The best: This VCD setting kicks ass

[Bull]

My card: Pinnacle PCTV Pro
System: Win98 SE (to improve I close all programs down so only Virtual Dub and Explore is open)
Computer: 1.6GHz - GeForce 7700 32bit (now I don't know anything els)



Virtual Dub v1.4.8:
Format: 352x288 - 24 bit RGB
Video Compression: Huffyuv v2.1.1 (both at best and with "Always suggest RGB format for output" and "Swap fields on decompression" on)
Audio Compression: Cd-qality
Overlay and preview off



TMPGEenc v2.51.33.127
Load: Video-CD
Video:
Motion search precision: Highest Quality

Advanced:
Video Source type: Non-interlace (progressive)
Field order: Bottom field first (field B)
Source aspect ratio: 4:3 Display

Noise reduction: 20-2-100 (High quality-off)
(I have some "Simple color correction" and "Clip frame" on...but thats just me)

Audio:
Use Audio edit: Change volume (200) and Normalize (100)



I think this is the best settings but if anybody knows how to improve let me know.
Oh yeah, I tryed "Quantize matrix" - "Soften block noise" (35-35), but with all that noise reduction on it just made it worse.
I would like some response so please write back

[Piper]

You'll get even better results if you capture at full res with Huffyuv rather than at 352x288.

[TheInformer]

Field order: Bottom field first (field B)

You're only going to capture the bottom field if you capture at 352x288. Capturing at 352x576 allows you to capture both fields per frame. You're missing half the video information right off the bat.

TOP FIELD (288 lines)--------\
BOTTOM FIELD (288 lines)----\___ together these give you one frame

Video:
Motion search precision: Highest Quality

I'd change that to "Higher Quality". There's a negligible difference in the output, IMHO, and the encoding time will be markedly shorter.

Everything else looks ok though if you have the hard drive space and I/O speed to handle the video and audio data streams.

[Anonymous65]

Yes, capture at 352x576, it will "greatly" increase your picture quality. Let Tmpeg do the resizing and deinterlacing (no need to specify deinterlacing, it is automatically done when compressing to VCD).

[sean madison]

there is a difference between high and highest in motion search precision cuz i have noticed.

U wont see any difference unless u are encoding vids 90 min+ onto 1 cd.

If u encode like a 100 min movie, i think my temps are set to use high quality. If u use high quality on 110 min though, blocks occur. In this case, if u use highest quality, the blocks are very visibly either completely disappeared or by far better than using high quality.

But if u are just encoding straight vcd standard 80 min per cd, u never need to use highest quality.

[Bull]

TheInformer:
If I capture at 352x576 I would get both fields or something like that, so it doesn't matter if I go with Top or Bottom at this resolution?

skittelsen:
How do I do that with out getting a realy strechy picture?

sean madison:
Can you explain this for me again.


I realy thougt I would get alot of response on this "Video Compression: Huffyuv v2.1.1 (both at best and with "Always suggest RGB format for output" and "Swap fields on decompression" on)", but I don't so great.

Thanks for the help

Nicolai

[kinneera]

I realy thougt I would get alot of response on this "Video Compression: Huffyuv v2.1.1 (both at best and with "Always suggest RGB format for output" and "Swap fields on decompression" on)", but I don't so great.

"Always suggest RGB format" is just a compatibility issue (to attempt to resolve a bug in certain software really), and "swap fields on decompression" could make for a mess if your target format wasn't dropping one field anyway. In most cases, you shouldn't need either option enabled.

As for the stretched nature (squeezed really) of a 352x576/480 capture, it will go away either (1) when you smart-deinterlace and resize to 288/240 vertical resolution or (2) when you encode it as MPEG and specify the display aspect ratio as 4:3.

[sean madison]

ok, if u are just using the standard vcd temp, u need at MOST EVER use high quality, but normaly, just use normal for motion search precision.

Highest quality is only used if u want to put 110 min or more on 1 cd. Trust me, i have noticed the difference between high and highest when it comes to low bitrates to fit those lengths of movies on 1 cd.

