Decline in image quality from clip to dvd

[Bill Slammon]

Sorry if this has been asked about already. I did several searches with no success.

I am finding a notable degradation in image quality from the clip (when previewed in Vegas) to the final product on DVD using Architect 4. The footage is being shot with a Sony A1. I record in HDV and downconvert to standard definition in camera when captured by Vegas 7. I then render the final edit to mpeg2 and ac-3. It's all widescreen footage. I have tried several of the template setting for rendering a DVD Architect 4 ready file. I have chosen the best quality render where ever possible. Still, the image quality of the video once burned to the dvd is not as clear and crisp as the vidoe clips are when previewed from the project media window. I am new to this, so would attribute the the video quality to my skills if the image wasn't as clear and clean as it is when previewed. (Note: I am talking about the project media window preview, no the timeline preview.)

Am I missing something? Is this the best I can expect? Or, are there settings or render formats (e.g., mpeg 2, avi, etc.) that would produce better dvds? I know there are things I could do that w

A possibly related questions: Does it make sense that a 1 1/2 hour edited video with audio and no major effects would take up about all the space on the DVD (this project is coming out at 4.5 gigs)?

By the way, the DVD image quality is the same whether the disc is played on a computer or in a DVD player connected to a tv.

I am working on a year old HP laptop with 2.13 Intel M processor, 1 gig RAM, and 125 meg ATI video card. I have had no problems running Vegas 7 or DVDA4

Sorry for the basic question. I am just getting into all this and want to make the best looking stuff possible. Thank you in advance for any advice.

Bill Slammon

[Carl Downs]

I can render out a 2Mbs .wmv file strait from the timeline and get a lot better quality than a "Main Concept" .mpg DVD at 7Mbs when playing on the PC monitor of course (playing DVD on monitor too). I agree with you... I thought the video quality was going to be at least a little better (I shot with a Z1U HDV, editing all with ConnectHD .avi files) but after using the template to export from vegas (and by the way, DVD Architect DID NOT have to re-render so template was correct), burning a DVD and watching it on my TV... it... sucks... just regular old DV quality. Just as a test I rendered a 6Mbs (not even 7 or 8) .wmv HD template file and man... now that is NICE!! yet the Mbs and file size is %20 less than the DVD?? I don't understand... Now... I know Hollywood uses real film and all, but they too have to render to SD and their DVDs look really nice. I know I know... they have very expensive machines and codecs to render there products but... I thought HDV rendered to SD would be a slight improvment since staring with a lot more data. Is it DVD architect and/or the "Main Concept" codec that is no good? Is it just me doing somthing wrong? Is there a suggestion? Another Method?

[Chris Barcellos]

1. One and a half hours on a 4.7 gig DVD is a reduction in bitrate from Hollywood movies. They are on a commercial dual layer disc, usually running about two hours.

2. I do not know what your work flow is, but I have been editing in HDV in Premiere Pro or in Vegas, then rendering to SD DV, and converting from there, to the final DVD files.

In my latest project, I did have some issues. I would normally keep every thing interlaced, but the picture just was losing to much, so when I rendered from HDV to the SD DV .avi file, I did it deinterlaced. The resulting video was much better in terms of sharpness.

3. I also have just downconverted from camera, shooting HDV, and then using camera downconversion capability. I end up with better video from the FX1 than from the VX2000 that way.

Remember, though, standard DVD is standard definition, and while it can be tweeked, it will never be high definition.

[Bill Slammon]

Thank you, Chris.

I still have a question, though: Why do the in-camera downconverted (HDV to SD dv) clips look so much better when previewed in Vegas 7 than the resulting mpeg2 DVD? This is SD dv to SD dv rendering. I would think there shouldn't be so much loss. Does the loss in image quality come from the mpeg2 render?

I'll try capturing HDV and then render to SD DV and see if there's a difference. Never thought of that. Thanks. Any other tips?

Which deinterlace method did you use, "blended" or "interpolate"?

You also answered another question I had about the quality and quantity of video on Hollywood dvds. Thanks.

Bill

[Seth Bloombaum]

Well, this may not be happening, but if it is it can have a pretty dramatic impact on quality.

A 4.5GB MP2 plus AC-3 audio is perhaps too close to the maximum size of a DVD. DVD-A might recompress the video in these circumstances, which will indeed be a huge quality hit.

So, assure that DVD-A is not recompressing by carefully reviewing the dvd-prepare dialog boxes, look for those messages in the "review message list" that indicate whether your video is being recompressed.

