You should never do colourspace conversions. If you think that there is no real loss then you might find this interesting.
If I use HuffYUV in YUY2 mode for each coversion and did as follows:
- Ripped a dvd and made HuffYUV clips in virtualdub in Full Compression mode
- Edited in Premiere
- Prerendered a section in HuffYUV and put it back in the timeline
- Exported whole amv to HuffYUV
- Compressed the final encode in virtualdub using Full Recompress mode
that would all be "lossless" except for the last compression wouldn't it? No - it is VERY lossy. Look at this example:
http://www.absolutedestiny.org/yuy2rgbconversion.jpg
The first image here is the source, the second is the final image and the bottom image is the difference between the two.
You'll notice that in the 2nd image you can see that
a) Both top and bottom subtitles have faded
b) The background has become more like grey
c) The kanji now has a dark red shadow at the edges
and so on. The third image shows all the colour that has been lost and it is quite substantial really considering that you are hoping for it to be lossy. This is not an uncommon editing procedure for some people either but remember that just because you are using HuffYUV doesnt mean that everything is lossless as it will do a colourspace conversion every time it is given data as rgb.
Virtualdub Full Processing and Premiere work in RGB so everytime you use those you convert the colours. Also vfapi plugin uses RGB as well.
So, if you are using an RGB process - use HuffYUV's RGB mode wherever possible... it might be bigger but it will look better in the long run.