Minimum system upgrades for HDV on Vegas
 

HDV continues to drive me nuts in post. So *many* little things that work fine in DV and stop me cold in HDV. Tried many approaches and workflows...

Today's issue: Need to extract about 1.5 hours out of 18 hours of M2T footage already on my drives. I've finally decided to punt - this little subproject is headed for the web, so I set two overnight renders to DV25, .5 and 1 hour, respectively.

Color correcting, reduce interlace flicker, video quality best - obviously a longer render, but DV25 is an intermediate for this subproject, not a proxy.

Surprise! (not) The longer render is still going this morning, it will total about 17.5 hours.

I notice that both halves of my hyperthreaded processor are maxed at 100% - good, thanks Sony, 6.0c really does better with threading.

I notice that page file usage is consistent at about 1.15GB.

**********************************
Is this a clear indication that I need another gig of RAM - is paging slowing down renders significantly? Or am I just fooling myself that my P4/3.06 with external ATA100 drives with no raid can even keep up with proxy and intermediate rendering?

I am rendering from one external drive to another.
**********************************

System is:
Vegas 6.0c
P4 - 3.06GHz, hyperthreading "on"
Intel D845PEBT2 MOBO
1GB RAM
ATA-100 system drive
3 ATA-100 externals in FW/USB cases, currently running USB2

i think that a number of people out here have posted about using the womble mpeg editor with hdv, for simple tasks like viewing footage and cutting chunks out of it, like you just asked for... there is a downloadable demo you can try.

>>but DV25 is an intermediate for this subproject<<

you are rendering from hdv to dv25 first, then rendering the same footage again, to a web video codec? why not go directly from hdv to the final web render? i'd imagine that you'll get much better quality that way.

Well, cutting chunks out is really just the start of a real edit. You think Womble would give better performance for the transcode? I don't know much about it, but I suspect it doesn't have the color correction capability that Vegas does. Better for me down the road if I color correct now and take the hit on render time at this first trancode. I will check out Womble though - if just for logging HDV.

Two reasons:
1) M2T playback in Vegas is anywhere from 2-12fps.
2) Vegas is much less stable for me in HDV than in DV.

Have you had a chance to edit HDV in Vegas? I'll agree with you on the theory of somewhat better quality direct from M2T to delivery codec, but Dan, if I can't put it into practice with a good workflow it doesn't matter much, does it? My render is finally done, and I judge the DV intermediate as quite good enough for today's project. Just as important, it satisfies the goal of getting me into a familiar workflow where I can more accurately estimate how long it will take in cutting and rendering. If I can't predict these things, I really can't make a commitment to my client.

My question is really about system specs to support this new world of HDV - with that much page file useage, am I stupid not to upgrade RAM right away (these subprojects are only the tip of the iceberg of what I will be doing with this footage), or is that a somewhat meaningless upgrade against potential upgrade of processor by a few MHz and larger FSB... or is it RAID architecture that is the greatest need?

sorry for the delay in getting back on this... the situation above could be tailor-made for womble, because it previews mpeg2 really fast... and if you just want to condense chunks of footage down like that, you won't have to do any rendering at all, because it's cuts-only edits... womble will do frame-accurate cuts, but i'd be looking to see if there is a program setting to cut on the gop's instead, since you are just roughing it out.

i don't think that womble will transcode to dv, but it could save you a whole bunch of time, and a lot of wear and tear on your computer hard drive.

the 30 day trial is free, you have nothing to lose by trying it out.

I have the following system and it's impossible to edit the m2t stream. I use the Cineform interm avi's and it works much better

Dual Core P4 Dual Core 3.2ghz, 3gb of Ram, 10k SATA OS with 4 TB of storage, of which 1 gb is raid.

It takes me approximately 3 to 1 to render to m2t when I have effects, slow motion or pan/crop.

When it's just cuts, no extras I get 2 to 1.

Jon

Seth, this is EXACTLY what GearShift was designed to do.
Put your m2t on the timeline, select "keeper" regions, run GearShift, it only converts what you want converted, to both CineForm and DV proxy at the same time (if u want)

You can also do this with the Batch Render tool in the Scripting folder, it just doesn't allow you to swap back and forth between the m2t, CineForm, and proxy

At womble.com, in addition to MPEG Edit, they have a program called MPEG VCR. If you by MPEG Edit you also get MPEG VCR for free.

MPEG VCR is the program I use for extracting the "good footage" from m2t files. You just load in the m2t file you want to cut into MPEG VCR, select your first in and out point, hit the little "create a new file" icon, hit the "batch" option and assign it a file name. Then you move on to the next file you want to split. When you're done selecting all the video sections you want, you run the batch file. It doesn't really rerender it, it just inserts a new header on each clip and file copies the part you need. A half hour or so later you have a directory full of just the parts you need in first generation m2t format.

MPEG VCR is a wonderful tool for splitting m2t files. You can set up all your split points really fast, and instead of waiting for a render, all you have is basically a file copy. All the trimmed files are exactly the same quality as the originals.


From there I run Gearshift and work in the proxy/m2t render format from that point on.

I was hoping to run Gearshift as DSE describes from the timeline, but ONLY create a DV proxy, not CineForm. But Gearshift doesn’t support that (at least not yet) – it only works from the timeline if you create a CineForm file (plus optionally a DV proxy as well).

I see 3 ways I could work, I’m trying to decide which is best:

1. Do as Laurence does, and buy MPEG VCR, use this to do a “pre-edit” of the m2t files, then use Gearshift to create the proxies (from source files, not the timeline). Edit the proxies, shift gears to m2t for render.

2. Have Gearshift create CineForm files and DV files from the timeline. Edit with the proxies, shift gears to the CineForm files for render.

3. Have Gearshift create CineForm files only from the timeline. Edit these and render from these.

MPEG VCR seems good value, but I’d like to avoid adding more software if possible. Option 2 seems a waste of space if I’m not editing the CineForm files, I don’t believe there is any quality difference, and requires more rendering. I didn’t think my pc (1.6 Ghz Pentium M laptop) was up to option 3, but actually it does pretty well editing CineForm directly, as long as I set the preview to “preview” quality (at good or best it can’t keep up the framerate). Both options 2 and 3 take more disc space, but I’m going to add disk so that won’t be an issue.

My current thought is to try option 3… and if I find my PC struggling (which it may if I do more complex composting) then move to option 2.

Mark

GearShift will create ONLY the DV proxy if that's all you want. Just leave the second file format option drop down box set to "None."

Yes, but this not supported from the timeline.
So I would have to create a proxy of the entire source files.

What I'd like to do is just select the footage I want to keep from the timeline, and create a DV proxy for just that footage.

Quoting the user guide, for both the Timeline and Region options:

"Both of these options require you to render both a DV proxy and an intermediary format to keep the files one-for-one. (i.e., DV proxy and M2T Only is not available for this option)"

Mark

Which version of GearShift are you using? 1.0? I'd recommend you upgrade to the latest version.
You can create DV proxies, and then convert to the m2t's on the timeline in the Shift if you wish. You do not have to build any form of intermediary should you not want to.

I'm using 1.5.2.

According to John Rofrano, this isn't currently supported.
http://www.dvinfo.net/conf/showpost....0&postcount=26

Sorry, I missed the "timeline" in your post. We'll have that updated soon.

That would be great! Thanks!