Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Capturing and Editing Video > Capturing Video

Reply
 
Thread Tools Search this Thread Display Modes
Old 16th January 2003, 21:18   #1  |  Link
dilly
Registered User
 
Join Date: Jan 2003
Location: maple leafs country
Posts: 72
Deinterlacing, RGB questions.

Hi. Just got into capturing a few weeks ago. I am capturing from digital cable through S-Video input. TV-card is Leadtek 2000XP, not that it matters particularly.

I'm capturing shows with iuVCR in huffyuv at 512x384. I encode two-pass (int) Xvid, 128kbit MP3 with my files at about 220Mb for a 22 min show. I'm using Virtualdub to do all my encoding/editing. I also apply a deinterlace (blend) filter.

Now, my questions are...

- When I capture from iuvcr, what format should I be capturing in? I did all my captures in YUY2, and I tried RGB555 and noticed the filesize was much bigger. What's the quality difference?

- Is my resolution not appropriate for capturing (NTSC) TV? If I capture in a different resolution will I eliminate the interlacing effect?

- I think this deinterlace filter is affecting the "smoothness" of the video. Framerate is the same but it appears choppy. Is this filter no good?

Any help is much appreciated.
dilly is offline   Reply With Quote
Old 21st January 2003, 21:40   #2  |  Link
dar1us
Registered User
 
dar1us's Avatar
 
Join Date: Apr 2002
Location: Europe/Oceana
Posts: 332
Ideally capture vertical at multiples of 480 for NTSC or 576 at PAL(proper.

Some cards capture at ...x480 and then resize that to 384, this isn't ideal, you really want to do resizing yourself. If you are loosing frames because your hardware is too slow, try and use a better codec (huffyuv, pivcideo2) or possibly divide your FULL vertical res by 2, so 480/2=240 (576/2=288)

though it my mind blend is yucky, it should create smooth (motion)results

with my card, I capture in RGB24 (555 is tempremental(sp?)), YUY2 for me creates worse quality, or is this a brain trick, is my mind fooling me, telling me the lower the number the worse the quality, because I know I can see perfectly well

ALERT... RANDOMNESS: I just had a haircut, damn it is itchy, but man I look sexy
dar1us is offline   Reply With Quote
Old 21st January 2003, 23:26   #3  |  Link
Xesdeeni
Registered User
 
Join Date: Aug 2002
Posts: 467
Only use 24 bit RGB or 16 bit YUV for video. Anything less (like 16 bit RGB: 5:6:5 or 5:5:5) will look bad (gradients in smooth areas) and won't compress very well.

In general, it is a good idea to capture at the highest resolution which your capture card can handle.

HORIZONTAL:
When you convert analog to digital, you must filter the analog to avoid aliasing. The frequency of the filter is called the Nyquist frequency. But hardware filters cost money, so most capture devices only include one for their highest resolution. Any time you capture at a lower resolution, the card must scale down the samples somehow. If you are lucky, they use an interpolated filter. If you are not, they just throw out samples. In the latter case, they will introduce undesirable aliasing.

VERTICAL:
Since (PAL or NTSC) video is interlaced, the card captures two fields at half the requested resolution (240 for NTSC, 288 for PAL). If you decide to capture at a lower resolution, the card must scale down the lines somehow. To interpolate vertically requires on-board (or on-chip) memory, which means more money, so most cards don't do this. Instead, they just skip lines. The problem is that they skip lines in each field separately. For example, if you ask for 400 lines of 480, then you'll get 200 from each 240 line field. The capture device will skip every 6th line in each field. But when these two fields are put back together into a frame, the skipped lines from each will line up and you will have effectively skipped 2 lines out of very 12:
Code:
Original
Odd
 1
 3
 5
 7
 9
11 <- line to skip
13
15

Even
 2
 4
 6
 8
10
12 <- line to skip
14

Resulting frame
 1
    2
 3
    4
 5
    6
 7
    8
 9
   10 /__Big discontinuity here
13    \
   14
15
   16
Unfortunately, they can't do it any differently, because if they skip lines based on the whole frame, they get the lines out of order:
Code:
Original
 1
    2
 3
    4
 5
    6  <- line to skip
 7
    8
 9
   10
11
   12  <- line to skip
13
   14
15
   16

Resulting frame
 1
    2
 3
    4
 5
    7 /
 8    |
    9 |__Lines in wrong field (and wrong time!)
10    |
   11 \
13
   14
15
   16
Anyway, the best thing to do is capture at the maximum resolution that your card will go. Then, depending on what the video is and the desired destination:
Code:
             /--Source--\
             Film    Video
     / VCD     1       2
Dest | SVCD    3       4
     \ DVD     5       6
Video = interlaced or hybrid

1. IVTC (NTSC) or deinterlace (PAL); scale to 352x240 (NTSC) or 352x288 (PAL); encode at 23.976 (NTSC) or 25 (PAL) Fps

2. Deinterlace (a smart deinterlacer will look better, but throwing out a field will work as well); scale to 352x240 (NTSC) (352x288 PAL); encode at 29.97 (NTSC) or 25 (PAL) Fps

3. IVTC (NTSC) or deinterlace (PAL); scale to 480x480 (NTSC) or 480x576 (PAL); encode at 23.976 (NTSC) or 25 (PAL) progressive Fps

4. Either
-a. Deinterlace (a smart deinterlacer will look better, but throwing out a field will work as well); scale to 480x480 (NTSC) (480x576 PAL); encode at 29.97 (NTSC) or 25 (PAL) progressive Fps
OR
-b. Scale to 480x480 (NTSC) 480x576 (PAL); encode at 29.97 (NTSC) or 25 (PAL) interlaced Fps

5. IVTC (NTSC) or deinterlace (PAL); scale to 720x480 (NTSC) or 720x576 (PAL); encode at 23.976 (NTSC) or 25 (PAL) progressive Fps

6. Scale to 720x480 (NTSC) or 720x576 (PAL); encode at 29.97 (NTSC) or 25 (PAL) interlaced Fps

Xesdeeni
Xesdeeni is offline   Reply With Quote
Old 22nd January 2003, 15:14   #4  |  Link
Ookami
Xe-Rotaredom
 
Ookami's Avatar
 
Join Date: Oct 2001
Location: Croatia
Posts: 1,029
Only one thing to add to Xesdeeni's excellent posting...

AFAIK, all of the capture cards work in the YUV colorspace, so when you capture in RGB, it's just a waste of time and resources (YUV - RGB conversion and the picture surely will not get better).

http://www.virtualdub.org/docs_capture (linked in the FAQ, and linked in this forums, seems no one bothers to read it)

And, dilly, you would found all your answers if you'd used the search engine a bit (and followed the links in the FAQ).

But, as Xesdeeni answered so great (this thread will be linked in the FAQ), I will shortly answer some things.

Codec: Lossless, I recommend Huffyuv
Colorspace: YUV (see link above)
Resolution: See Xesdeeni's posting (max res. then resize)
Interlace/Deinterlace: If your format supports it don't deinterlace, for more deinterlace stuff look at the IVTC article by Manono et al.

Cheers,

Mijo.
__________________
"Only those who attempt the absurd achieve the impossible."
Ookami is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 00:05.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2017, vBulletin Solutions Inc.