Ways to increase rendered MPEG-2 Video Quality?

[Reciever80]

I'm using Ulead videostudio 10.0 SE (the one bundled with Easycap) and I can capture videos perfectly, and in perfect quality. Then once the video gets rendered, it becomes all pixelated, and was not the quality it once was. Is there anyway i can change this? I have it set to deinterlace output, and I have tried it at 4:3 and 16:9, at 640x480, 720x480, and 1280x720, however I still get the same problem. I see others that use easycap and they have good quality. What should i do to fix this problem? Anything you ask, i will do my best to fulfill, so please, shoot away.

[2Dogs]

OK, start by giving us the properties of the captured video you are starting with.

Right click on one of the clips, and post the clip properties.

Then right click on the best of your pixelated rendered output files, and give us its properties.

Ordinarily, you would be looking to convert the captured clips to 720x480 mpeg-2, using a video bitrate of between 4000kbps and 8000kbps.

You would also usually stick with the original aspect ration, either 4:3 or 16:9. There would be nothing to be gained from converting 4:3 source material to 16:9 or vice versa.

[Reciever80]

I dont really have any good captured video or bad captured video, but I can put up the properties anyway...

Would bitrate have anything to do with it?

edit: i found a clip that looked better than another one: I'll put up the properties..

O.o the difference was 16:9 and 4:3, although the clarity still doesnt match some of the other stuff i have seen. Here's the thing for the good one:

File Format: NTSC DVD
Video Type: MPEG-2
Attributes: 24 bits, 720x480, 16:9
Frame Rate: 29.97 frames/sec.
Data Rate: 5400 kbps

[2Dogs]

Those clip properties look pretty normal.

You haven't shown if the video is upper field first, lower field first or "frame based". If your intention is to output for playback on a pc, you should use frame-based, otherwise you should retain the field order of the source clips. If you get it wrong, the video can look jerky.

When you first insert a clip into a VS project, you should be prompted to match the project properties to the video clip.

You should get the best output picture quality by matching the source clip properties. In fact when you have things set correctly, VS should "smart render" the clips, which in effect simply entails writing them to a new file without any re-encoding.

So you should check to see that your output video file properties match the input file.

For example, my old P4 2.8c pc takes only a couple or three minutes to smart render a 48 minute mpeg2 file - a 60 minute TV show with the commercials cut out from it. Re-encoding the same video might take my pc about 75 minutes.

Re-encoding is unavoidable at any transitions, or where you use video filters or effects, or add titles.

[Ken Berry]

2Dogs -- Reciever80 said in his first post he had set it to 'deinterlace' the video, though he did not explain why... I suspect that in re-rendering it is applying one of the field orders, and this could be at least one reason why the output is not what he wants.

[Reciever80]

2Dogs -- Reciever80 said in his first post he had set it to 'deinterlace' the video, though he did not explain why... I suspect that in re-rendering it is applying one of the field orders, and this could be at least one reason why the output is not what he wants.

||

My field orders are Upper field, and for Deinterlacing video, I was under the assumption that it increased video quality.

[Ken Berry]

Can you tell us exactly when you set "it" to deinterlace -- during the capture process, or during the output/rendering process? What are the exact settings you used.

FYI, deinterlacing does not improve quality. If video started life interlaced (in your case, apparently using Upper Field First), then by and large you should leave it that way. It will play perfectly well both on your computer screen and on all TV screens, including the newer high definition 'progressive scan' HDTVs which can play native 1920 x 1080p (p for progressive, as opposed to i for interlaced) video. But essentially, progressive scan video is high definition video captured in a certain way. Otherwise, most amateur high definition video cameras are usually set to capture in i for interlaced mode too with no negative effects quality-wise in the final edited video from this. Progressive scan roughly equates to frame based in analogue video terms, and is not interlaced. But progressive scan HDTVs also have no difficulty reading interlaced video and playing it back perfectly.

So in a word, you have no reason to deinterlace. If you are wanting to use your captured video for, say, playback on the internet, then you would convert it to one of the highly compressed formats which are for the most part deinterlaced. Yes, okay... But you would not output it to deinterlaced mpeg-2 for playback on the web as it would be too large to be practical. If, however, you intend to burn it to DVD, then leave it as Upper Field First, do your edits, and output it also using Upper Field First. You will get best quality if you match the output properties exactly to the original capture properties.

But depending how long the project is and what the original bitrate was, the bitrate you use for rendering might need adjustment. If for instance, the capture bitrate was 8000 kbps, you could fit about 1 hour of that on a single layer DVD and get very high quality if you rendered using 8000 kbps. But if the project is much longer that an hour, then you would need to reduce the bitrate to around 6000 kbps, but you would still get good quality for around 90 minutes of video. (Equally, if the original capture bitrate was around 6000 kbps, it would mean automatically you could burn a 90 minute project to a single layer DVD.) You would in both cases get around 10 minutes more on a disc if you used one of the more compressed audio formats such as mpeg layer 2 or Dolby, instead of the standard LPCM.

[Reciever80]

I have it set to "deinterlace output" which i assume is the output from the TV to my capture card. ...But why would it matter size wise? nothing is more than about 2-9 minutes, so...

[Ken Berry]

Deinterlacing has nothing to do with file size. The file size is dictated by the format of the video (in this case mpeg-2) and the bitrate at which it is broadcast and captured. A higher bitrate means a larger file (and usually, higher quality). I find that most of my captures of standard def TV programs seem to use their apparent broadcast rate of around 6000 kbps.

[Reciever80]

My TV is a standard TV, although there is a way to record in wide screen, so I normally do that. It seems when I do that, it becomes slightly better, wheras if i recorded in 4:3, whenever the screen shifts position, it would pixelate a bit. However, what are settings that I can change, in general, that will increase video quality?

[Ken Berry]

OK, so we are back to your original question. The simple answer is to capture using the highest quality settings. Once you have captured, though, there is nothing you can do to actually improve the quality of what you have. All you can do is *maintain* that quality, and the best way of doing that is to maintain the same properties of the captured video throughout your project and in the final render. And implicit in all that is the message: don't deinterlace!

[Reciever80]

awesome! I'll try that, and then I'll see how it works. If something is wrong I'll make sure to report back here. Thanks!