Vegas Video discussions from 2003

[Peter Jefferson]

5.1 mixing can also go beyond the standard keyframed panning within a stereo track...

vegas also allows for a myriad of bus tracks which can be assigned to seperate channels ;)

[Glen Elliott]

I recently finished a commercial for a local jeweler which needed to be transferred to BetaCam prior to airing on local cable. I finished the edit and output to DV AVI and for simplicity burned it to a CD. Beings it was only a 30 second spot it was only 200megs.
Anyway I get a message from the guy thats doing the transfer to Beta stating the AVI is not the right size. I checked and sure enough, when played in Windows Media player it only plays in half res.
Now the only place I see any sort of setting that seems remotely related to this is under the Vegas Capture options under the Preview tab. It has a check box entitled, "Base DV decoding resolution on preview window size" which is checked. Under it it has, "Allow full-screen file playback via Preview on Device or Record to Device" which isn't checked. They seem related to my problem but beings they are located under the "Capture Options" I'm a bit weary as to if they will help me.

All of your input will greatly be appreciated. Happy Holidays to all!

[Rob Lohman]

There is no problem. There is an annoyance though. The Microsoft
DV Codec is setup to default playback DV at half resolution.
Probably for speed reasons.

Althought you can change this I can only find how to do this with
the old mediaplayer and not the new one. However, this player
should still be installed on your system. Do this:

- Start -> Run -> mplayer2
- drop an DV AVI file on the player
- hit pause
- File -> Properties -> Advanced -> DV Video Decoder -> Properties
- Full (...), Save As Default, Apply, OK
- Close
- CLOSE PLAYER DIRECTLY

It might be that the dialog to save it as default will not close or
does some other weird stuff. Try closing it with the X in the right
upper corner etc. Play with it. You will need to close the player
and then open the file again to see if you did it. If succesful it
should remember your setting.

I know. Handy this hidden feature that you can't easily change.
Just the way the Microsoft world works...

[Glenn Chan]

Ok I just tested using RamDiskXP (uses RAM as a hard drive) versus an IBM deskstar 80GB 2MB buffer. The RAM drive should be a lot faster than the IBM (by a factor of 60 or more). My results show that there is a small difference in rendering speeds, but only on renders that would normally take a short time to render. Practically there isn't much reason to pay attention to drive speed unless:
-hard drive speed is a bottleneck (i.e. you can do a lot of streams of RT or want uncompressed) If your drive is really fragmented it could be a bottleneck. If it's a firewire/USB2 drive or if the drive is in PIO mode it could be a bottleneck.
-you spend a lot of money on very small performance increases.

Results:

Testing methodology:
I tried to keep as many factors the same as possible. I used the sample projects and rendered material out (file --> render as...). The render times are as reported by Vegas (presumably it's accurate).

Test 1: Color Correct
The project actually renders faster if you go file-->render as... instead of selectively prerendering. Anyways...

With the IBM drive: :24, :24, :24, :22, :24, :24
With the RAM disk: :21, :23, :21, :23, :21, :21
About a 9.2% difference.

Test 2: Backlit Shadow
I just rendered the loop from region marker 2 to marker 1.
IBM: :46, :46, :46
RAM :46, :46, :46
No difference.

Test 3: Intercutting film and video
I added the gaussian blur filter to all four clips in the project, default settings.
IBM: :51, :51, :50, :50
RAM: :50, :50, :49, :50
About a 1.5% difference

On the CPU-intensive renders (tests 2 and 3) there is practically no difference between the two drives. On test 1 there is a small difference, but probably something you wouldn't notice. Practically speaking, getting faster hard drives isn't worth it. Short renders will show a difference but they are fast renders anyways. For long renders, there won't be a difference. RAM disk isn't really a very practical storage solution so you will need to go with a drive array (RAID). I don't expect that a RAID will perform better than a RAM disk so performance will fall somewhere in between. That makes the performance difference even smaller. However, performance will take a hit if your storage is a bottleneck (see above for the situations where this happens).

[Glen Elliott]

Hmm, I'm wondering now if the guy who's doing the transfer opened it up in Windows Media Player and thought it was half res because of that. You can easily see that it IS the correct resolution by right clicking on the file and choosing properties- 720x480 is the listed resolution.

I guess I have to personally talk to the guy I guess to work this out. So...those options I mentioned have nothing to do with this?

[Edward Troxel]

Glen,

Find out WHAT he is using to play it. If it's not the program that prints it to tape, make sure he tries it there. Windows Media Player can definitely give some incorrect results.

[Edward Troxel]

Testing methodology: 1 hour program, about 10 minutes needs rendering.


PIII 750 MHz, 256 Meg RAM, 7200 RPM drive
PTT render only the changed sections: about 50 minutes
Full render to new AVI: about 1.5 hours


PIV 2.8GHz, 1 Gig RAM, 5400 RPM drive
PTT render only the changed sections: about 15 minutes
Full render to new AVI: about 1 hour

Connecting a 7200 RPM drive to via firewire to this computer also speeded up access tremendously.

My conclusion is tht the drive speed DOES make a difference BUT it depends on how may effects and CPU speed as well.

[Dave Largent]

How do I set up V4 so that captures go to
the file of my choosing. Poked around in
V4 capture and didn't see how to set this
up.

[Randy Stewart]

Dave,
Open Vidcap (click on the camera in the media bin), click on options at the top, then click on preferences at the bottom, click on the disk management tab, click on add folder, locate the folder on your hard drive and click okay, then check the box in front of that folder to make it active. Your captured vid's will go there. Don't forget to name the tape so all of your clips will begin with that tape name and a number in order of capture. Hope this helps.
Randy

[Dave Largent]

Thanks, Randy, that did the trick.

