Multiple pass rendering - how?

[ruggy1]

I may sound crazy, but how is it that movie studios can produce DVDs with almost 4 hours of top quality video on a single (probably dual layer) DVD when the best I can get with MF is one hour of mediocre video on a single layer DVD. I have tried the 2-pass option when rendering my high quality .AVI file but it does not seem to improve the quality. I have heard they use up to 20 passes to get the quality we see on DVDs. How can I do this with Ulead, or with any other product? Thanks

[Ron P.]

Their equipment to do multi-pass encoding costs just a little bit more then $100, probably somewhere around $50,000 upwards. So you too can do 20 pass encoding, just go buy one of them...

[heinz-oz]

I'm afraid you can't. Commercial DVD's are not burned, they are pressed, different technology alltogether. Having said that, if you only get 1 hour of mediocre quality onto your single layer DVD, from a good quality source DV AVI, there is something wrong with your process.

We need a lot more detail on source format, project settings etc. to judge what it is you do wrong.

Dual pass encoding only reduces file size while retaining good quality if there is fast moving action in your video mixed with slow moving or stationary scenes. The idea of dual pass is that the system will inspect your footage and decide where a higher bit rate is neede (because of movement) and where a lower bit rate will suffice. If you only have one hour of footage, you can go for a constant high bitrate, single pass, CBR, to get good quality DVD.

[ruggy1]

Thanks guys. My rendered .AVI file is as good as it could be using Ulead - VBR, 9000Kbps etc. But I am still curious as to how they do a 20 pass DVD. They may do it quicker on a $50,000 machine but it must still be a linear process that should be reproducable on a high powered PC, no matter how long it takes. I know they sometimes use banks of Itanium blades, each chewing on a bit of video at a time, but it still must be possible to do the same thing on a PC, even though it may take a day or two?

[maddrummer3301]

.

[ruggy1]

Thanks again MD. I assume as no-one has been able to answer my multi-pass question, then it must be a trade secret! So moving on to MD's suggestions.

I am between customers and have experimented. I used 2 different inputs, firstly Video8, captured on the fly, and then a good .AVI file from a DV camera.

Video-8 on the Fly results with MF5+:
1) Preference setting 'good', single-pass VBR at 7000, 88% quality setting
2) Preference setting 'best', 2-pass VBR at 7000, 88% quality setting
3) Preference setting 'best', 2-pass VBR at 9000, 100% quality setting
Burnt them as 3 chapters on a DVD - absolutely no difference in quality when played back on a good quality 100MHz TV. Cannot tell them apart.

DV .AVI captured by Firewire on disk then rendered with MF5+:
1) Preference setting 'good', single-pass VBR at 7000, 88% quality setting
2) Preference setting 'best', 2-pass VBR at 7000, 88% quality setting
3) Preference setting 'best', 2-pass VBR at 9000, 100% quality setting
Burnt them as 3 separate .iso images - absolutely no difference in quality when played back on a good quality 100MHz TV. Cannot tell them apart, except that the 3) option produced a file that was 25% bigger

So maybe MD is right - is does not matter what you set the properties to, MF5 just goes right ahead and does its own thing anyway!

I have P4, 3.4GHz, 1TB disk, 1GB RAM.

[maddrummer3301]

.

[DVDDoug]

They may do it quicker on a $50,000 machine but it must still be a linear process that should be reproducable on a high powered PC, no matter how long it takes.

I agree. But, I'll bet the MPEG encoder software is very expensive too. And, their software may allow them to intervene manually, telling the software what parts of a frame/scene are critical and need more data.

In fact, I doubt that speed is that important to the studios. They have several weeks between the film release and the DVD release, and it's not going to matter if the encoding takes 24 - 48 hours. Whoever is actually doing the encoding, probably does want to complete it quickly so that they can move on to their next project.

Also, most movies are still shot on film which has higher quality than DV. They may use something better than AVI/DV for the telecine (film-to-video) process. I'll bet they at least use uncompressed DV (60GB per hour). If you feed higher quality to your MPEG encoder, higher quality is going to come out.

And, you are going to get diminishing returns. A 20-pass 2500kbps MPEG is not going to be as good as a 2-pass 5000kbps MPEG.

[ruggy1]

Thanks guys, I will try CBR.