Yea, definitely some flavor of AVI. Cineform, Lagarith (which is free and what I use), Huffyuv. or something similar.

CineForm AVI are Vegas 9 work fine together since Sony released a patch in their knowlegdebase : http://www.custcenter.com/cgi-bin/sonypictures.cfg/php/enduser/std_adp.php?p_faqid=4558 This patch should not be needed with 9.0a.

Also CineForm 64-bit is available as a public beta:
http://www.sonycreativesoftware.com/forums/ShowMessage.asp?MessageID=660368

David Newman
CTO, CineForm

what's wrong with mpeg-2? HDV is that & i find it very easy to work with.

You mean besides it being lossy and long GOP?

HDV editing has it's place and it's uses. You can shoot a home or school event video, put the HDV clips right on the timeline, edit it, add some music and narration, and smartrender it into a master that will burn onto a Bluray disc without a single destructive rerender throughout the whole process. That's the good news.

The bad news is that if you do any kind of color correction, filters, text overlays, motion stabilization etc. you have to rerender and m2t rerenders look pretty bad even after a minimal number of rerenders.

That's where a codec like Cineform comes in. You can add color correction, filters, text overlays, motion stabilization and all the tricks it takes to get a really wonderful looking final product without the damage associated with multiple generations and a long GOP codec.

What about the Sony MXF option? Is that good enough for editing and then rendering out a release?

It certainly plays back at full frame rate.

The Cineform 1080-60i reduces the AVCHD 1920x1080 frame to 1440x1080. It takes 3 minutes to render a single 12.5 second clip on my ancient Pentium 4 machine.

The resulting 229MB Cineform intermediate will play back at full frame rate, but only in Preview. By comparison, the original 37MB AVCHD 24 mbps clip will only play back at 2 fps in preview.

Isn't MXF using the Mpeg2 codec ?

I have found .AVI to be the best for me.

I concur with others, having tried several other formats, CineForm is my main intermediate format. For HD work I upgraded to their NeoHD which can encode to 1920x1080 (unlike the Vegas in-built one that, in V8 at least, only goes to HDV 1440x1080, and is a significantly older version). Same size implies faster render.

David Newman, thank you for highlighting the existence of that patch - that was my very question before spending any time entertaining the new version 9. I was just about to search/ask on this forum.

Also I use Cineform (the same NeoHD) for lower def resolutions, even down to 320x240 - that gets rendered onwards to a variety of streaming formats. I just wish the license could be extended to double-PAL, i.e. 1440x1152, for storing the results of resolution-enhanced DV (a few low-cost tools do this).

The only nuisance with these intermediates, same for HuffYuv etc., at least in PAL (LFF) land or when progressive, is when using such a media in a Vegas project, you have to remember to change its properties manually. As far as I have been able to figure this out, this is a deficiency of the AVI container (at least via VFW) which doesn't say what the media format is (e.g. 16:9 or progressive) so Vegas appears to use "rules of thumb" which, sadly for me at least, defauts to 4:3 UFF.

Rough-Science Experiment: To compare formats for accuracy, put a pair of them in parallel tracks, synch them (e.g. helps to temporarily reduce the top one's level/opaqueness) (put the top track back to full level opaqueness), put the top track in "Difference" mode, add a "Levels FX" as a "Video Output FX" and tweak down its "Input End" slider (almost down to the minimum) until weird black-and-white outlines etc. appear in the Preview. Now these differences are clearer you may need to synch the two clips better (a frame or three either way). Note the Levels setting for any future comparisons, so that they will be consistent with this. It is interesting not only to confirm that there lower differences but also too see a distinctively different quality of differences in the case of (wavelet-based) Cineform as compared e.g. to (DCT-based) Mpeg. Even HuffYuv shows some miniscule differences under this "microscope" - it really picks out minute details. Of course, what matters is not this microscopic study so much as the pleasingness of the end product, which is subjective / psycometric / subliminal. But it's good to get the "multiple generations" potential pitfall out of the picture.

AVI and MXF are just container formats. They know how to interleave audio and video but don't know/care how that video and audio are encoded. Same for Quicktime (.mov) and some other more obscure ones.

AVI typically contains video encoded as DV but it could also be uncompressed, HuffYuv etc., Cineform, Indeo or any of manyothers. Same for MXF e.g. Sony EX ClipBrowser generates MXF containing Mpeg2-encoded video.

