Comments

fldave wrote on 7/30/2006, 7:34 PM
Any effects, croping, etc? Captured m2t to Cineform with no timeline/media effects or resizing should not take that long.
johnmeyer wrote on 7/30/2006, 8:14 PM
If you are just generating an intermediate, for use in subsequent editing, then this is WAY too long. On my 2.8 GHz single-thread, single-core, single processor, 3.5-year-old PC, I can render from m2t to Cineform intermediate in about five hours for every hour of video. Therefore, your 30 minutes would take about 2.5 hours. On your PC, it should be less than an hour, I would think.

If you have the newer version of Cineform (which I don't) I think the conversion is even faster.

You definitely do NOT want to do anything except the conversion (i.e., don't do color correction or anything else) in order to keep the conversion as clean as possible. Also, you do NOT need to use Best, although if you have the complete version of Cineform, I understand there are more options, and I can't advise you on what to set. This was asked, and the Cineform CTO answered, about six weeks ago, so if you search on his user name, you'll find the answer.
frazerb wrote on 7/31/2006, 5:25 AM
Thank you for your replies. No, I did not have any effects or transitions.

My rendering times are usually very good. Sometimes I can can render a SD project with lots of transitions and filters to mpeg-2 in near realtime.

Any ideas on why this was so slow? (I have a recently defrag'ed hard drive with lots of free space that is not the same drive as the progam is on, so that's not it.)

Next time I will try re-booting first.

Buddy
fldave wrote on 7/31/2006, 5:46 AM
Make sure that your Project properties are HDV 1440x1080, in other words match the input m2t (not sure if you are PAL, NTSC, 25fps, 60i, etc).

If it's set at say, NTSC DV, then I think vegas will try to convert m2t to that size, then resize again for the final output. Theory, at least.

When the render is done, put both the m2t and the Cineform on the timeline one on top of the other. Then turn the top track mute switch on and off while closely viewing the preview. You should not see any difference between the two. If you do, either there was some resizing going on or the track opacity got bumped or something.
frazerb wrote on 7/31/2006, 3:55 PM
I have checked all of those things and rebooted. I ran a short test, with the same results--it is taking 10-12 times the clip length to render to the intermediary.

I have also noticed that the computer is not using much of the processing power. The Task Manager reports about 30% of the four processor threads and about 480mb of RAM out of 4GB in the system.

Any other ideas?

Buddy
Chienworks wrote on 7/31/2006, 5:40 PM
The low processor value is troubling. You must have something else hiding in the background that's sucking up cycles.

Defragging makes zero difference, for all practical purposes. Don't ever do it again. You're endangering your data for no measurable gain.

Project settings are pretty much ignored while rendering. Render settings override project settings.

HDV just takes a long long time. Vegas has to process about 5 times as many pixels as for SD. That alone will nearly quintuple rendering time. Find out what's stealing the other ~70% of the processor and you'll be just about on track.
fldave wrote on 7/31/2006, 6:34 PM
Norton/other antivirus running?
Drive DMA at max setting?
Render "to" drive different than Project media?
What are your "Video" tab settings under main Vegas options?
Jay-Hancock wrote on 7/31/2006, 11:20 PM
As was written before, make sure the output format exactly matches the input format. But also, make sure the output format isn't progressive, that the frame rate doesn't change, and that video render quality is set for good (not best). In the preferences there is a video tab that includes a setting for max # of threads that can be used for the render. If yours is any less than 4, you'll get less than optimal results.
frazerb wrote on 8/1/2006, 5:34 AM
When I render the same unedited m2t file to an DVD Architect NTSC mpeg-2 file, the processor utilization runs about 75% and all four processor threads show equally high usage. A 5 minute clip takes less than 10 minutes. Rendering an avi file with the Cineform intermediate preset shows processor utilization at less than 30% (again all 4 processor threads are in use).

So why would the Cineform codec use less processor?

Buddy
fldave wrote on 8/1/2006, 6:39 AM
I just tested a 1 min m2t clip. I have a 3.2 pentium HT.

To DVD Arch NTSC widescreen, it took 2:44 at best, file size 42MB. Both "cpus" were averaging about 80% usage.

To Cineform intermediate, at "good", it took 3:15, file size 640MB. Only 47%-55% utilization. The avi rendering is single threaded, (I think it even says it in the manual, at least for the DV codec). So for 30 minutes, you'll need about 20GB?