If so, you'd need to either do another compression run in Vegas, or, give DVD-A a DV-25 AVI file for it to compress to size.

[Bill Slammon]

Thank you, Seth. DVD-A did recompress the file once. But the last few runs, and I've tried several to try to get the image quality up, the final mpeg2 file was near 4 g, and, according to DVD-A, the AC-3 brought it up to around 4.5g. With the last two runs, DVD-A said everything was good, no recompression necessary. By the way, I've achieved the same lower quality results on smaller files of about an hour in length.

I appreciate your suggestions, though. Thank you.

Bill

[Christian de Godzinsky]

Hi,

I am puzzeld about the same problem - quality loss. Preview looks great, and after MPG2 conversion the video quality drops a lot - too much compared what Hollywwod is able to put on those DVD's. When I say preview I refer to FireWire preview via my CAM out to a good quality analog CRT-monitor.

First I thought that the reason is the fairly low bitrate (only about 6Mbit/s) to fit my 90 minute video on a single layer DVD.

Normally I render in Vegas to best quality AVI and the let DVDA do the final rendering to fit the DVD with the max bitrate possible. To test this if it has to do with bitrate I rendered just a 30 minute part of the same project to DVD - using DVDA and the highest possible bitrate.

There is still a very visible quality loss in the video. I am talking about lots of still scenes with little movement. Here the compressed MPG2 compression should excel and still produce good sharp images. But no.

I am very disappointed that I cannot get that good quality that I see in the preview to the DVD. What is wrong? Maybe the MPG2 decoder is nog that good in DVDA? BTW - I am working in PAL.

So you have another person here to join the group of people not satisfied with the quality. What else can I test? Maybe some of the very experienced moderators could throw in a comment or two??

Christian

[Chris Barcellos]

Originally Posted by Bill Slammon

Thank you, Chris.

I still have a question, though: Why do the in-camera downconverted (HDV to SD dv) clips look so much better when previewed in Vegas 7 than the resulting mpeg2 DVD?

My simple explanation is this, but don't ask me about the technical aspects:

Because Mpeg2 is a further compression of the video information. So you are taking compressed HD in the form of HDV, reducing it to DV in camera. Information is lost in the compresssion. Then you take the Resulting DV material, and compress it more, more information is lost. Plus, you introduce in .mpeg the necessity for the player to reconstruct each frame.

I'll try capturing HDV and then render to SD DV and see if there's a difference. Never thought of that. Thanks. Any other tips?

Which deinterlace method did you use, "blended" or "interpolate"?

Camt recall whether I used blended or interpolated. Try small clips of each to get a range.

Tips: Experiment and report back. We are all looking for better ways to do things. Thats what this forum is about.

[Seth Bloombaum]

Some general pointers about shooting for later compression. Any particular point may not apply to your project.

Comparisons to Hollywood feature films are bound to disappoint. Features are shot under very controlled conditions, and are compressed with dedicated hardware. You'll need to do the same to get similar results. Laser Pacific is one of the L.A. digital post houses highly regarded for compression and other services. It's worth a visit to their web site.

Comparisons to camera previews and DV25 file previews won't be close. Comparisons to MPG2 previews to an NTSC monitor via firewire should be very close, but bear in mind that computer display of your MPG2 preview is actually a hardware-assisted upconvert to HD-like resolutions.

Compression throws away most of the bits to fit within the DVD datarate and size specs. The SD-DV that you're ingesting is 5:1 compression. MPG2 for DVD is at least 25:1 compression and frequently higher. There are techniques to assure that what's left is most desireable. The primary concern is to reduce the number of pixels that change value on every frame.

1) Avoid video noise in dark areas, typically produced in low-light shooting and especially with added camera gain. One method to deal with this after the shoot is (very) careful use of the black restore filter in Vegas. Other techniques include blurs and denoisers, but they can very easily reduce resolution.

2) Avoid rapid camera motion, and/or jerky hand-held shooting.

3) Use the exposure tools you have available in the camera in conjunction with lighting to produce well-exposed tape. Use the scopes in Vegas to assure that you are working within NTSC limit of 100 IRE. Use color correction, levels, or brightness and contrast to bring clips into spec.

4) There are scenes that will stress the weaknesses of the compressor. For example, a tree that fills the frame, leaves blowing in the wind, or other scenes in which every pixel is constantly changing. The tree might be particularly challenging depending on lighting, if the leaves are constantly going through a broad range of color and brightness values. Rippling water might change every pixel as well, but through a narrower range of values.

