This is a question for the tech heads in the community.

I did an experiment last night. I rendered a project to a .avi file in Vegas 9, then placed that .avi file in DVDA 5 to burn to disc.

The project was set at 8mbps, in DVDA's properties.

Once the disc was burnt, I went back to the original timeline and used Vegas 9 to render it to MPEG2, again at 8mbps. Then used DVDA to burn that MPEG file to disc.

Checking both discs on a 42 inch plasma, the vision quality of the first disc was visibly superior to that of the second.

Any ideas why?

I was always under the impression that the Vegas MPEG engine was superior to that of the one in DVD Architect. Having seen the difference with identical vision on the same screen, I am swayed to change my workflow and let DVDA do the MPEG conversion.

Thoughts?
Arguments for and against?

Ben

Hi Ben

Many moons ago I discovered that doing MPEG2 and then allowing DVDA to re-render gave me inferior DVD's. I may be wrong but it might be something to do with the GOP they use. Typically DVD video has a GOP only every couple of frames and even Vegas itself does a lousy job of re-rendering MPEG2 video unless it has a GOP of 1

I swapped over to using DVD Lab and the results were way better!!! Now I render in Vegas and author in DVD Lab.

Instead of getting technical why not try another DVD authoring package and compare the DVDA DVD and a DVD made from another package?? That way you can isolate whether the problem lies in DVDA's rendering of an MPEG2 file. Otherwise you just might have to render out AVI from Vegas.

Chris

Thanks Chris...
I might do just that, but the vision off the DVDA encoded disc is almost as good as watching the vision out of the composite port on the camera..

It makes me want to further explore possibilities... What was astonishing was there was naff all motion artifacts on the DVDA version. certainly it took an hour and half for DVDA to burn a 26 min project (I only have an old 3.4Ghz chip), but the result is stunning..

Ben

Can you post a screen shot of your Mpeg2 settings that you used to make the second disk? This is quite curious and might cause me to change my workflow as well. Would save me a step in my DVD workflow.

Okay I'm lost. I'd always heard you're supposed to render the video to mp2 and the audio to ac3 and import both clips to dvda. But you're feeding dvda a avi file and the results are superior? What kind of file is your camera recording natively and are you editing in that native format then rendering to avi?

People go wild for what someone else tells them about quality. Kudos to Ben for putting "what everybody knows" to the test.

The shocking truth is that the rendering engine in DVDA and Vegas are the same.

What is different is that DVDA will automatically choose settings that it thinks is best. With Vegas, you get to choose settings.

Something as simple as DVDA using Variable Bitrate encoding (VBR) but you choosing Constant Bitrate (CBR) in Vegas could account for the differences in picture quality, even at the same overall bitrate.

DVDA's defaults are OK, especially for up to 60 minutes of content.

THE WORST quality may be acheived by making a mistake in the Vegas MPEG2 encode sufficiently large that DVDA says "I can't put this on a standards-compliant DVD, must re-encode it to MPEG2". You'll know this has happened, because it will look horrible, it will take extra time to build the disk while the video is re-encoded, and DVDA will tell you it's going to re-encode if you read the messages closely.

Okay so need to render separate mp2 and ac3 files, just render to avi and import to DVDA and make DVD?

Hi,

I tried many things to make a better quality. You can dump .avi or m2t file and DVDA will handle it, but it will take longer to re-render (if Im not mistaken DVDA is a single core app).

I took a different route. I use the best setting (in vegas) while rendering to DVDA (mp2 and aac). In DVDA i burn to .iso file. The Iso file size is about 6GB. Then I use DVD Shrink program which is free DVD Shrink v3.2.0.15 (freeware) (http://www.afterdawn.com/software/video_software/dvd_rippers/dvd_shrink.cfm)
and hack my own DVD by compressing (automatic setting approx 10 to 15% compression) DVD. The compressing process takes approx 2 to 5 minutes and builds an another iso file (4.7gb) which i can burn to dvd disk.
This method will eliminate guesswork on final size of the file and it looks better (in my opinion)

while I agree that DVDA may have a decent render quality when doing small projects, I prefer to do all rendering (mpg and ac3) in Vegas by setting the right bitrate for the size of the project. All are VRB and some are 2 pass depending on the job. I test run all my DVDs on at least a 42" screen and have never had a problem with quality. I've had them run on big screens (10' wide) and they look fine.
I have also put as much as 3 hours on a DVD and while I really had to crush it, it looked pretty good on the 42 to 52" screens.

I never go above 8,000,000 as a top BR and my lowest BR is set to 2,000,000. Its the average BR that gets changed unless I'm doing CBR.
I'ts been working for me since DVDA 1/Vegas 3

When you are trying squeeze the maximum content onto a DVD, it really helps the encoder to have low noise video. Noise really increases the burden on an encoder. It has to use a great deal of the available bitrate to encode the noise which is high frequency and changes frame to frame. This leaves less available bitrate for the actual video content. It's worth it to use a product like Neat Video to remove the noise first. The resulting encoded video quality difference is surprisingly better when you have "bit rate starvation" on a 1+ hour DVD. The only problem is that Neat Video is processor intensive which results in a slow render.

Another option to consider is using TMPGEnc via the Frameserver from Vegas. I've always been very happy with my results. It also includes a noise filter, if you need it, that can take advantage of CUDA processing on a modern nVidia graphics card for speed improvements in the heavy processing noise filtering.

Both the older TMPGEnc Plus 2.5 and the latest TMPGEnc 4.0 XPress work great with the debugmode frameserver out of Vegas Pro 8 or 9.

Mmmm.. I wasnt aware the MPEG engine was the same in Vegas and DVDA..

But to help things;

Perrone, The vision is shot on a Sony DSR-570, http://img.photobucket.com/albums/v355/Cally460/DSR-570.jpg
and ingest into the system is via a Sony DSR-11 VTR linked using firewire.
Ingest is done using Sony Video Capture 6e, which is a part of Vegas 9.

In the media manager, the files are listed as an .avi file, with the following attibutes
http://img.photobucket.com/albums/v355/Cally460/DVfileattribute.jpg
and are cut on the timeline as such. Rendering is done using the following:

Rendering to MPEG2 in Vegas is done using the following preset;
http://img.photobucket.com/albums/v355/Cally460/VegasMPEGsetting.jpg
Which as you can see is a two pass render, VBR max of 8Mbps.

Rendering to .AVI in Vegas is done with the following preset;
http://img.photobucket.com/albums/v355/Cally460/VegasAVIsetting.jpg

Rendering to MPEG in DVDA is with the following;
http://img.photobucket.com/albums/v355/Cally460/DVDAdiscsetting.jpg
What is not specified is if this 8Mbps is a constant or variable bit rate.

Looking at the vision, on the same screen and same player is the areas where motion artifact is noted on the Vegas render is not apparent on the DVDA render.

No filtration, such as sharpening, NEAT or correction was carried out on the project. It was 'untouched by human hand'.


Brian; Screenshots are above as requested.
Yes. I just took the .avi and placed it directly into DVDA and went from there.

Don; I have noted a two pass VBR render gives better viaual quality than a CBR. Also I have noted that Vegas will do a 9.8Mbps render, the max readable with most players is 8.0Mbps, and so I have stuck with that as a maximum in VBR and CBR renders.

Seth; Ive never encountered a video recompression issue when placing an MPEG2 file into DVDA. I also never knew the two programs used the same MPEG engine. How can you tell in DVDA what 'fine tuning' its doing?

Ben

Looking at the vision, on the same screen and same player is the areas where motion artifact is noted on the Vegas render is not apparent on the DVDA render.

The Bitrate is higher in the DVDA version. I am certain that would make a difference.

This is a question for the tech heads in the community.

I did an experiment last night. I rendered a project to a .avi file in Vegas 9, then placed that .avi file in DVDA 5 to burn to disc.

The project was set at 8mbps, in DVDA's properties.

Once the disc was burnt, I went back to the original timeline and used Vegas 9 to render it to MPEG2, again at 8mbps. Then used DVDA to burn that MPEG file to disc.

Checking both discs on a 42 inch plasma, the vision quality of the first disc was visibly superior to that of the second.

Any ideas why?

I was always under the impression that the Vegas MPEG engine was superior to that of the one in DVD Architect. Having seen the difference with identical vision on the same screen, I am swayed to change my workflow and let DVDA do the MPEG conversion.

Thoughts?
Arguments for and against?

Ben

Ben:

There is not enough information provided about your source material and Vegas settings to provide you any reccomendations.

There is no doubt that Vegas Professional will provide better or at least equal results to DVDA since they have the same basic rendering engine but the Vegas Pro version has far more tuning available.

...Ive never encountered a video recompression issue when placing an MPEG2 file into DVDA. I also never knew the two programs used the same MPEG engine. How can you tell in DVDA what 'fine tuning' its doing?

No recompression = good!

DVDA really doesn't tell you anything about the tuning it's doing, except the alerts it displays when you finally build the DVD.

Regarding the settings you've posted - what's the total length of all video segments you're putting on a DVD?

Regarding the Vegas MPEG2 render settings -
* Did you start with "Mainconcept MPEG2 | DVD Architect PAL Widescreen Video Stream" template, (which you then modified)?
* Your minimum and average bitrates are quite high. Usually, for content under an hour, I'd set 4Mb - min, 6Mb - avg, 8Mb - max. But it isn't clear to me that lower min and avg bitrates would make more bits available for bursting to cover short scenes that need it. Maybe so.

What kind of content are you working with - is it high-motion subjects (sports)? Static subjects? Shot handheld? Shot on a tripod? Well lit? Underlit? Is there video noise from camera gain?

In your first post, you referred to "vision quality" differences. Can you be any more specific about what you're seeing? Common playback errors include pixellation aka. macroblocking, banding, frame drops aka. freezes, mosquito noise (noise outside an object), video noise (in darker areas).