[Anonymous65]

The TV set is 4:3, so make sure you select 4:3 aspect ratio in tmpeg. It will then make sure the picture looks correct when you play it back on a TV.

[TheInformer]

Bull:

If I capture at 352x576 I would get both fields or something like that, so it doesn't matter if I go with Top or Bottom at this resolution?

The order you capture in does matter. Different capture/TV cards capture the fields in different orders, so you're going to have to find out which way yours works.

Here's what you do - make a short capture of a scene without much movement, say 10 seconds or so. Encode the file with "Top Field First" checked in TMPGEnc. Now change that option to "Bottom Field first" and encode the clip to another file. Watch both encoded files. Use the option that gives you the better output. I'm trying to avoid too lengthy a discussion on field order here

[Bull]

OKay, thanks for the replys. THere whern't very much to improve but there were some. I stil can't get the 352x576 to resize in TMPEGEnc with out just getting a realy long picture.

[TheInformer]

Bull:
Capture the video at 352x576, then use VirtualDub's 2:1 Vertical Reduction filter. It will resize the captured video to 352x288 for you and you can then either save it and load the resized file into TMPGEnc, or frameserve it to TMPGEnc.

[Bull]

So in conklusion tis should be the best.

---------------------------------------------------------------------------

Captur in Virtual Dub v1.4.8:
Format: 352x576 - 24 bit RGB
Video Compression: Huffyuv v2.1.1
Audio Compression: Cd-qality
Overlay and preview off

Frameserve or resize avi in Virtual Dub v1.4.8:
Filter: VirtualDub's "2:1 Vertical Reduction" filter (will resize the captured video to 352x288)

TMPGEenc v2.51.33.127
Load: Video-CD
Video:
Motion search precision: High Quality (if it's under 90min to 1cd) or Highest Quality (if it's over 90min to 1cd)

Advanced:
Video Source type: Non-interlace (progressive)
Field order: (Different from ripper ti ripper)
Source aspect ratio: 4:3 Display

Noise reduction: 20-2-100 (High quality-off)
(I have some "Simple color correction" and "Clip frame" on...but thats just me)

Audio:
Use Audio edit: Change volume (200) and Normalize (100)

And maby youse tooLame for better audio

---------------------------------------------------------------------------

You're only going to capture the bottom field if you capture at 352x288. Capturing at 352x576 allows you to capture both fields per frame. You're missing half the video information right off the bat.

TOP FIELD (288 lines)--------\
BOTTOM FIELD (288 lines)----\___ together these give you one frame

Exlplane this to me again, please?


And now for something comletely different:
How can you insert pictures in this forum, couse if seen people do it.[/quote]

[TheInformer]

You're only going to capture the bottom field if you capture
at 352x288. Capturing at 352x576 allows you to capture both
fields per frame. You're missing half the video information right off
the bat.

TOP FIELD (288 lines)--------\
BOTTOM FIELD (288 lines)----\___ together these give you one frame

Exlplane this to me again, please?

Ok.

When you look at a TV screen, it isn't one solid image after another like you have with film. A TV image is made up of many lines which are drawn xx number of times a second: NTSC=60, PAL=50 (I believe). These lines are drawn starting from the top, with line #0 and progressing 2,4,6,8,etc. Once the bottom of the screen is reached, then the other lines are drawn (1,3,5,7,etc). This is called an interlaced image. The 'top' field on a screen includes the lines 0,2,4,etc, while the bottom field includes the lines 1,3,5,etc:

FIELD TYPE

0 TOP field
1 BOTTOM field
2 TOP field
3 BOTTOM field
4 TOP field
5 BOTTOM field
6 TOP field
7 BOTTOM field
~ ~
~ ~
574 TOP field
575 BOTTOM field

All of the top fields go together, and all of the bottom fields go together. When you combine them together, you get a frame.

When you capture at 352x288, you're only getting the 288 lines that make up the TOP field (remember the TOP field lines are drawn first). The next 288 lines after those are the BOTTOM field and are not being captured. Capturing at 352x288 means you're only getting every other line.

Now, to those who are more experienced, I know that this explanation is not the more technically accurate - it was only meant to illustrate the point