I've never had to re-encode my MPGs to use them with DVDA. I encode them from DV AVI into MPEG2 using the Vegas-supplied template, and DVDA uses them as is.

I think one thing that you've missed, and relates to the question about why you have to re-encode even if you only snip out a little bit, is that MPEG does NOT encode every frame in isolation. In otherwords, most frames are encoded as the difference between that frame and either the preceeding one or the following one. What this means is that if you remove a frame from the middle of the MPEG, there will be no way to recover the following frame since it most likely requires the missing frame inorder to be decoded (unless the removed frame is the last frame of a GOP). In other words, each successive frame in a GOP builds on data from the previous frame. Only the "I" frames stand alone. This is one of the reasons that chapter markers need to be on an I frame, because that's the only place you can start a decode.

As for hardware encoders - I'm sure there are some, but they are expensive, and you are locked into a specific encoding format/technique. With software encoders, as new techniques allow the data to be more compressed without noticable quality loss, we can take advantage of them.

Of course, that's not to say that I ENJOY the long rendering times. Like you, I wish it were faster. Just hopefully now I've shed a little light for you on why it works the way it does.

bcbarnes. Oh yeah, I see that now. In one para you've taught me a lot. Now I remember why fractals were so different...

So, what do you do is just use Vegas to "render" to MPEG2. How long does it take on your box, roughly?

Maybe it's because I have 20 large AVI files I need to work on. I wish there were a BATCH way to convert them all to MPEG2, and leave it running perhaps whilst I'm on vacation :)

As for my replayTV MPEGS, I have no idea why it wants to re-encode each one again for 5+ hours for 30 mins :(

Glad I could be of some help.

I'm currently running an AMD Athlon 1800+ XP system w/1Gb memory, and renders generally run about 2:1 (i.e. 2 minutes to render each minute of video). This varies somewhat, but generally is true.

I tend to render the audio separately, using the AC-3 render settings, and the render time for that is negligible - i.e. a couple of minutes to render out a 30 minute timeline.

Do you have the specs for your replayTV mpeg files? i.e. resolution, CBR or VBR, bit rates, audio settings, etc.. If you post them, someone here might be able to pick out why DVDA doesn't think they are DVD compatible and re-encodes them.

One more thought (or maybe two):

Canopus has just released their new MPEGPro hardware, real-time MPEG-2 capture devices (http://www.canopus.com/US/products/MPEGPRO_EMR/pm_MPEGPRO_EMR.asp) based on the MVR-1000 (hands down the BEST hardware encoder for less than $1,000). These new products (the PCI card version is not out yet) will retail for $499. The one that's released is the external version that offers composite and S-video inputs and a USB interface. If you're running USB 1.1 you're limited to 6 Mbps capture rate. USB 2.0 is capped at 15 Mbps which is the limit of the chipset.

I've used the MVR-1000 since the 3rd quarter of last year and it is terrific. It offers a wealth of built-in processing features and is a breeze to use. Often, I use the print-to-tape feature of Vegas to a JVC SR-VS30U MiniDV and then out to the encoder. It works a treat. However, ProCoder (also from Canopus) is the best and fastest software encoder of just about every technology that I've yet seen. The Mastering Quality setting is as near to a hardware encoder as I've found. Second choice is the much more rare Vitec DCM 3. This card offers video blanking during capture (terrific for tapes with streaks at the bottom of the sceen due to head placement differences) and real-time AC-3 192Kbps 2.0 audio capture. You don't get to fool with the AC-3 settings but for the cost of a simple AC-3 encoder you get a real-time MPEG-2 capture card as well.

There a lot of available solutions, but since I use one or all of them every day, I spent about 2 months testing various solutions including Cinema Craft SP, but found that the products from Canopus are the best price performers around. Of course there's the SD-2000 from Sonic Solutions for $25K but I'll wat till my feature is done for that one.

>>I wish there were a BATCH way to convert them all to MPEG2, and leave it running perhaps whilst I'm on vacation

TMPGEnc does have batch encoding capabilities. (v2.5)

It's pretty slick.

While it may not be a nice as Canopus' batch encoder, it is a lot more affordable.

There's a Script for Vegas which will allow you to Batch encode - even lets you output to more than one format, e.g. MPEG and WMV, without you being there ...

I guess it all depends on where the data is coming from. I have (in essence) a realtime MPG encoder. It is the Dazzle Digital Video Creator. But, it works with my external camcorder (or other RCA plugable device) and when it is recording the data from the tape - it save to MPEG1 in realtime.

So, sure - it's possible.

But when you are in an editing application - you are not editing in real time or rendering in real time (so that blows that). I would have to print to tape from Vegas, to turn around and play back in real time to make an MPEG....which, depending on the length of the video would take just as long to render directly to MPEG.

Of course, I guess, if you really really really wanted to (and had enough RAM power) you could run your video from your timeline and do a screen capture of the preview window to some screen capture application...but then you would run the risk of the video not playing smoothly and such.

Even the big pro's using computer generated graphics for movies and such know that rendering takes time - even if you have a super computer. But the faster your processor - the faster the rendering.

My $0.02

(Of course, I say all this after testing a rerender of an video....2 hour vid I edited....took about 27 hours to render. But it looked great when I was done.)

Q: How come some software re-encodes for hours JUST if you snip an itty bit from it?
A: This is one of the biggest sins commited by SOFO and other companies. It is pure laziness on the part of the software engineers. Cutting MPEG should only require re-rendering a few frames on either side of the cut (unlike DV, where every frame is compressed individually, MPEG frames depend on their predecessors). SFO has the technilogical expertise to do this, and IMHO they SHOULD do this, both in Vegas and in DVD-A.

Q: How come Vegas and many other software products re-encode an MPEG when it's already an MPEG?
A: This sometimes is required to make the video fit in the 4.7/4.37 GByte space of your DVD. Again, Vegas currently lacks a tool to correctly set the bitrate to make the MPEG-2 stream fit a given storage size, so if you encode in Vegas, and later try to make it fit a given space in DVD-A, DVD-A re-encodes. I assume this will be fixed in a later release.

Q: How come some MPEGS in some products require no re-encoding and burn straight onto DVD whereas some products, Vegas included requires a re-encode.
A: If the video has been encoded into MPEG-2 that is of the correct format to put on DVD, no re-encoding is required. However, as pointed out above, Vegas is currently not smart enough to only render the few frames around a cut point if the original video was MPEG.

Q: Since re-encoding is a huge process are there any hardware products that speed up encoding x format (MPEG or otherwise) to MPEG?
A: Not that work with Vegas.

Q: Is the industry working to solve this problem (assuming it is a problem and judging by the questions it is) and is it solveable (standards/hardware/specials etc.,)?
A: The industry is a long way down the learning curve, because MPEG has been around for quite some time and is used in satellite, DVD, and some cameras. Therefore, don't expect major changes, other than what faster computers will bring. The biggest change -- and something within Sonic Foundry's grasp -- is to make Vegas and DVD-A MUCH smarter in the way they handle MPEG to avoid re-encoding. In other words, the encoding is always going to take a lot of time, but unless the video frame is changed, that video should never have to be re-encoded again. If SOFO makes that a priority, then we will see substantial improvement in a relatively short amount of time.

Q: A cheap $100 capture product can create MPEG real time, is it any good and if so, why isn't there a card that can "CAPTURE" MPEG/MOV files from a PC and create an MPEG - real time...
A: The MPEG real-time capture is actually not too bad. I have it in my ATI 8500DV All-in-Wonder capture card. For TV shows captured off satellite, the result, when played back, looks FAR better than VHS, and much better than S-VHS. However, if you start with really good video, like well-lit DV shot with a 3-chip camcorder, you will appreciate the better quality provided by the SOFO/Mainconcept non-real time MPEG-2 encoder.

Q: If money was no object, can a PC ever encode QUICKLY with a special hardware upgrade/card/chip....
A: Satellite uplink facilities use such hardware, but it costs many hundreds of thousands of dollars. You did say money was not object ...

Q: How come Vegas and many other software products re-encode an MPEG when it's already an MPEG?
A: Actually there is a way in Vegas. Use the custom templates to create differing bitrates of encoding. I've been successful in changing the bitrates at which my video encodes into SVCD. This now enables my 40 - 42 minutes of video to fit on a standard CD-ROM.

I have to question the assertion that simply removing a few frames from a GOP shouldn't require a complete re encode. I suspect there could be quite a few issues jump up doing this. Certainly just building a few GOPs around the cut would produce a valid MPEG stream but would it still be valid for playing on a set top DVD player?

And whats going to happen to the audio thats possibly muxed in with the video?

Certanly there are mpeg2 editing systems that don't re-encode every time you do a cut but as I understand it they use only all I frames which makes it much easier to handle. Also there is a big difference is there not between editing a straight mpeg video stream and one that already muxed for DVD, its MAY (one day in DVDA) have multiply audio streams, camera angles etc also muxed into it.


On another point though, cheap, good hardware encoders should be coming, the chip used in the GR-HD1 camera is also going to appear in standalone encoders (if it hasn't already).

Anyone have a good source of info on the audio linkage to the video? The video is quite straight forward. You have a complete JPEG-like image (I frame) at the beginning of every GOP so when splicing the editor should not have to look at more than 15 frames then create new P and B frames. However with variable bit rates on the video and audio, how is the audio segmented in AC3 and in PCM. It either must exhibit the same variable rate or else it must be keyed to the I frames of the video. This is one area I have never totally understood (the keyframe on audio and how frames are time tagged for variable bit rate). Is there a good free web reference or can someone explain how the two separate streams are synchronized? I've noticed splicing programs often slip audio at the splice when combining two mpeg files of variable bit rate.
thanks in advance

Maybe to answer that you should buy the 'Bible' from the DVD Forum or buy a new car instead.

ftp://ftp.avc-pioneer.com/Mtfuji5/Spec/
There are lots of specs out on the web for free, like the above mt fuji specs.
http://www.pioneer.co.jp/crdl/tech/dvd/5-3-e.html
The above is a chapter from the pioneer technical guide (with links to the rest of the guide) which was helpful on dvd audio

surely there are others that are free?

Everyone please welcome me to am first DVD burning session, thank you, you may all sit down now.

I have a newly edit piece of work in Vegas that I just rendered MPG2 (mainconcept)withou changing any of the specs. The file is a little over 1 gig. Is this going to work as is? (I want the answer to be yes, so I can take my new 1 gig MPG2 file, put it on a portable hard drive, drive over to my office where I have DVDA and a DVDR, pop it in and watch it burn at 4x. Thats it! done! or is there more waiting and headache?
I have been told that DVDA will want to separate the a/v tracks and re-render... help me please..

Re-rendering is the main reason I gave up on DVD-A. I capture through Canopus hardware and use a couple of low end authoring programs to build my products in minutes rather than hours.
peter