The bad news is that if you do any kind of color correction, filters, text overlays, motion stabilization etc. you have to rerender and m2t rerenders look pretty bad even after a minimal number of rerenders.

most ideas listed are lossy, I wasn't assuming a lot of re-renders.

For space, QT-PNG. Uncompressed @ a lot less space (maybe less then a compressed file).

But if you're just doing intermediates then the format won't REALLY matter as you'll replace them with the good files when you're done. I suggested mpeg-2 because you can use the HDV template & Vegas edits that pretty fast vs other formats. Or go down to DV just for editing?

I still have the occasional problem with HDV believe it or not. Usually its just a frame that won't smart-render and causes Vegas to freeze. I can then go and render it to Raylight or Cineform AVI, but not back to HDV.

I use Raylight, just cause I have it, I think it compares well to Cineform, but I have not done the extensive tests that have been done recently.

Other than those, I would vote for Logarith and Sony MXF before HDV.

kairosmatt

Different people understand different things from the term "Intermediate", I was confused by this originally:
a) A temporary disposable lower-rez format for offline editing, where the file is temporarily replaced, to be re-replaced later on for final product rendering. I used to use MJPEG for offlining (but better machine now removes the need for this at all).
b) A stepping-stone file, rendered out from some quality-enhancing project, to be relied upon for any further editing, e.g. better than the original in some way e.g. color-graded.

Intermediate-b may involve several steps e.g. Source -> QualityGraded, Graded -> Denoised, Denoised -> Effected, each as a separate project taking in the file rendered by the project of the previous step. That's where "multiple generations" comes in. In the "old days" (a couple of years ago) I tried using DV for this but noticed some detail getting blurred, so it was the problem not the solution that drove me. Tried lots of formats to find something better. For Vegas, could instead just use nested projects (no need to render any intermediates) but then the framerate gets frustrating and it generally stretchs the system more (makes me concerned about stability).

Right now I have a 3-hour 4-cam project, using CineForm NeoHD. Interestingly, for one of the cams ( Z1 - HDV) the denoised version was 97GB as compared to 122 GB for the non-denoised version, so that's another factor to take into account.

Sounds a lot but only about double the size of the original source, unlike which this format is good for several generations. Hate to think how many times bigger HuffYuv or Lagarith would be. Wish (back in less demanding days) I had bought a 1TB GRAID instead of a 400 GB one though...

I've done (and posted) a number of experiments using the "difference" methods. It's how I came to use DNxHD for intermediates rather than anything else. I found the proxy level of that codec was nearly lossless for my XDCamEX footage.

I've just completed a 25 scene training video for work. End to end 1080p and used DNxHD all the way through. Three generations of rendering, and the end product is absolutely fantastic.

Cineform would probably be faster, but I had a number of problems with it over the past couple years and just wanted something different.

I understand that DNxHD is wonderful except for the fact that the only way that Vegas can use it is in a rather inefficient .mov container, which makes it quite lot slower than working with Cineform in an .avi container.

Absolutely. It is a significant drawback to using the codec. Another reason to cut on proxy files (or Cineform).

Maybe the new 64bit version of Cineform will allow some new workflow for me.

I often use PicVideo - it is faster than Cineform and the file sizes are smaller (if you put the quality slider down to 19 instead of the 4:4:4 option at 20 on the quality slider), and quality is good at 19 (although after 6 generations, you can see that Cineform is better - see http://www.sonycreativesoftware.com/forums/ShowMessage.asp?ForumID=4&MessageID=655619); after only a generation or two, it still looks visually lossless. It costs $40 if you buy it from www.resolume.com/shop/codec.php

I like PicVideo and there is a 64 bit version available. My problem with it as compared to Cineform is that it doesn't smart-render so every single time you redo something you need a lossy rerender.

Laurence, what is your workflow to get from HDV to Bluray without re-rendering? The only way I have found is to render to the BluRay 1440x1080-60i template that has been modified to High 1440 level and 25 Mbps CBR. That will smart render in Vegas and then I have to use Encore to make the bluray disk without recompressing.

I have not found any combination that will smart render HDV from Vegas and that DVDA will not insist on recompressing.

thanks,
Rocky

I haven't done it with DVDA yet. I did it with Ulead Movie Factory plus. There must be a way of doing it with DVDA too though. Have you tried separating the video from the audio?

Eugenia the Lolly-Lady posted a comparison of intermediates: http://aggregator.foolab.org/fooaggregator/source/216?page=4

I always render audio and video separately.