Eugene, I too love TMPGEnc . My question is which is best to burn afterwards : ISO file or DVD folder. What is the difference. Thanks, Bill

Eugene, I too love TMPGEnc . My question is which is best to burn afterwards : ISO file or DVD folder. What is the difference. Thanks, Bill
An ISO file is a virtual image of your DVD, while the DVD Folder is a folder with the Audio_TS and Video_TS folders in it. Either way should work fine for later burning.

Myself, I go with the ISO file, since it is just one file to keep track of instead of a directory, and it is prepared by my authoring program. I tend to trust having the program that authored the DVD decide on the exact layout of the DVD versus having my burning program create its own DVD layout just from the DVD folder of files.

Also, with the right software, you can mount an ISO file as a virtual DVD on your computer.

Ben:

There is not enough information provided about your source material and Vegas settings to provide you any reccomendations.
.


Hi Jeff,
Check the screenshots listed above. If the links dont work, PM me and I can send them to you.
Ben

Answering your questions, Seth;

Seth; Regarding the settings you've posted - what's the total length of all video segments you're putting on a DVD?
Ben: Most are sport and general shot for the local News. Average length is 25 minutes vision shot per game. Non TV gigs are 60 mins per DVD max.

Seth; Regarding the Vegas MPEG2 render settings -
* Did you start with "Mainconcept MPEG2 | DVD Architect PAL Widescreen Video Stream" template, (which you then modified)?
Ben: No. MPEG2, with "include video stream' box ticked.
When I first started using DVDA, I was told to render vision separate from audio, but I always stuffed it up, getting one without the other. I abstracted that an MPEG2 file should carry both and went from there. The only change to the recipe Ive been using (as in the screenshot) was to go from an 8Mbps CBR to the current two pass 8,7 and 6Mbps VBR.
But now you mention it, I will have a look see at the template you mentioned.

Seth; What kind of content are you working with - is it high-motion subjects (sports)? Static subjects? Shot handheld? Shot on a tripod? Well lit? Underlit? Is there video noise from camera gain?
Ben; High motion sports shot on a tripod, well lit. Others that have improved include news vision of burning houses, lit with available light from the fire trucks, no gain, and on a tripod. If I do switch gain on +18db with DPR (dual Pixel readout) on, the noise increase is marginal. If its at 36db or 42db, then the gain noise is like a snowstorm.

Seth; In your first post, you referred to "vision quality" differences. Can you be any more specific about what you're seeing? Common playback errors include pixellation aka. macroblocking, banding, frame drops aka. freezes, mosquito noise (noise outside an object), video noise (in darker areas).
Ben; The biggest change I have noted is the ring around objects is less apparent in the AVI to DVDA render, as is the blocking around text on titles supered over moving vision. Mosquito noise is also lower.

Ben

I abstracted that an MPEG2 file should carry both (audio and video) and went from there.
This may account for the quality differences - i've not closely examined this, but, presumably your 8Mbps peak VBR rate includes video and audio in Vegas, but is only limiting video in DVDA.

This may be important, because when you do an MPEG2 render with audio in Vegas, it includes PCM audio, which is uncompressed wav at 16/48, umm, what is that bitrate? Around 1.2Mbps if I remember. If true, that means you're comparing apx. 6.8Mbps Vegas MPEG2 to 8Mbps DVDA MPEG2. Which would ordinarily be extremely hard to differentiate, but, you're doing high-motion sports content, which is going to stress any encode.

Most people say, right or wrong, AC3 audio at, umm, 228Kbps (default, stereo) is good enough, and gives more disk space than PCM for video when rendering 1hr.+ programs.

Which is why you'll see continual reference to rendering video and audio separately - it's the only way to get AC3 aka. Dolby Digital out of Vegas. There are scripts that can run both renders automatically...

The biggest (vision quality) change (Vegas vs. DVDA renders) I have noted is the ring around objects is less apparent in the AVI to DVDA render, as is the blocking around text on titles supered over moving vision. Mosquito noise is also lower.

The "ring around objects". If this is white on dark objects, and black around light objects, you could be seeing the results of the camera detail level reflected as compression errors. I don't shoot with much detail, or at least haven't since the time I was working in TV news and sports (!) where high detail settings are routinely used. In those ancient tymes, detail was adjusted under the side plate of the camera, by an engineer with a tweaker, so us cam ops didn't have much choice about it.

Of course high detail, (and I mean a high level setting of a cam's detail processing, not small things the camera sees in the scene), is characteristic of much of TV news/sports. It does good things like outlining a golf ball or a hockey puck, players seem more distinct against backgrounds.

All this is a long way of saying that since my work started in video compression, I've not had to deal with much in the way of cam detail. Mostly a function of the markets I work in, where everybody wants their footage to resemble film, not tv.

But I'd certainly try a camera detail level setting of 50 or 60% of what you have now, and see if that too affects the look of your dvds.