Acquisition in HDV for DVD release is a good workflow, but won't increase apparent sharpness much. These issues may not be relevant to your particular video, your mileage may vary.

***edit***
Be sure to look at this thread http://www.dvinfo.net/conf/showthread.php?t=64309 for some great explanations about VBR and CBR.

[Bill Slammon]

Chris and Seth, thank you both so much. This has been a very informative discussion. I really appreciate the time and thought put into your replies.

My plan is to in fact keep experimenting with shorter pieces and capturing in HDV and play with it to find ways to get the best results.

By the way, what method will produce the best quality video if you want to output to other than DVD, like back to tape for example? In other words, what SD format in Vegas 7 retains the highest output quality? I'm sure the answer begins with "it depends", but I'd love to know what you all think.

Thank you.

Bill

[Jim Montgomery]

Try this workflow

Shoot in Cine 30

Capture footage using HDLink with de-interlace checked

Open a timeline using HDV 1080-60i. On the video tab change the field order to none (progressive).

Render as mpeg 2, template = DVD Architect NTSC Widescreen and in "custom" change the field order to progressive only.

Jim

[Bill Slammon]

Thank you, Jim. I'll try it.

Thanks to all for your input.

Bill

[Christian de Godzinsky]

Hi,

Should we come to the conclusion that the DVDA MPEG2 encoder never can produce the same quality as some hardware and/or hollywood based encoders - independent of how good quality material you throw at it?

The Vegas manual mentions that the o n l y way of producing broadcast-quality material is to stick to DV alias AVI files!! Yes, that is what is says. That silently indicates that the MPEG2 encoder is not broadcast quality (in Vegas and/or DVDA).

And many tv BROADCASTS (if not all - excluding HD) are MPEG2.... There is something that I don't get here... Sony claims that this IS a professional product...but it seems that it is lacking a professional MPG2 encoder...

Regards,

Christian

[Seth Bloombaum]

Encoding MPEG2 for DVD and encoding for broadcast transport streams are two different animals entirely.

Vegas' MPEG2 codec, licensed from Main Concept is as good as any and better than most software encoders for DVD specs.

DVD spec MPEG2 video is in no way "broadcast quality", broadcast transport streams commonly used for program distribution range up to 50 Megabits per second bandwidth, DVD is (about) 5 Mbps.

Should we come to the conclusion that the DVDA MPEG2 encoder never can produce the same quality as some hardware and/or hollywood based encoders - independent of how good quality material you throw at it?

IMHO the encoder is quite well suited to the market it was created for. Really, it is excellent. However, it is not a magic box. Whether we're talking about high-end hardware or inexpensive software, how the operator uses the technology has everything to do with the final product. For example, best results will usually not be obtained with DVDA's encode of a DV-AVI, but by compressing in Vegas (using the same codec), where the operator has many controls of how the material is compressed.

Likewise, it takes considerable skill and experience to compress hollywood features for the mass-market.

The Vegas manual mentions that the o n l y way of producing broadcast-quality material is to stick to DV alias AVI files!! Yes, that is what is says. That silently indicates that the MPEG2 encoder is not broadcast quality (in Vegas and/or DVDA).

That's right. No MPEG2 encode for DVD is broadcast quality. Some would say that anything shot in DV or HDV is not "broadcast quality" (not me). The fact is that some broadcast networks will accept DV and/or HDV acquired material and some won't. But people argue about this, nevertheless. I'm sure a search of the forum archives would turn up lots of these arguments.

The real question is whether you are acquiring in the right format, using the right post-production workflow, distributing in the right format, and applying appropriate technique and craftsmanship along the way for your target audience.

Any other discussion is almost always about "mine is better than yours."

[Mike Kujbida]

My 2¢ worth here.
I've done a number of DVDs that are around the 2 hr. mark (some with lots of motion and some with very little motion).
In all cases, I've been quite happy with the end result.
I do shoot with a JVC 550 (1/2" 3 CCD) camera which definitely helps.
I also try, as much as possible, to ensure that the scenes are properly lit as I find that makes a big difference. Noise is the enemy of any form of encoding.
The old saying GIGO (Garbage In, Garbage Out) is very true for a process such as this.
When you think about it, you're taking a file that's (approx.) 26 GB. in size for a 2 hr. video and squashing it down to 4.3 GB., including audio.
With all the compression happening, something HAS to give and it's the picture quality that suffers the most.
When the "big boys" do it, they're using hardware encoders and quite often doing it a scene at a time in order to maximize the end result. They also have twice the storage capacity we do so there's a lot less compression going on.