horizontal resolution with analog tv caps

[warbird]

I capture off my AIW Radeon tv tuner, off analog cable tv and usually off local broadcast stations, and encode to standard vcds. I was wondering does it really make much of a difference which horizontal resolution i cap with? See I am struggling trying to decide whether its worth using picvideo @19 just so i can cap at full d1 720x480 rez, or if it would be better to cap at 352x480 using huffy. Full d1 with huffy takes up SO much space, but i usually have enough space for most of the stuff i capture at 352x480 with huffy. My point is, is it worth sacrificing lossless huffy compression to picvideo so i can capture at full d1 rez, or would it be better to sacrifice the extra horizontal rez so i can use huffy?

-------------------------------------------------------------------------------------
My specs:
Athlon Tbird 1400mhz/266
Abit KT7A-RAID
512megs Crucial pc133 cas2
All-In-Wonder Radeon 32mb agp
Turtle Beach Santa Cruz
30gig maxtor ata/100 7200rpm
Plextor 12/4/32 SCSI
400w pwr

Pioneer DV-343

[g0dZpr0fi7]

If your encoding to standard VCD 352x480 is fine.

[Gameshow Host]

Hi, I'm a newbie so I might be wrong about this, but here's how I see it...

Horizontal resolution
If you're grabbing video to make into a VCD then you only need to capture at the horizontal resolution of VCD (352). If you capture 720 pixels wide then you will have to scale the image afterwards and you will lose quality in doing so. It's always ideal to capture at the resolution that it will end up as.

Vertical resolution
As for the vertical resolution, it's different because you need to capture both fields in order to get 60 fps. So capturing at full screen (480) makes sense. Then it can be scaled down to VCD resolution (240).

I may be wrong about this. Someone please correct me if I am!

[shardison]

My only comment would be that, as long as your computer is fast enough, capturing at higher resolution and then re-encoding to a lower resolution always looks better than capturing directly to the lower resolution. (Like 640x480 to 352x240) Try it. You'll see. There's a good reason for it; when you encode directly to 352x240 you drop half the interlaced information right off the bat. When you start at 640x480, much of that interlaced information is combined and incorporated into the resized file. Movement is smoother and details are sharper.

[TheInformer]

warbird:

Check this thread out:
http://forum.vcdhelp.com/viewtopic.php?t=87984

[shardison]

I should read more carefully. My comments are about going from 640x480 to vcd resolutions (vertical resolution) Not quite the subject.

Still, I get better results if I do 640 horizontal instead of the conservative 352 horizontal capture x 480. And it's very true for hi-8 captures.

And there is no reason to capture 720 horizontal; that's more lines than the capture chip has.

[warbird]

thx for all of your replies.. I spent a couple of hours testing and ive noticed that the output mpgs look considerably better when capped at 720x480. When capped at 352x480 the quality just isnt as good, i notice more noise and images arent as solid and clear looking, and the detail isnt as fine.. i guess im going to just forget about huffy and use picvideo with manual quality settings (lum quality 1, chrom quality 2). i get around 3:9:1 compression that way so its not as lossy. An hour of capure is around 20-22gigs and i can handle that. I guess i got confused with all kinds of ppl telling me that it makes no difference whatsoever whether you cap at 720x480 or 352x480 because your just going to resize down to 352 anyway, when theyre wrong. With vhs it might not make much difference, but with tv there is quite a noticable diference.
I guess i just need to go buy a bigger hard drive


-------------------------------------------------------------------------------------
My specs:
Athlon Tbird 1400mhz/266
Abit KT7A-RAID
512megs Crucial pc133 cas2
All-In-Wonder Radeon 32mb agp
Turtle Beach Santa Cruz
30gig maxtor ata/100 7200rpm
Plextor 12/4/32 SCSI
400w pwr

Pioneer DV-343

[therick]

@warbird - it is defnitely ok to cap at 720 and go down to 480 or 352...as you have found out it DOES improve your video quality...

@Gameshowhost - capturing at 720 (or 640) and going down to 352 is a GOOD thing...you are wrong about it hurting video quality. Try it and see for your self...as warbird did. If you have the space, definitely capture at full resolution and go down...

[Shabubu]

Just be careful, remember that downsizing, especially with a Bilinear filter is in effect doing a very simple noise reduction. Check that you can't get the same effect with a 352 cap and a smoother filter.

Also, try sticking with huffy and capping slightly less than 720 (640 or 480 for example). The bilinear resize "noise reduction" will still kick in, you'll save space and drop less frames.

I did tests and I did find an improvement using higher than 352 (pal), but not much difference between 480 and higher.

Again it's personal preference, but huffy is definately the way to go if you can.

[kwag]

@warbird

You may also want to use the decomb.dll filter to do IVTC on your captures, to bring back the original 23.976fps.

That way you'll get the original picture as it was filmed.

Check this link:
http://www.inmatrix.com/articles/ivtcsynth.shtml

kwag

[therick]

I second that...decomb is a godsend

[warbird]

Shabubu - I use BicubicResize(352,240,0,.50), i dont like billinear because it tends to wash out colors and ghosts sometimes.. with avisynth, bicubic resize isnt limited .60, .75, 1.0 like with vdub. i find that .50 doesnt leave as much distortion around images that bicubic sometimes causes, and i like it way better then billinear.

kwag - Ive tried decomb a few times but it allways seemed to decimate wrong after a few mins into the clip and jumps very badly.. i dont know if dropped frames cause it or what, so i usually just use greedyhma with my svcds. I dont ivtc my vcds though, but ive been considering it. I use area based deinterlacer @65,0 interpolate with vcds. btw, i love your template! Its trully a godsend.. I dont know how many vhs' ive capped that ive fit onto a single cd, with hardly no quality loss. Great job!

[Shabubu]

Warbird:

Don't worry, I know the differences between Bilinear and Bicubic I was just saying that if he used the bilinear filter on his 702 cap that the quality might have been affected by the inbuilt "smoothing" this process performs. You don't get it with bicubic (bicubic sharpens in general, so especially for noisy caps the bilinear is probably the better way to go, good quality bicubic will be preferable). The ghosting is probably because your using the bilinear filter to effectively deinterlace your clip, if you deinterlace well first (disable the resize filter and run thru vdub) then resize the ghosting will be gone. The washed out coulours may be apparent if you cap in Mjpeg, there's a fix somewhere so I'll see if I can find it.

[warbird]

Oh.. ya, i use bilinear when the source is really noisy, but thats about the only time i use it.. Ive heard ppl talking about an avisynth resize filter called simple resize. Supposed to do no image processing, and makes the picture look better from what ive heard but im not sure if i believe it. Ive never tryed it myself.

-"The washed out coulours may be apparent if you cap in Mjpeg"

ya, i hate using mjpeg.. thats why i was curious about capping at 352x480 using huffy instead of picvideo, as ive never capped at that rez before.. In my case, if i want to use huffy i have to cap at 352x480 (not enough space for 720x480, 640x480, or even 480x480 most of the time). At 1st, i couldnt tell any difference between the 2, (thats why i wanted to hear other ppls opinions about it) until i capped a show at 352 that i cap regularly @720. Then i noticed the difference. Looked better capped at 720x480 with picvideo@18 then it did using lossless huffy @352x480. Go figure.

[Shabubu]

I acutally only started using simpleresize when I got DVD2SVCD, it's actually really nice, plus it's got support for interlaced sources and really helps your encode speed. Here's the link to that MFPEG fix for avisynth...
http://www.inmatrix.com/articles/ivtcsynth1.shtml

It's on the second page of the guide... good read too.