My page file usage was about 390MB, Dynamic RAM render setting at 0. When I bumped it to 1024MB, my page file usage went to 1.39GB (I have 2GB RAM).

It sounds like your hard drive can't keep up, or you are spending a lot of time swapping in the page file. Run some HD Tach-like tests to see what the throughput on your hard drive is.

Edited:
Dumb question, but are you sure that the m2t is NTSC and not PAL, and it is really HDV, not SD?

frazerb wrote on 8/1/2006, 11:43 AM
Thank you for the data, I'll keep running tests.

Yes, I am sure everything is NTSC and the recording is HDV.

Buddy
frazerb wrote on 8/1/2006, 1:08 PM
I ran an HD Tach test on the hard drive to which I am rendering the files. The results were:
Burst speed: 227.8MB/s
Ave. Read: 52.3MB/s
CPU Utilization: 1%+/- 2%
Random access: 15.2ms

I am no expert at reading the numbers, but using the comparison feature, the results looked good for a non-RAID device.

Anybody have an opinion on these numbers?

Buddy
frazerb wrote on 8/1/2006, 1:08 PM
I ran an HD Tach test on the hard drive to which I am rendering the files. The results were:
Burst speed: 227.8MB/s
Ave. Read: 52.3MB/s
CPU Utilization: 1%+/- 2%
Random access: 15.2ms

I am no expert at reading the numbers, but using the comparison feature, the results looked good for a non-RAID device.

Anybody have an opinion on these numbers?

Buddy
fldave wrote on 8/1/2006, 1:56 PM
The numbers look fine to me.

The only other think I can think of is to review these two pages for optimizing your pc for video, they have been referred on this site before.

http://www.videoguys.com/WinXP.html
http://www.videoguys.com/WinXP2.html

Also, regarding anitvirus, many people have had lots of problems with some with video editing (like nnnooooorrrrttoon). I have mine set up to only check incoming data from internet, lan, email. Some AV programs additionally check every file that is read by your system, which could greatly drag down performance.

Edited: by the way, my WD Raptor has 67MB sequential write
frazerb wrote on 8/1/2006, 2:11 PM
fldave,
I changed my Dynamic RAM Preview to 0 (from 64) and now a 5 min m2t file renders to the Intermediate avi in 17 min. That is a massive improvement, but I think I should get better.

I'll check my anit-virus settings and look at the Video Guys pages.

Thanks a bunch,
Buddy
fldave wrote on 8/1/2006, 2:19 PM
Wow, huge. 5min-17min is about the same ratio as my 1 min/3:14 test render above.

If you want to see more about what I think about Dynamic RAM settings, you can check out my testing here:
http://www.visualretreat.com/vegas/2005_V5V6_Compare.htm

I think Dynamic RAM should never be between 0 and 128MB. 0 is fine, as is 128, but not in between.
frazerb wrote on 8/1/2006, 2:52 PM
My Dynamic RAM Preview was at 64. So, I would suggest nothing higher than 0 when one is rendering. It can be set back to something higher during editing.

Buddy
Jay-Hancock wrote on 8/1/2006, 3:31 PM
are you sure that the m2t is NTSC and not PAL, and it is really HDV, not SD?

At the risk of sounding like a wiseguy, it might be worth mentioning that in the HDV world there is no such thing as NTSC or PAL. For 1080 size HDV, only there is 60i and 50i (and maybe 1080p). Unlike NTSC vs. PAL, the color system is identical for HDV regardless of whether you are shooting for Europe or USA or Asia or whatever. Only the framerate changes. Effectively the terms NTSC and PAL are practically meaningless in the HDV world.

I would suggest nothing higher than 0 when one is rendering.

This doesn't always hold true; I have many projects where setting RAM Preview to 0 drops CPU utilization on the additional threads and greatly lengthens the render. And it is 100% repeatable. But the Veg file that fldave posted does need the setting at 0, so there doesn't seem to be a reliable "rule of thumb". Basically, I found it best to do a real short test render on my projects to see what works best. This is discussed in this thread
fldave wrote on 8/1/2006, 4:36 PM
"m2t is NTSC and not PAL"

Yes, I meant 50i. I have an FX1, so I am locked in the 60i world for now.
frazerb wrote on 8/1/2006, 7:44 PM
I hope Vegas 7 makes some improvement--or at least creates more consistency--in this area.

Buddy