[Randy Stewart]

My pleasure. Merry Christmas!
Aloha,
Randy

[Peter Moore]

It's very simple. If the rendering of each frame takes longer than writing that frame to disk, higher disk speed is not going to affect your rendering times. The DMA controller is sent data and then takes over and writes the data independently of the CPU. If the DMA controller is ready for more data before the CPU is ready to give it more, it's not going to benefit from any HD speed increase.

So, yes rendering time is unlikely going to benefit directly from faster hard drives. Would I still encourage everyone to get the fastest disks they can get? Absolutely! All program loading times are quicker, all file read/writes are quicker, and most importantly, all virtual memory read/writes are faster. If you're doing DV, you're going to have huge files, and you're going to be needing to move them, process them, manipulate them etc. You're going to want the fastest drive you can get.

[Peter Jefferson]

speed is also determined by the amount of dives you have instaled and how your rendering ios configured...

I have a "raw" drive, its a WD 120gb with 8mb cache..
i then use the a drive of teh same model for my master renders...
The temp drive directory is a partition of my main sustem drive.
and the main application ois on the system drive

in effect i using 4 seperate drives (3 physical) for rendering.


this saves ALOT of time, as you have one drive controlling eveything, then you have the temp drive housing the prerendered frames, then you have a read only drive (raws) then the writing drive.

Using a P4 2.4 without HT, and 1gb Corsair DDR 3200 onan abit it7 max 2 series 2 mobo, i usually get my rendering flying at about 1/4 realtime (a little faster actually) if no effects are being used. basically 1 hour and a half takes abotu 15 minutes... transtiions add about 10 minutes...

When using Widescreen mask, colour correction, or basically any effect or transitions, it is usually 8 times slower than realtime.

Today i did a 1 hour render of a wedding using
Widescreen mask,
Colour correction,
titles
frame positionin of almost all clips using track motion
slow mo
bump map
gausian blur
b&w
blur
glow
convolution
and afew others i cant remember right now

this took about 8 hours and 16 minutes and left me with a 13gb file.. ALL frames were redrawn...

with a HT CPU, you can cut about 20% of the render time...

this is all this AVI output, obviously mpg2 would take longer (about 20 or so hours)

[Edward Troxel]

There are also a couple of other "file locations" you need to change. Under File - Properties and Options - Preferences in Vegas is a couple of other file locations which should be changed from the default.

I have a lot of the configuration options, including these, explained in my newsletter http://www.jetdv.com/tts

[Glenn Chan]

Your RAM has timings that can affect performance. RAM timings are most conveniently represented as a string of numbers. The best is 2-2-2-5. The worst is 3-4-4-8. These numbers represent the four most important timings. There are some others but they aren't too important.

Long story short, RAM timings don't seem to have any impact on rendering times. I tested 3-4-4-8 versus 2-3-2-5 and saw no difference. RAM timings make a 20% in winRAR compression and around a 2% difference in games, but for Vegas they don't seem to make any difference. It's possible RAM timings make a 2% difference but the performance increase would be very underwhelming. You probably can't feel any performance difference that's less than 10%.

Memory bandwidth

By changing the memory dividers on my computer, I can lower the frequency at which the RAM runs. Some computers run the RAM at a memory divider (i.e. the computer is runs 800mhz front side bus while the ram is DDR333). The effect of the lower RAM speed is less memory bandwidth. Instead of 1.6GB/s bandwidth I'm getting either 1.28GB/s or 1.06GB/s. All these speeds are absurdly fast. Running at a divider also kind of lowers the memory timings since the front side bus and the memory have to be in sync before anything can happen. The 3:2 divider makes the memory be in sync more often than the 5:4 divider.

Results (rendering out shadow.veg from "marker" 2 to 1)
No divider (1.6GB/s) :46, :46, :46
5:4 divider (1.28GB/s) :47, :47, :46
3:2 divider (1.06GB/s) :47, :47, :47
(3:2 divider with low/fast timings) :48, :47, :47

The results aren't very exact but as you can see, memory bandwidth barely makes a difference. The last test I ran to see if tight timings (2-3-2-5 versus 2.5-4-4-8) would make the 3:2 divider as fast as no divider. The results show that it is slower, but keep in mind that there's variation in Vegas rendering speeds. I could measure the difference more accurately but it seems to me there's little point. The performance difference is just too small to care about!

Please note that I'm running single channel RAM, which is slower that dual channel RAM. Dual channel RAM would generally show the same thing: tight timings and memory bandwidth doesn't matter that much.

Keep in mind that these results only apply to Vegas on a Pentium processor. Other programs and processors will have different results. For benchmarks of other programs, see "Choosing Optimal Memory to Match Intel Pentium 4 Processor" http://www.xbitlabs.com/articles/mem...ay/p4-mem.html.

Machine specs:
Pentium 2.6 800FSB hyperthreading
Asus P4P800 (FSB=200.5X4), i685PE chipset, MAM and Turbo on AUTO
PMI RAM 1X512MB
winXP Service Pack1a
Vegas 4.0c

Actual memory bandwidth:
1519MB/s (no divider, SPD/default 2.5-4-4-8 timings)
1641MB/s (no divider, 2-3-2-5 timings)
1226MB/s (5:4, SPD/default timings)
1082MB/s (3:2, SPD/default timings)

-----------
While running the tests I did find out that Folding@home causes a noticeable hit to performance (15-25%ish). Folding@home is a distributed computing project which uses your computer's "unused" CPU cycles to simulate how proteins fold.