Not really. If you want to dither it, do it at the 24-bit level if you want a 24-bit result, but really I would not bother about the dither at that level. Going from 24-bit to 16-bit is a very different matter, so the very last stage of the mastering process should be to take it down to 16-bit with dither.

Awesome. I knew 24 to 16 was important, just wasn't sure about 64 to 24.

Thanks for clarifying.

I know some people that up sample the tracks to 192k while mixing as they believe the track reacts better to the plugins added. What I am actually saying is that at 192k sample rate they are claiming that the tracks will not smear as badly from all the plugins added during the mixing process. Question: Is this true or rather does it make any sense at all? My question has less to do with bit depth, except do plugins work better on the mastering process at 64 bit too? Just picking the idea to see what responses we get here...

Since more and more plugins are now doing oversampling, recording at 88.2 96k or 192k seems less important. Don't forget that 32tracks at 192k will be a huge amount of hdd space and way harder to process for your computer at the mix time. I see it right away when I need to ajust for a longer latency when processing 88.2 projects vs 44.1.
On the other hand, if one would do OTB mixing, I'd go for Higher rate since the computer does only read and write work.

One thing I noticed recently... I was mixing in 48k sample rate as my mix files were all configured that way. I dumped the mixdown file as 44.1k and of course there is a bit of high end that is lost immediately. Personally I feel that sticking with 44.1k is the way to go and the idea that it is taxing my computer less is even more incentive to mix files in this manner. It seems to me that when you convert a file from 88.2 or whatever back down to 44.1 it is gonna smear the image and lose some high end. So why do people continue to think that up sampling a 44.1k to 88.2k is gonna give them less smear? I really am curious about this and maybe I have hijacked this thread... Oh well... I think if someone was concerned about dither they would want to talk about sample rate too as that is very relevant indeed.

PS was not saying 88.2k I meant that they are up sampling all the way up to 192k!! So is that even a good idea if the main files are recorded at 44.1k and 24 bit??

Double PS: I really don't think up sampling increases anything at all except noise that did not exist before. Seriously, going from 44.1k to 88.2k is one thing but all the way up 192k makes no sense and no good can come from up sampling. I mean if you sampled the original recording at 88.2k then yes maybe, but up sampling 44.1k to 192k is not recording a sample. Rather it is destroying a 44.1k sample in an attempt to claim missing information that will Magically emerge just by a computer process... This make no sense at all! Nothing about up sampling makes any sense to me at all and in fact down sampling is hard enough, so why make it harder?

The majority of audio systems are not capable of reproducing the differences between double speed and quad speed so I wouldn't bother. If you want the most realistic and accurate audio recording and reproduction, that is 1 bit DSD. The problem there is editing the tracks.