Joined: 13 Jan 2005 Posts: 2 Location: New Zealand
Posted: Thu 13 Jan 2005, 23:22 Post subject: DVD Overscan Problems
Much of my DVD source material is digitally recorded by capture card from local TV and directly encoded to MPEG-II during capture. (These days I use Winfast PVR as a computer based recorder). On playback, all the set-top players I have tried produce significant (around 8%) horizontal & vertical overscan. This is much greater than the normal live TV picture overscan (perhaps 4%) and results in significant loss of picture material, some titling etc.
Re-encoding with TmpgEnc and use of Video Arrange Custom Sizing settings around 672x536 pixels fixes this problem, but results in some loss of picture quality, and greatly increases DVD production time.
I would be very interested in any suggestions of how this problem might be addressed by hardware or software changes of any form at the time of recording. All my work is in the PAL standard. _________________ Reactance
SATISFY YOUR NEEDS, BUT SUPPORT THE NEEDS OF OTHERS ALSO
Joined: 04 Feb 2003 Posts: 587 Location: Lisboa, Portugal
Posted: Thu 13 Jan 2005, 23:52 Post subject:
DVD players don't change the image at all; it's the TV set that stretches the image and doesn't show the edges. Your problem seems to be that the image is being stretched twice - once during capture and once during playback.
I suspect your TV capture software is zooming in on the image during playback / capture (to hide the timcode, sync lines, etc.). If it doesn't have any option to turn this off, try other TV software (ex., DScaler).
Also, if you can capture to a less compressed format first (ex., HuffYUV AVI), and then encode to MPEG-2 using TMPGEnc, you'll probably get better results (real-time MPEG encoders are usually CBR and optimised for speed, instead of quality).
Joined: 13 Jan 2005 Posts: 2 Location: New Zealand
Posted: Sat 15 Jan 2005, 7:42 Post subject: DVD Overscan
Thanks for the thoughts. I downloaded a copy of DScaler & had a play with it today. (I had never heard of it.) It certainly is a Tweaker's Dream! Am not sure if it will enable me to capture at reduced size yet. I intend to play around with it further.
I have used Huffyuv & AVI capture in the past, but find it difficult to store much video on my computer in this format. The machine I now use for video is reasonably fast, and I usually capture at 8 Mb/s CBR using a high quality setting. Have not found single pass VBR to offer any further advantage here. My capture quality, is mainly dependant upon the quality of the TV transmission source material. (It is much sharper than VHS).
By the way, all the video capture setups I have seen here in NZ show this overscan problem once I point it out to people. Maybe the local version of PAL transmission has some peculiarlarities with respect to how much of the 64uS line width is used for picture content. ?? It does make it less attractive to turn directly captured video into a DVD.
One last question. Can you suggest some programs (editors or whatever) which identify the nature of each frame (I, P, B) as you look at it on screen, or advise what the GOP structure is in a video from an unknown source. (I am sure I have missed something obvious here.) _________________ Reactance
SATISFY YOUR NEEDS, BUT SUPPORT THE NEEDS OF OTHERS ALSO
Joined: 04 Feb 2003 Posts: 587 Location: Lisboa, Portugal
Posted: Sat 15 Jan 2005, 18:25 Post subject:
Most TV playback software enlarges the image, so that when you play it on your PC it looks like what you would see on your TV. This is not a problem with the transmission (if the transmission didn't include the overscan area, your TV wouldn't be able to display the image properly, because it's in that area that the sync signals are).
The problem is that, when you record it, the software captures the enlarged version, and not the actual source frames. You should have an overscan slider in DScaler (sorry, I don't have it here right now and don't know the exact menu option), that you can use to reduce or eliminate all resizing. To make DVDs, you'll want it set to zero.
If you want to capture directly to MPEG, naturally, using CBR at the highest bitrate gives you the best possible quality (VBR does not improve quality; it simply lets you keep the same quality while using a lower average bitrate - you still need a high maximum bitrate, for the complex scenes). But capturing to a less compressed format (ex., HuffYUV, DV, etc.) and then converting it to MPEG (at the same bitrate) with a better encoder, will produce better results. The bitrate alone doesn't determine the final quality. Some encoders give you better quality at 6 Mb/s than others can manage even at 8 Mb/s. Also, an I-picture-only MPEG stream at 8 MB/s will look a lot worse than an 1I-14P stream at the same bitrate, using the same encoder. I-picture-only streams really shouldn't be used at less than 25 Mb/s or so.
There are several programs that will give you information about a MPEG file's structure. One of the most popular seems to be Teco Bitrate Viewer, but if you look in the "utilities" section of videohelp.com, you'll find a few more (some are free, some have trial periods, etc.).
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum