How To Capture PC Video In High Quality To DVD And Blu-Ray Disc

[blazin-j]

Hello All,

I am new to the board and may be asking a question that has been asked before. I did some searches before posting so please don't flame me. The reality is this is a HUGE message board and difficult to find the answer(s) you are looking for.

I am capturing video from my PC monitor using FRAPS at 30 fps and saving it as an uncompressed .AVI file so I do not lose any quality. The video is being captured from a 720x480 pixel window. This is because I want to export the video onto a DVD-R and I don't want to lose any quality downsizing.

I have tried Adobe Premiere Pro 7 and Cyberlink Power Director 5 thus far, but the problem is the MPEG2 codecs that come with the program compress and cause the resulting DVD video to look crappy (i.e., the lines are not as crisp and clear, resulting artifacts occur, objects don't move smoothly but jitter somewhat across the screen, colors are not as vibrate but appear washed out).

Is there a better way to do this? A step-by-step process, including what tools to use, would be appreciated.

I simply want the video on the DVD to look exactly like it does in the uncompressed .AVI file.

Also, the next step would be to capture the video at a higher resolution and the save it to a BD-R. I know I'll need a Blu-Ray burner and associated media, but what I am not sure of is what resolution HD/Blu-Ray is? In other words, instead of capturing at 720x480, what would I capture at? My monitor can display 2560x1600.

Thanks in advance for your help. I respect the vast expertise of the members on this board. Please remember too that eveyone was once a newbie.

[aedipuss]

you might try upping the fraps setting a little. go full sized 720x480 at 60fps. there shouldn't be much loss of quality going to 720x480 mpeg-2 30i that way. blu-ray hd is 1920x1080 which is above what is allowed in fraps.

cce, mainconcept, tmpgenc, procoder, henc, etc. are all other encoders you might try.

convertxtodvd is pretty good at making usable dvds from weird sources like the fraps codec.

[Baldrick]

1.
How many minutes are you trying to squeeze in on a DVD?
What settings are you using?
You should be able to get very good quality if you use high bitrates(6-8MBit/s) and you can then fit around 1 hour on single layer DVD. You could also try other mpeg2/dvd converters and compare like the very good hcenc.

2.
Read www.videohelp.com/hd#tech , you can use up 1920x1080p with blu-ray video.

[blazin-j]

you might try upping the fraps setting a little. go full sized 720x480 at 60fps. there shouldn't be much loss of quality going to 720x480 mpeg-2 30i that way.

OK, will do and see if this produces a smoother animation.

1.
How many minutes are you trying to squeeze in on a DVD?
What settings are you using?

Current project is about 1 hr.

As far as settings, are you referring to FRAPS? I am using full size, 30 fps.

If you're referring to the codec in Premier Pro, I tried both the NTSC DV High Quality 7Mb CBR 1 Pass setting as well as the NTSC DV High Quality 4Mb VBR 2 Pass setting.

In Power Director, I just used the High Quality setting for DVD burn. I don't remember it giving me many choices, so I chose the best available.

Any suggested programs for transcoding to DVD other than Premier Pro or Power Director? Something that will use uncompressed .AVI files as input?

[aedipuss]

1 hour of dvd mpeg-2 can be encoded at 8000kbps and keep the filesize under 4.3gb.

the film machine is another choice to try.

[blazin-j]

1 hour of dvd mpeg-2 can be encoded at 8000kbps and keep the filesize under 4.3gb.

I'm not too concerned about filesize since I can burn a 8.5Gb dual-layer DVD+/-R, if necessary. Quality is more important than filesize. So, does that change your encoding recommendation?

As for the FRAPS setting of 60FPS, my PC can't capture at that rate. I can only get 40-45 FPS. It's a Pentium 4 that runs about 3.2Ghz with 2 GB DDR2 RAM. My hard drive is a Raptor, which should be fast enough. I'm hoping to do this without investing in more hardware.

[jagabo]

Be aware that MPEG uses 4:2:0 subsampling for the image data. This means the gray scale image is saved at 720x480 but the color information is stored at only 360x240. So colored text (or other small details) will look much worse than b/w text -- especially with colors of similar luminance:



The original color text was as sharp as the b/w text. This is after conversion from 4:4:4 RGB to 4:2:0 YUV but without any MPEG compression or low pass filtering that a DVD Player would apply.

[aedipuss]

1 hour of dvd mpeg-2 can be encoded at 8000kbps and keep the filesize under 4.3gb.

I'm not too concerned about filesize since I can burn a 8.5Gb dual-layer DVD+/-R, if necessary. Quality is more important than filesize. So, does that change your encoding recommendation?

As for the FRAPS setting of 60FPS, my PC can't capture at that rate. I can only get 40-45 FPS. It's a Pentium 4 that runs about 3.2Ghz with 2 GB DDR2 RAM. My hard drive is a Raptor, which should be fast enough. I'm hoping to do this without investing in more hardware.

ok. set fraps at 29.97 720x480 and encode to mpeg-2 at 9000kbps. give some thought to trying some of the other encoders. cce being about the best, then mainconcept and procoder, with the rest - tmpgenc, henc, etc as ok. my opinions only of course.

vegas pro 8 may still have a 30 day trial, it uses the mainconcept mpeg-2 encoder and would be a good choice, but it's expensive. convertxtodvd has a free trial w/watermark, does surprisingly well and is fairly cheap.

[blazin-j]

Be aware that MPEG uses 4:2:0 subsampling for the image data. This means the gray scale image is saved at 720x480 but the color information is stored at only 360x240. So colored text (or other small details) will look much worse than b/w text -- especially with colors of similar luminance:

I understand what you're saying, but it makes me wonder then how Hollywood or even small business film/animation companies make their videos look so crisp and clear. Seems there has to be a way. Is it simply a matter of h/w or s/w equipment; that is, commercial equipment vs consumer products?

Aedipuss, thanks for the tips. I'll have to try these programs and encoders out and see which tests produce the best results.

[jagabo]

I understand what you're saying, but it makes me wonder then how Hollywood or even small business film/animation companies make their videos look so crisp and clear.

They don't. Have you ever watched the closing credits on DVD? They're never sharp and crisp like a computer monitor. When text needs to be readable on a standard def TV it is large.

Some things you can do:

Make sure any text is well antialiased. If necessary, capture at a larger resolution and downsize. This may give you better antialiasing.

Make sure there is a lot of luma contrast between text and the background. If possible, use white text on a black background or vice versa.

Use fonts without serifs. Those tiny little serifs will flicker on SD TV.

Capture at 23.976 fps and encode progressive with 3:2 pulldown flags. Don't capture at 29.97 fps and encode interlaced (to do so further blurs the color channels).

Use sufficient bitrate to avoid loss of small detail.

Don't have too much stuff moving around or changing brightness and colors. One of MPEG's most powerful tools is to only encode the changes from frame to frame. Parts of the frame that don't change consume almost no bitrate. If everything is changing the video will require a lot of bitrate.

That also means you want low noise captures. That probably isn't a problem for your screen caps on a computer. If your source is low noise, don't use a spatial noise reduction filter.

[blazin-j]

Jagabo, thanks for sharing your wisdom and experience. I did have one question, however.

Capture at 23.976 fps and encode progressive with 3:2 pulldown flags. Don't capture at 29.97 fps and encode interlaced (to do so further blurs the color channels).

I'm assuming you mean this should apply if I am rendering text only over a static background (as in credits)? I am capturing the video from my PC at 30fps currently, as described above. If I wanted to overlay text on top of this, then would what you say still apply? The 'background' animation will not be static, but fairly 'busy'.

Finally, is it better to encode progressive vs interlaced? Will the viewer will have to have a progressive scan DVD player if I encode progressive?

[jagabo]

is it better to encode progressive vs interlaced?

Interlaced frames should be encoded interlaced. Image quality is always better if you encode progressive frames progressive. Colors get more smeared if you encode progressive frame as interlaced. Here's an 8x enlargement (point resized so you can distinguish each pixel) of some small red RGB text on a blue background. The top is the original RGB, the middle is encoded progressive MPG, the bottom interlaced MPG:



Normal video doesn't have such sharp detail so the difference is less visible. Here's a web site that discusses the old DVD chroma upsampling error. They also talk about this fundamental problem with interlaced MPEG encoding with 4:2:0 subsampling. See the "4:2:0 Interlaced: Fundamentally Broken" section.

http://www.hometheaterhifi.com/volume_8_2/dvd-benchmark-special-report-chroma-bug-4-2001.html

Will the viewer will have to have a progressive scan DVD player if I encode progressive?

No. The analog signal that travels over composite or s-video is always interlaced (one field is sent at a time, 59.94 fields per second). The DVD player will peel the two fields apart and send them one by one whether the MPG data is encoded is progressive or interlaced. A progressive player will send the progressive frames to the TV as progressive images.

Capture at 23.976 fps and encode progressive with 3:2 pulldown flags. Don't capture at 29.97 fps and encode interlaced (to do so further blurs the color channels).

I'm assuming you mean this should apply if I am rendering text only over a static background (as in credits)? I am capturing the video from my PC at 30fps currently, as described above. If I wanted to overlay text on top of this, then would what you say still apply? The 'background' animation will not be static, but fairly 'busy'.

Any time you want to preserve small color details you want to encode progressively. The lower the frame rate you use the lower a bitrate is required too. You have to balance the need for smoothness (higher frame rate = smoother video), readability of any small text, and compressibility.

[blazin-j]

Thanks again, Jagabo. The examples you provide are compelling and this is making a lot of sense to me, but I still have a few remaining questions.

Interlaced frames should be encoded interlaced. Image quality is always better if you encode progressive frames progressive. Colors get more smeared if you encode progressive frame as interlaced.

.
Question: How do I know if the original video captured by FRAPS on my flat panel is interlaced or progressively scanned? When I think of interlaced, I usually think of tube TVs that used phosphor screens. But if you have a computer flat panel monitor, can I assume it is progressive?

Any time you want to preserve small color details you want to encode progressively. The lower the frame rate you use the lower a bitrate is required too. You have to balance the need for smoothness (higher frame rate = smoother video), readability of any small text, and compressibility.

I do want to preserve small color details, so it sounds like progressive is the way to go. But I am still unclear about what process/settings I should use to optimize the video capture and DVD encoding. Am I correct to assume that I should capture the original video in uncompressed .AVI at the highest possible fps rate my h/w can support, at 29.97 or at 23.976? I don't even think FRAPS will let me capture at 23.976 fps so wouldn't I have to change the fps speed with some other program?

Any extra clarity you could provide would be greatly appreciated.

Just keep in mind that the video I am capturing is highly dynamic, not static, and I want the resulting DVD to display the best possible quality; that is, smooth, clear, crisp, vibrant colors and text. Compressibility and filesize are not important considerations to this particular project.

[jagabo]

How do I know if the original video captured by FRAPS on my flat panel is interlaced or progressively scanned?

Almost anything you capture with FRAPS will be progressive. The only exception I can imagine is an interlaced DVD or AVI playing without any deinterlacing. You will see comb artifacts if the video is interlaced.

I do want to preserve small color details, so it sounds like progressive is the way to go. But I am still unclear about what process/settings I should use to optimize the video capture and DVD encoding. Am I correct to assume that I should capture the original video in uncompressed .AVI at the highest possible fps rate my h/w can support, at 29.97 or at 23.976? I don't even think FRAPS will let me capture at 23.976 fps so wouldn't I have to change the fps speed with some other program?

How fast is the thing you're capturing? Is it generating 60 different frames per second? 30? Less? More? Capturing at a frame rate different than the source will lead to jerkiness. If you view one of your 30 fps caps frame-by-frame you can get an idea if the source was generating less than 30 fps -- you will duplicate frames. If it was generating more than 30 fps, and not a multiple of 30 fps, you will see jerks in the motion. If you post a short sample I can take a look at it.

One thing I'm not sure about is whether 29.97 fps progressively encoded MPEG2 is legal for DVD. And if not, if it becomes legal if you use DgIndex to apply 2:2 pulldown flags.

[blazin-j]

How fast is the thing you're capturing? Is it generating 60 different frames per second? 30? Less? More?

Well, I'm not sure, but I can tell you that when FRAPS is running (but not capturing), there is a yellow number in the corner that I believe indicates the fps -- it is about 55 or so. But once I start capturing and the number turns red, the number decreaes to about 40-45.

One thing I'm not sure about is whether 29.97 fps progressively encoded MPEG2 is legal for DVD. And if not, if it becomes legal if you use DgIndex to apply 2:2 pulldown flags.

This flew over my head. Do you mean legal as in violating some law or as in not being compliant with some MPEG2 standard?

Different question: Does the monitor you are using make any difference in the capture? For example, would a CRT monitor capture look different from a flat panel capture? Or is it capturing from the video card and the monitor you use to view irrelevant to capture quality?

[jagabo]

Well, I'm not sure, but I can tell you that when FRAPS is running (but not capturing), there is a yellow number in the corner that I believe indicates the fps -- it is about 55 or so. But once I start capturing and the number turns red, the number decreaes to about 40-45.

Since what you're capturing isn't generating frames at 30 fps, or an integer multiple of that, capturing at 24 fps probably won't be noticeably more jerky than capturing at 30 fps. Even decimating from 30 fps to 24 fps may look ok. How critical is playback speed? You could also slow the 30 fps video down to 24 fps making it 20 percent slower.

One thing I'm not sure about is whether 29.97 fps progressively encoded MPEG2 is legal for DVD. And if not, if it becomes legal if you use DgIndex to apply 2:2 pulldown flags.

This flew over my head. Do you mean legal as in violating some law or as in not being compliant with some MPEG2 standard?

I meant "legal" as in supported by the DVD player specification. I know that 23.976 fps progressive with 3:2 pulldown flags is supported. And 29.97 fps interlaced is supported. But I don't know if 29.97 fps progressive is supported.

Note that analog NTSC video is always transmitted as 59.94 fields per second. One field at a time is transmitted down the wire or over the air. When you watch on a interlaced display you see one field at a time. Frames don't exist until that video is digitized and pairs of fields are woven together. At that point you have 29.97 fps interlaced video.

DVD players can be rather dumb devices. They don't know how to convert 23.976 fps progressive frames to 59.94 fields per second. 3:2 pulldown flags are instructions added to the video stream that tell them how to produce the requisite 59.94 fields per second.

Someone discovered that it's possible to use other pulldown patterns and these playback properly on most DVD players. For example, 25 fps video can have 3:2:3:2:2 pulldown flags to create 59.94 fields per second. 20 fps video can use 3:3 pulldown flags. The program written to add these unusual flag patterns is called DgPulldown. So, if 29.97 fps progressive isn't officially supported by the DVD spec, you could add 2:2 pulldown flags with DgPulldown.

Different question: Does the monitor you are using make any difference in the capture? For example, would a CRT monitor capture look different from a flat panel capture? Or is it capturing from the video card and the monitor you use to view irrelevant to capture quality?

Generally, the monitor doesn't make any difference. The video data is captured (copied from) the graphics card's frame buffer. There is one case where the monitor can make a difference. It's possible in Windows to sync a program to the refresh rate of the monitor. Many games and media players use this sync option to prevent updating the screen while it is being displayed. If this isn't synchronized you will get tearing -- part of the image you see will be from one frame, and part from another. If your program is timing its screen updates this way there would be a difference in smoothness when captured at different frame rates.