVCD! My settings are as follows:

- I capture at 480x576 using Huffyuv (which I find to be excellent) in VirtualDub.
- I compress using TMPEGEnc keeping the size at 480x576
- I use 2-pass VBR with the maximum at 2500 and the minimum at 1500.
- I use the project wizard to calculate the average bitrate so that I can get 2 episodes onto each 700mb CD-R (49%). Usually avg bitrate is about 2000-2050.

I find in the main that these settings are fine. However, sometimes the picture can be very blocky during high motion scenes - is this a limitation of the bitrate I'm using/MPEG-2 or is there anything I can do about it? I enabled the logging screen as well, and sometimes I can see it encoding at rates below the 1500 I set - what's that all about.

Also, do I gain any benefit in encoding at 480x576 or should I resize to 352x576/288?

SatStorm - have you got a website or anything like that with all of the different settings you've used in the past for capturing off Sky digital?

[energy80s]

Instead of VBR, try CQ mode at about 85. Your average bitrate seems very low considering 2 episodes of "The Simpsons" run for about 42 minutes.

[baker]

Post now cont here with pic:

http://www.vcdhelp.com/forum/viewtopic.php?t=113873&highlight=

I have to reply to some stuff there also.

Baker

[baker]

Good post satstrom

You see I want my video at 352x576 as it is my preferred res. But a video at 352x576 looks crap, not sharp enough. So some tricks need to be used. I capture at 480x576 then bicubic resize down to 352x576 to give the video a shrper look. Its also the trick behind my cvd guide. Without bicubic CVDs look shit. All shrpeners are crap as they introduce noise into the picture. Bicubic actually gets rid of noise.

Baker