I did extensive tests on this about eighteen months ago, but got so confused, I finally gave up. The test involved setting the event properties to upper, lower, and progressive, and then setting the project properties to upper, lower, and progressive, and then putting the resulting VEG into another project and then -- you guessed it -- setting those project properties to upper, lower, and progressive. I then rendered the result to MPEG-2 using, well you know. This resulted in a LOT of permutations.

The reason for doing all this is that was getting some very weird judder effects. In addition, the source material was HDV and I was going to SD MPEG-2.

I finally got something that looked OK, but I never was able to get something that looked as good as rendering directly from a project without using nesting. If I can find my original test results (I sent them to Sony), I'll post them here.

OK, this is a test someone else did:

Text flicker nested

I traced through his steps and have found that if you set the original text event's field order to match the project (lower field first for standard NTSC DV, rather than the default Progressive), and then drop the resulting VEG into another project, there is no problem.

I also seem to remember that the problem went away if you set the preview resolution to best, so it was not clear that the problem he was tracking down actually made its way to the final rendered result. In fact, I just found a response from Sony and they said that this particular problem was the result of how they downsize the nested VEG for displaying at "preview" resolution.Now, here is the last of a very long series of emails with Sony. Sorry this is long (even by my bloated standards), but as I say in the email, perhaps it will help someone.I just spent over an hour doing various tests. I started by choosing one of the original VOB files (from of the high school's game DVDs). I then tested that file to confirm that it is really upper field first, and not just being mis-reported. There are several ways to do this, but they all involve separating each frame into fields, thus doubling the number of frames, and then walking through one frame at a time and looking for backwards motion on alternating frames. I confirmed without a doubt that the VOB file definitely is upper field first.

I then put that VOB on a fresh Vegas timeline and saved various VEG files. For each file, I created permutations by changing the field order in both the media properties and project properties. This gave me four VEG files. I then imported these VEG files into Vegas and created permutations by rending to MPEG-2 files using the DVD Architect MPEG file template. For these permutations, I changed the project properties, and also the field order in the render template.

This gave me 16 MPEG-2 files. I put these into DVD Architect and created a test DVD that I watched on a standard NTSC monitor.

All eight of the files where I had changed the field order of the source VOB created video that juddered. Thus, "lying" about the true field order by changing the media properties of the original VOB screws everything up. Not a big surprise, but I thought perhaps this might be "fixed" later in the process.

Next, any attempt to render the final result using upper field first in the render template (which is what I think you were recommending below). resulted in field-reversal jitter. Not softness, but jitter.

The two situations that resulted in no jitter, but soft results (indicating some sort of operation that halved the resolution by using the same scan line twice, or something similar) were:

 

|         VEG                  | project that uses the VEG

          |                                 |

          v                                 v

source media bff; project tff;   proj tff, render tff

source media bff; project tff;   proj bff, render tff

Thus, if I "lied" about the proper order by setting the media properties for the original VOB to lower field (bff), but changed the project properties to top field, when I put the resulting VEG into a nested project and then rendered to MPEG-2 using upper field first (tff), I didn't get jitter, but the result was soft.

My conclusions are that the project settings in the original project from which the VEG gets created don't seem to matter. The media gets passed through based on the media properties. If you change these media properties, you can create bad things, just as you would if you did so without nesting. I guess this was the first thing that was interesting (the fact that project properties in the source project doesn't matter).

Second, I was unable to find any combination where rendering to the final MPEG-2 using upper field first (tff) produces an acceptable result. This worries me, because you know FAR more than me and this is what you recommended. However, out of the eight tff renders, two were very soft, and the other six juddered.

If I had more time, I would describe what happens when you try to view these files on the Vegas timeline though an external monitor. In the end, this is what was actually confusing me because what you see when played from the timeline is not at all what you see when you render, burn a DVD, and then view the result. If there is any issue with Vegas, it is probably here.

For instance, in one of the earliest tests, which I am re-doing now as I type this, just to make sure, I stacked all 16 MPEG-2 files back on a new Vegas timeline. The Project properties were set to the default NTSC DV 29.97 template. 14 of the 16 MPEG-2 files looked "soft" when viewed on an external monitor, with the Preview quality set to Best-Full. There were three different degrees of softness (which I could describe, but this is too long already). The two renders that produced sharp results back on the timeline were both when the source VOB and the source project properties were both set to lower field first and then rendered to bottom field first (the project properties of the nested project didn't matter). I could not tell, however, if they were playing with or without judder, because the playback was only about 20 fps. These two files looked terrible when actually prepared to a DVD and played on an NTSC monitor.

In thinking about this, the "sharpness" and "softness" of the results viewed on an external monitor from the Vegas timeline may simply be an artifact of the field reversal. I looked really hard to see if this was the case, and I think it is. I have attached two timeline snaps. You'll see immediately how I was fooled. Look at the horizontal lines in the empty portions of the grandstands in the background. On the sharp but juddery version, the stands appear much more clearly. I guess if I were ever to try to understand this subject well enough to teach it, this would be a very interesting lesson in apparent sharpness vs. real sharpness.

So, I think my "problem" is solved. Now that I "understand" it all, most of it was caused by seeing something on the external monitor from the Vegas preview line that doesn't actually reflect what happens when the file is rendered.

So, the only real feedback is that upper field first rendering does not create a useful DVD, no matter what combination or permutation of setting is used.

John