New DI Codec

Question

I wanted to start a new discussion so that others might have the opportunity to post without drawing an earlier thread off topic.
For a while now I've been using Lagarith as my codec of choice for Digital Intermediate work.  Lags is a free, lossless video codec that provides all the quality of Uncompressed, but at smaller file sizes (though still somewhat larger than DV and other camera formats).  Lags has served it's purpose very well as an intermediate export format that I use to create my MPEG files with.  (HC Encoder just does a much better job than AME, but it needs an exported file to work with.  It can't encode directly from a Premiere sequence.)  And recently, under the premise that fewer compression steps is better, Lags has served well as a source format for editing.  I convert my DV clips to 24p Lags files using the dv2Film process, then bring those Lags clips in to Premiere for editing.  I then create a Lags export for MPEG conversion.  This process has only two compression steps, the original DV (which is unavoidable when shooting with a DV camera), and the final MPEG stage.  All in between steps use the lossless codec Lagarith.
After some recent testing, I have switched over to the UT codec (v 7.0).  While I am relying somewhat on the testing of another regarding it's visual quality*, I have not noticed any problems with the picture using my own media and tests.  What I HAVE noticed is that the final MPEG encodes at about 2.5 times the speed of a Lags DI when I export a UT_YUV420 file from Premiere.  I have also noticed that the UT_YUV422 option works very well for my dv2Film process.  The files it creates are smaller than Lags, use less resources during playback than Lags, and seems to have all the visual quality that I need for the job.
Now one might ask why use a YUV 4:2:2 or 4:2:0 'compromised' color space format instead of the full 4:4:4 of RGB?  There are a couple of reasons.  While theoretically 4:4:4 might seem the better option (and in some very professional cases it may well be), given the media I work with and export to, it's just plain overkill.  DV is a 4:1:1 format (NTSC) and MPEG2 on DVD is a 4:2:0 format.  While technically inferior, these color spaces are yet capable of producing some excellent results.  Witness DVCPRO HD and even AVC-I are both 4:2:2 color space codecs.  And AVC-I is comparable to Master Quality D-5!  So On the one hand, I figure if it's good enough for such professional codecs, it's good enough for me.  And indeed I cannot see any difference between a UT RGB export and a UT YUV export.  What I DO see is considerably smaller files in the YUV export, as well as faster encoding times to MPEG.  So, that is what I use.
To sum up:  Lagarith is an excellent codec for intermediary work.  But I feel UT is every bit as good vidually, and offers performance and file size improvements that make it hard to ignore.
Comments?  Opinions?  Questions?
*"I did a pretty extensive study of...video quality with Lagarith,  Huffyuv, and UT, and...UT-RGB and Lagarith-RGB provide 100%  identical output quality (confirmed with the MSU video quality  measurement tool)"

Answer 1

My notes:
1. Exporting at HD resolution from a Pr sequence using the UT codec is 3 - 4 times faster than exporting the same sequence using the Lagarith codec!
2. With 8 cores pegged in the AME, there was absolutely no difference in encoding speed between Lagarith SD source files and UT SD source files when encoding to MPEG2 DVD.  Both were very, very fast.
3. Decoding UT HD files using MT AviSynth crashes the host application, in my case, VirtualDub.  Conversely, decoding UT HD files using AviSynth with MT disabled works perfectly in VirtualDub.
4. Lagarith HD source files, in concert with hd2sd() and MT AviSynth, only export from VirtualDub twice as fast as UT HD source files with hd2sd() and AviSynth with MT disabled.
5. UT SD source files can be used with MT AviSynth without crashing VirtualDub.  It's only the HD UT files that cause a crash.
Summary:
Since the time saved by using UT on initial export from a Pr HD sequence overcomes the speed limits imposed by non-MT AviSynth as compared to MT AviSynth and Lagarith, and since all SD workflows with UT that I've tested are solid and faster than comparable Lagarith workflows, my new DI codec of choice is UT.
Thanks, Jim!
-Jeff