mpeg-2 encoding problem

[Rick Mac]

What render template settings are you using?
Do you project settings match those of your source footage?

[Greg Mcdonald]

I’m using Vegas Movie Studio Platinum 7.0 which is what I thought originally was the problem – that with the cheaper software, they put a cheaper encoder in. But that’s not the case as both Platinum 7.0 and Vegas 7.0 use the exact same MainConcept mpeg-2 encoder. Just to make sure, I downloaded the free trial version of Vegas 7.0 and I got the exact same results with it as I did in Platinum 7.0.

So to get to your question, in both Platinum 7.0 and Vegas 7.0 I used the “DVD NTSC” template. I also tried in both programs the “DVD Architect NTSC video stream” template, then pulled the file into Architect and output a DVD. I used the default settings for the templates and in all the output files, the contrast increased by about 15%.

And yes, the project settings matched the render settings as the test source file/project was DV NTSC 29.97, 720×480, 4×3.

Here’s a 2 second clip from the .mpg output file from Vegas 7.0 where no adjustments were made to contrast or anything else:

https://icorptv.com/assets/multimedia/Vegas7-NoAdjustment.mpg

Here’s a 2 second clip from the .mpg output file from Vegas 7.0 where I reduced the contrast by 15%:

https://icorptv.com/assets/multimedia/Vegas7-Minus15PercentContrast.mpg

Here’s a frame grab from the source .mov file (the source file is uncompressed and it’s 632MB which is obviously why I didn’t upload it. But I assure you this frame grab is an accurate representation in terms of contrast of the source .mov file):

https://icorptv.com/assets/images/Source-Frame.jpg

Open the files with Windows Media Player or Real or any other program you use and you can see the mpeg file where I reduced the contrast by 15% looks like the original, while the mpeg with no adjustments – just imported into Vegas and output – shows the contrast increased pretty radically.

Any idea what it could be?

[Greg Mcdonald]

One thing I should add that I take for granted that maybe someone else won’t – I’m opening the .mpg output file in Windows Media Player, Real, PowerDVD, and Nero Showtime and comparing that to how the source .mov file looks when played back in Quicktime – so the files are compared within the same RGB set up.

I’m not comparing the output file to the timeline as it’s played back in Vegas – it looks fine in Vegas which has a different RGB set up.

[Edward Troxel]

The BIG question: How do both of your versions look on a TV? When one look right after burning a DVD?

Edward Troxel

[Douglas Spotted eagle]

Do you understand gamma? WMP has a different gamma.
Additionally, you can burn a DVD that has a 0-255 range or a DVD that has a 16-235 range. What you are describing is the difference between these ranges.
There is nothing wrong with the MainConcept encoder in Vegas, but rather what/how you’re feeding it or with what application you’re viewing the media.

[Greg Mcdonald]

jeditdv – yes as I mentioned I watched the DVD’s on TV sets and the increased contrast is present. I tried 2 different TV’s using 3 different DVD players connected to those TV’s and the increase in contrast is there. I can say with absolute certainty that the increase in contrast is hard coded in the file and is not the result of any one particular player.

DSE – I have a basic understanding of gamma and if WMP has it’s own gamma setting that can change the look of a DVD it’s playing then forget about WMP. I just listed WMP as one player I tested among 4 players on the computer and 2 TV sets with 3 DVD players.

I understand that you can change the setup to 0-255 or to 16-235 and that typically still photo files like .bmp and .jpg files have the full range of 0-255 while video is 16-235.

In all my testing I didn’t change change the RGB or apply any changes whatsoever to the video track. I simply imported a file, like a .mov file (but as I mentioned I also captured DV footage into Vegas and got the same result). After importing the file and putting it in the timeline, I then rendered to mpeg-2 for DVD. Really, that’s all I did. I DID NOT apply any plug-ins or filters to the video track, such as the “Levels” or “Color Corrector” plug-in where you can apply “Computer RGB to Studio RGB” or vice versa.

So I guess my question is, if you DO NOT change the RGB set-up or apply any plug-ins or filters to the video track and Vegas by default, because of it’s native settings, increases the contrast in a mpeg-2 encode, what plug-in do I need to apply in order to get a mpeg-2 file to come out properly and look like the original?

But even that is just another work around to solve the problem. I’ve already discovered that I can get the mpeg-2 output to look like the original by applying the “Brightness and Contrast” plug-in and reducing the contrast on the video track by 15%. So it’s the same difference if the answer is to apply the “Brightness and Contrast” or to apply some other plug-in.

I’m used to using Final Cut Pro and Premiere only occasionally, but with both those programs, you can import a file or capture footage, then export to DVD and the DVD looks exactly like the original source – you don’t have to change the RGB setup or apply any filters to adjust for what the encoder does.

In fact Premiere also uses the MainConcept mpeg-2 encoder (at least 2.0 or earlier does, CS3 might not) and in Premiere it does not increase the contrast in the output files.

Did anyone open the output files I posted (they’re small, 1MB or less) in the program of your choice (other than Vegas that is) and compare them to the frame grab from the source .mov file?

If you do that you can see what I’m talking about.

[Greg Mcdonald]

jeditdv – in my post I only mentioned how the DVD output from Vegas looked on a TV where I made no adjustments to it.

The DVD output from Vegas where I reduced the contrast by 15% looked great on both TV sets – like the original as it should. As well as the DVD output from Premiere and Final Cut Pro looked like the original and didn’t have any increase of contrast.

[Rick Mac]

Lunchmoney,

Forgive me, a few more questions.

1) Your DV playback while editing is good? Contrast looks correct While editing?

2) You then render from the timeline your mpeg file. Does the contrast still look correct pulling the rendered mpeg file back onto Vegas timeline?

3) When you pull the mpeg file into DVD Architect, does DVD Architect tell you that it needs to recompress the mpeg file?

Don’t give up, we will do our best to find the answer.
Just to let you know, I edit DV and I use DVD Architect to author DVD’s and my video looks the same throughout the entire process. So this is not normal for Vegas.
What we may do is have you post a few seconds of your DVD Architect project and try a burn from here. I will try to get a hold of Edward (JetitDV)for what we would need you to post.
Edward, If you see this perhaps you could tell him what to post so we can experiment with it.

Regards, Rick.

[Edward Troxel]

I know I’m definitely not seeing that here. I capture DV, render to MPEG2, output to DVD, and it looks the same as the original DV (comparing both on the same TV).

[Greg Mcdonald]

Hey Rick & Edward,

Thanks for your help with this. Yea, I’m sure it’s something I’m doing or not doing right. I’m glad to hear you don’t have this problem. I couldn’t imagine Sony would put out a program that does this.

So, to try to narrow it down to what may be happening;

1. Yes, the DV playback while editing is good. Contrast is correct while editing and playing back in Vegas. In fact, when I reduce the contrast by 15% in Vegas, the video looks like the contrast has been reduced.

2. When I pull the rendered mpeg files back into a Vegas timeline, the file where I reduced the contrast by 15% looks like the contrast has been reduced by 15% and the file where I didn’t adjust the contrast, looks correct and normal.

3. When I pull the mpeg-2 file into Architect, it does not recompress the file.

Since the .mpg files I posted aren’t recompressed in Architect, Architect just converts them to standard .ifo and .vob DVD files, wouldn’t they suffice for a test?

If not, I’ll post up a .vob file output from Architect.

Thanks again for your help with this.

Greg