It's a question of what your priorities are. There is no obvious difference to me in quality between original and re-rendered HDV images. I have not done a careful A/B comparison, but I can guarantee you that my customers have not either, given that all they see is the final standard-def video on DVD anyway.

I just finished editing a 2-hour programin V7 with two cameras, and all clips were color-corrected. I used the raw .m2t files, not intermediates. It worked OK.

This may not help the way the timeline plays but if you add marker and ensure the Render I-frames at markers check box is selected before rendering (in the Render As dialog, choose MainConcept MPEG-2 from the Template drop-down list, and then click the Custom button. In the Custom Template dialog, select the Video tab and select the Render I-frames at markers check box) you may find a work around.

I just looked an I am surprised that if you zoom in on a frame and cut that frame to make a single event the event properties does not indicate the frame type.

jbeale....how long did it take you to render it out? How many times did you render it before you released the fInal version? core 2 duo's are fast, but render time is render time. If vegas had "smart rendering", you could save most of that render time. And avoiding generational loss is gravy.

mhstevens...hmmm...rendering new i-frames at markers will still require interpolating new i-frames from B-frames. This is the issue as I understand it.....only the i-frames are close to the original image. In an uncompressed file, the B-frames are used to reconstruct frames based on the I-frames....losses occur on the reconstruction. Everytime you re-render, you create new I frames from reconstructed B-frames, which is why the re-rendering process causes losses. If I can cut the clip at known I frames, I don't have to reconstruct I frames from reconstructed (interpolated) data. Maybe I'm just talking geek, here.

>jbeale....how long did it take you to render it out? How many times did you render it before you released the fInal version?

Don't know how long; I ended up with a few final renders but they were overnight while I was asleep and it was done by morning, so it didn't count against work time. Other than that I worked direct from the timeline, no intermediate renders. Using intermediate formats, the render times were still long, and it would have taken 3x as much disk space.

Ah, possibly stupid point here but I think HDV specifies a fixed length GOP of 15 frames. You might get something different to play off the T/L but it'd not print to tape. That might be one reason why Vegas will insist on re-encoding it.

Bill, the comments direct from Sony on this forum indicate that you'll never get anything better than a one-time, direct render from the original HDV to whatever your end format is. Sounds obvious, but has lots of implications.

For absolute maximum purity, you're probably right that you should do it in Womble (for "smart rendering"). But if one generation worth of loss isn't a heartbreaker, you could do it straight from Vegas without using an intermediate. Because either way, (direct or intermediate) you are going to incur at least one generation of recompression. And most intermediates aren't 100% lossless.

Sony insists that in any case where you aren't doing multiple generations (i.e. more than one) of rendering, you should use the original HDV, regardless of color correction, compositing, etc. As Wolfgang commented, the complexity of color correction makes it more viable to use an intermediate if your processes require multiple generations of rendering (which again becomes the decision point).

Remember it's not just a quality issue: it's a HUGE time issue as well. Womble will chop up an m2t file about as quickly as a file copy. If you batch file a bunch of Womble edits, they will chop in the background with next to no CPU overhead. A rerender like you need to do with Vegas when working with mpeg takes however long it takes to render the file. During this render time, your CPU is usually too busy to do much else.

ahhh...exactly my point!
So, this becomes a wishlist for vegas 8...smart rendering ala womble.

Someone explain to me how it is 'smart' to render a non compliant stream?

Sounds to me like a formula for all manner of grief down the track.

Bob.

If you're doing 'cuts only' V7 handles m2t very well and the worry about I-frames etc academic. If editing was restricted by the mechanics of the 15 frame GOP it wouldn't be practical and you'd have to use intermediates. Editing native m2t isn't my preference but I've recently done a 84 minute job (final length) that way with corrections and transitions and render time to DVD was 8 hours (overnight). Clients very happy.

The point is that if you are working on a documentary and you are shooting vast amounts of footage, you need some sort of a tool to chop out the usable segments from your raw footage so that you are submerged in hundreds of gigabytes of useless junk. In the DV world this is easy as Vegas smart-renders DV footage. With Cineform this is easy as well though the file sizes are a lot larger. With m2t footage however, the only way to losslessly pare down your footage is with some sort of native mpeg editing tool. If it is a simple project you probably don't need to bother with this, but if you are building up documentary footage over a long period of time, it really is a neccessity.

