I use Sony .MXF as well as Cineform .AVI for HD intermediates. My AVCHD camera produces .MTS files which I usually shoot at 30p.

I use basically four formats:

1/ Cineform for highest quality (smart-renderable)
2/ MXF for almost the same quality at a fraction of the size.
3/ Quicktime .mov with png compression for anything with a transparent alpha layer.
4/ Quicktime .mov with Avid DNxHD codec for Handbrake encoding intermediary and for working with the FCP world.

I find I'm using MXF more and more as I simply can't see the difference between it and Cineform and since it previews beautifully off small bus powered USB 2 drives which are so darned convenient. Cineform lets me go back and forth between After Effects and Vegas though.

I agree with you Laurence,

When working on my last project I used MXF 4:2:2 instead of Lagarith. I was very happy with it.

The big thing with .mxf is to make sure that you use it interlaced even though the chances are very good that you are using progressive footage. I set it using one of the interlaced templates but set the deinterlace method to none.

The reason this is important is that Vegas will only smart-render .mxf footage flagged as interlaced. If you set the .mxf render properties to progressive, it won't smart-render. If you set the properties to interlaced and select either blend fields or interpolate, it will screw up resizes and renders to other formats.

Just remember these two things: 1/ Set it to an interlaced template even though it is progressive. 2/ Set your deinterlace method to none.

.MXF with a smart-render is very cool. The format looks wonderful and no damage is done as you smart-render sections into a final piece.

.MXF without a smart-render isn't really good enough. .MXF will not hold up to successive rerenders like Cineform or a lossless codec will.

I shoot in AVCHD 1080x1920 60i. I edit the AVCHD clips on the timeline and then render out to mainconcept mpeg-2 SD for my final DVD.

If I was to convert the clips before editing to MXF 4:2:2 would this give me a better copy when rendering out to mainconcept mpeg-2 SD?

All it would really do is speed up your previews and renders. Since it adds an extra generation in there it is slight compromise in quality over the way you are already doing it.

In my case I am working with a Core2Duo that isn't really up to the task of AVCHD. Not only that, but the Nikon video is in a .mov container and in cRGB color space (at least in Vegas lingo). Converting it to .mxf solves all these problems in one go and I really can't see the quality hit.

I have shot in 720 60p. I will be outputting to DNX?? (Avid) per previous threads for web use and perhaps a smaller version for electronic distribution.

I assume that a stop in MXF, in between, won't shoot me in the foot?

Rob

The reason I asked if it's better to convert my AVCHD to MXF 4:2:2 HD before editing and then render out to mainconcept mpeg-2 SD is not so much for better preview playback but will it give a better copy reason being MXF 4:2:2 HD has better color space and video information is in ever frame.

The question is when down converting from MXF 4:2:2 HD to SD even though we went and extra generation will it look better? Will it have less aliasing, artifacts and moire?

Avid DNxHD 10-bit (4:2:2:) is lossless in my tests.
Avid DNxHD 8-bit is not. Nor is MXF, in fact the two are similar in that respect.
Not that you would ever notice a difference if you were doing one or two intergenerations.

Using a 4:2:2 intermediate will not make 4:2:0 source "look better." Personally, with all the CG you describe I would use an RGB intermediate, but then the losses would be the same rendering back to mpeg-2 SD anyway.

Have you read Nick Hope's HD to SD thread? Are you aware that just the act of downsizing from 1080 to 480 throws away 83% of the picture data? That is nonrecoverable.

The concerns in your last question (aliasing, artifacts and moire) are products of the resizing and deinterlace processing used, and the output itself, not the intermediate codec used. You won't be able to smart-render MXF after it has been changed for MPEG-2 SD output, if that is what you are asking.

I agree Mark, but there are times when some of these things still make a difference.

For instance, in my case with my Nikon D5100 camera. It shoots with 8 bit color in cRGB color space. When I convert to .mxf I use a 32 bit timeline and use a more defined 4:2:2 color space, the subtle changing of the colors between cRGB and sRGB will take advantage of that. If you were using an AVCHD camcorder which is already in sRGB color space there would be no advantage to using 32bits or 4:2:2 color for making the intermediates.

Yes you through away most of the picture when you downrez to SD, but somehow, the more pristine the original, the less damage this seems to do.

I can see that. So many people assume DSLR footage is AVCHD, when it's not. Along with hundreds of millions of Point-and-Shoots, Pocket HD, Gopro, Hybrids, iPhones, Androids, etc. etc.