Rendering Bit Rates

cpalermo wrote on 11/22/2004, 10:59 AM
I currently use Vegas 4 but I've played around with 5.

With 5, it says it has "Improved MPEG-2 rendering, including two-pass encoding". Anyone know to what extent this is improved? And how does two-pass encoding help- does it just decrease the size of the mpg file?

Also, with both, I noticed that when you render to mpg2 and select the custom button- on the project tab you can have video rendering quality be either good or best (among others), with good defaulted. Is there really any noticeable difference between good and best?

Thanks...

Comments

FrigidNDEditing wrote on 11/22/2004, 11:51 AM
I didn't use V4 but I do know that the Good and Best have a negligable difference if any with video, where it makes a difference is with resizing. So, with videos that are using Picutres or with projects where you are doing post production zooming (interviews etc...) you would want to go with best. It will take longer however.

Sorry I couldn't help with the other stuff.
John_Cline wrote on 11/22/2004, 12:17 PM
Two pass encoding helps by more intelligently deciding how to allocate the available bitrate "bandwidth" when using variable bitrate encoding (VBR.) On the first pass, it analyzes the file looking for "difficult" places, like high motion or highly detailed images. On the second pass, it will allocate a higher bitrate to these areas and a lower bitrate for those other areas which would not benefit from the extra bitrate. Two pass encoding is particularly important when you need to use a relatively low average bitrate in order to get a long program to fit on the DVD.

It does not decrease the size of the file, that is strictly a function of the average bitrate setting.

John

johnmeyer wrote on 11/22/2004, 12:21 PM
1. The V5 MPEG-2 encoder uses later algorithms and provides better quality.

2. Two-pass encoding will improve quality when you encode at relatively low bitrates (6,000,000 avererage bitrate and below). It uses the first pass to find the sections of video tha contain high motion and then encodes those at high bitrates. It then encodes low-motion video with lower bitrates. It has to perform the first pass because otherwise it cannot ensure that the average bitrate will exactly equal what you specified. Two-pass will give you exactly the same file size as one pass.

3. The Good and Best options on the first tab (the Project tab) of the Custom button have always been there and do not provide better quality for video. In some cases, best can provide better quality when your project contains still images.

You will get a HUGE difference in quality from the Video Quality slider found on the Video tab of the Custom button. Sony, for some totally irrational reason, chose to set this at 15 for the Default setting for MPEG-2. Since Vegas 5 does NOT remember the last template you used, you are always forced back to the default setting, and if you forget to change this setting, your MPEG-2 render will not look good. The differences are not subtle. It will render fast, but the quality will be poor.

All the DVD Architect templates set this slider to the maximum value of 31, and that is what I recommend you do for all renders intended for DVD.
cpalermo wrote on 11/22/2004, 1:02 PM
Very helpful, thanks... So with that video quality slider, does that affect the size of the mpg file once it's done, or does it simply increase the rendering time? Also, how can the video quality slider act independently from the average variable bit rate?

For example, what is the difference in a rendered video that has high on the video quality slider but low on the average variable bit rate, as opposed to one with low on the video quality slider but high on the average variable bit rate? If I'm making DVD's, you said you'd recommend 31 on the video quality slider, any recommendations on average variable bit rate?

And I guess I should always use two-pass, right? (Assuming time is not an issue)
Chienworks wrote on 11/22/2004, 1:30 PM
The Quality slider has nothing to do with bitrate. What this slider does is specify how "careful", so to speak, the encoder is about doing a good job. At a low setting, the encoder rushes through the job as fast as possible and uses "sloppier" algorithms. At a high setting more precise algorithms are used and it takes longer. In both cases the bitrate and file size will be essentially identical. The longer render with a higher quality setting should look noticeably better because the bits are put to better use more accurately.

Two-pass encoding allows the encoder to decide in advance which sections need more information to be rendered more accurately. It can then use fewer bits for more static or lower detail sections and save these bits for where they are needed. Once again, the final file size should be just about the same since this is deteremined only by the bitrate specified and the length of the video. The two-pass version should look better in high action and high detail areas than a single pass, at the cost of taking twice as long to render.
johnmeyer wrote on 11/22/2004, 3:58 PM
Two pass lets the encoder use a wider range of bitrates. When a single pass is used, the decision on when to use lower bitrates is done on a fairly local basis. For instance, to take an extreme example, suppose the last ten minutes of a thirty minute video was a static shot of a still room, taken on a tripod. The bitrate for this ten minutes could be almost zero and it would still look good, However, the single pass algorithm would be optimizing the bitrate just a few seconds at a time, increasing and decreasing as the motion changed every few seconds. It would have no idea about this big chunk of zero motion footage coming up later in the video. As a result, it would encode the faster, earlier action at far too high a data rate.

As to the video quality slider, my understanding is that this affects how many frames the encoding looks at in order to determine how to encode the frame differences. This is really the primary difference between real-time MPEG encoding and the kind of quality encoding that can only be done non-real time. The more frames the algorithm looks at, the more information it can use to determine how to track the motion of pixels from one frame to the next.

If you use the standalone version of the Mainconcept encoder, you can control dozens of other parameters that determine not only how many frames are part of this search, but also how many adjacent pixels, both in the horizontal and vertical direction, are used. If you are willing to spend the time to learn how to use these controls, and if you don't mind some extremely long encode times, you can get remarkable quality, even at lower bitrates, using this encoder. The other standalone encoders (Canopus, CCE, etc.) have similar settings, I am sure.