In this scenario I can sure see how things could be done better.
Given that you don't need frame accuracy a 'Smart / Rough Trimmer'
would be ideal. You just define the first and last frame you need and it finds the first I frame before the in and after the out and then creates a new clip / file, no encoding is needed, just a new header and all the frames are just copied, could even be done in the background or batched. The result would be fully compliant HDV stream with no re-encoding at all.

Probably the thing could be a standalone app for that matter.

>>>a tool to chop out the usable segments from your raw footage so that you are submerged in hundreds of gigabytes of useless junk<<<

Was that the substance of the original query for this thread? I read it differently. However if one does have a lot of useless junk then obviously one wants to delete it quickly and I see the value of the proposal. But you have got me into trying to imagine a scenario where this might happen. Obviously wildlife documentary would be one area where you might shoot for a long time to capture a moment. In less uncontrolled circumstances a lot of the deletion is achieved by not pressing record, and otherwise my experience is that clips or portions that end up "on the cutting room floor" are not to be pre-judged at a crude level of sorting. In a documentary, clips that at first viewing seem to be irrelevant can find an important place in another sequence. Even stuff like "kiddy football" must be difficult to sort the junk from the good until editing. Of course here I'm presuming that technical faults haven't created the junk.

no, Laurence and farss have it right. I capture on the order of 7 hours of footage from actors/musicians. Many times, this 7 hours, or so, has to be editted down to a 5 minute trailer, which is then used to get additional funding for a feature length video. So, as you can surmise, extensive cuts only editting is a real driver. It's ineffecient as hell to have to render-rerender many times over. Each "final" version goes up before a committee of reviewers. Needless to say, the re-editting can go on all night...it's frustrating as hell. I'm beginning to learn womble's editor...if only because it can cut/quantize to I frames only, as described by Bob and Laurence.

>>>Each "final" version goes up before a committee of reviewers.<<<

Sounds like enormous fun! I'd forgotten that was your line of production.

IMO the lack of "smart-rendering" in Vegas is a real short-coming. To get around this, suggest a couple of tools. First, TMPGENc's MPEG Editor will do frame-accurate cuts editing and re-render only the frames around the cut. It's very fast and produces compliant HDV m2t files that can be then be imported onto the Vegas timeline. Alternatively, you can go to the dark side and pick up a copy of Premiere Elements and then get the MainConcept MPEG plug-in which will do smart-rendering. The plug-in for Elements is only $60 vs $400 for the Premiere Pro plug-in. Again, it produces compliant HDV m2t files which can be imported into Vegas at a cost much less than Connect HD if you're only concerned with basic cuts editing.

Will Premiere Elements let you use plugins like the Dynapel Steadyhand one? Trimming m2ts and deshaking some of them at the same time would be interesting to me.

The key for me is which one of these smart rendering packages will accept an EDL, or some sort of intermediary edit list, from Vegas so that I don't have to cut all over again or manually type in "in" and "out" points. I suspect Premiere Elements might accept a Vegas EDL as is.

This thread seems to have raised a "new" argument in favor of Cineform over native m2t editing. I have done little native editing in V7 so I can't say for myself, but wondered if those of you who have done both (I think Serena has) thinks this is a valid argument in favor of Cineform or not.

even the cineform codec has me wondering. isn't the cineform 8-bit codec also mpeg format in an AVI wrapper? it's just a short form GOP, i beleive. so the re-rendering after a cut argument still applies. perhaps the short form GOP minimizes interpolation errors...but u still have to re-render every time a cut is made. wouldn't it be nice if you only had to re-render the frames on either side of the cut, rather than the whole bloody file?

>>>isn't the cineform 8-bit codec also mpeg format in an AVI wrapper?<<<

No, it isn't. A neat explanation can be found in cineform format

and advantages of the codec are nicely illustrated in image quality

EDIT: CineForm Intermediate uses a temporal Wavelet transform. It has nothing in common with MPEG, including MPEGs poor multi-generation characteristics. We have a brief description of our codec in our technology section here:
response from Cineform

ahhhh...thanx, so much, Serena. Looks much better than I thought....and, in fact, rendering is almost as fast as so called smart rendering.And, Cineform even supplies a registry switch to be able to play m2t files without skipping...sweet.