Capture question for AutoGK input

[dvd_maniac]

I am using the 2.18b version of AutoGK and use it mainly to encode my recorded TV shows. I capture Mpeg-2 at 12mbit/s with 384 Audio and 720x480 resolution. These files are typically 5.5GB/hour and I encode them to 300MB/hour. Most of the encodes results around 400x304 to 512x384. I leave everything in AutoGK at default except I use CBR->mp3 at 96kbps (So I can later edit out commercials in VDub).

My question is wether I'd be better off capping at a lower resolution and using that resolution as a fixed width? Would that noticably impact quality or encoding speed?

[CWR03]

Capturing at a lower resolution will only reduce the quality of your final product, but it should not affect encoding speed. You can set a width in advanced options with AutoGK, if that is your concern. The best way to improve the quality in your case would be to aim for a larger file size.

[dvd_maniac]

Hi CWR03,

Thanx for a quick response. I do not think I stated my question the right way.

I am please with the current quality of my encodes now. However, what I am wondering is if I would see any drop off in quality if I recorded at 512x384 instead of 720x480 using the same 12mbit/s bitrate? I'm not sure but SDTV is not 720x480 is it? So isn't capping at that resolution wasted bits anyways? Wouldn't I get a better source by using a lower resolution? And if this is the case, wouldn't I be better off capping at the resolution that AutoGK usually resizes it to and set a fixed width? Would I see a slight increase in encoding speed if it did not have to resize?

[jggimi]

I'm not sure but SDTV is not 720x480 is it?You should review the Analog TV Capture Guide (http://www.doom9.org/capture/start.html), in particular chapters 3, 4, and 5.

[dvd_maniac]

hey jggimi,

I tried reading that a while ago when I was still using ATI AIW to cap with. A bit over my head though.
Now I use Hauppauge's Mpeg Hardware encoders and I do not know if they are still as revelant to my question. My main question being: Will I see an improvement in encode times if I do not have to resize?

[CWR03]

If you're encoding to the same resolution you were before, there will be very little or no change in the speed - it's the output resolution that affects that, not the source.

Regarding quality/source resolution, the only way to find out is to try.

[len0x]

My main question being: Will I see an improvement in encode times if I do not have to resize?

yes, for two reasons:
- decoding of MPEG2 input also takes time, so the bigger resolution the more intense CPU load is
- filters that are doing deinterlacing and resizing would have to work with less data per frame and will also be quicker

Having said that - the quality will degrade though: its better to have larger source and resize down that capture at lower resolution (mainly because resizing removes small artifacts and produces crisper image).

[dvd_maniac]

Having said that - the quality will degrade though: its better to have larger source and resize down that capture at lower resolution (mainly because resizing removes small artifacts and produces crisper image).

Thanx Len0x. I will stay with 720x480 then. I was thinking I would get a better capture using the same bitrate at a lower resolution and that I was actually LOSING quality by resizing, as well as wasting time. Thanx for setting me straight.

One more question. Would it be better to record straight to 44.1 instead of 48khz?

[jggimi]

On resolution, from our Capture FAQ. For NTSC, the vertical resolutions would be 480 and 240, rather than 576 and 288:9.) At which resolution should I capture?

· First let's have have a detailed look at the input. There are two different systems which are commonly used (PAL and NTSC). PAL (Phase Alternate LINE) is used in most European countries (France and some Eastern Europe countries use SECAM which is closely related to PAL). On the PC the TV picture can be displayed in various resolutions, up to the native PAL resolution (768x576). The maximum resolution you can use to capture depends on your capture card and most cards support 7**x576 where the horizontal resolution only depends on your capture device. In order to keep the proper aspect ratio of the picture you can capture either at full resolution (768x576) or at 1/4 resolution (384x288). If you intend to put the video on a VCD use 352x288 pixels, for SVCDs use 480x576 (both times the picture will be horizontally squeezed a bit and then unstretched during playback).
· Using 1/4 resolution has a distinctive advantage. The output will not have any interlacing effects making deinterlacing unnecessary. For more information about interlacing please refer to (De)interlacing - Basics (German link)and this http://www.doom9.org/ivtc-tut.htm. On the other hand using this resolution you will sacrify quality as you're not storing all the information of the original signal.
· No matter what resolution you chose make sure that you use one of the abovementioned horizontal resolutions.

For further informations on the subject refer to the following threads:

http://forum.doom9.org/showthread.php?s=&threadid=21067
http://forum.doom9.org/showthread.php?s=&threadid=18143
http://forum.doom9.org/showthread.php?s=&threadid=16013
http://forum.doom9.org/showthread.php?s=&threadid=43198

Deinterlace/Resolution:
http://forum.doom9.org/showthread.php?s=&threadid=40940

A great guide about finding the right resolution is "Der Karl's Capture Karten aspect ratio fuer Dummies ;-) " PDF (sorry only in German language): http://forum.doom9.org/showthread.php?s=&threadid=32604