bouldersound, post: 429367, member: 38959 wrote: I'm not asking for empirical evidence that it sounds better, I'm asking for empirical evidence that it's different at all, and which is objectively less degraded. Once that is provided I'll be more open to spending time listening.

i suggest you do a test for yourself to see if it's "different" or not. what i advocate (and others too) is not a simple transfer. it involves taking part of the process (summing) away from the computer (which btw a computer doesn't do very well) freeing up resources that can be better used to process tracks and plugs. Chris takes it further by dong additional processing on the second DAW at 2 track capture, while listening to the output (D/A) of the 2 track capture DAW. this is no different from "sweetening" audio on the 2-bus with eq /compression/ limiting, while listening off the p/b head of an analog 2 track machine you are mixing into. nothing new about that. we've been doing it for years.

Chris is now summing ITB, isn't he? So even he has decided analog summing is not worth it. I'm not saying what you're doing doesn't sound better (though I have my doubts), I'm saying I don't see it as being a net gain. I think the benefits of decoupled transfer and the negative effects of SRC are being grossly overstated. So I'm not going to do experiments that cut into my limited time in the studio without good reason in the form of evidence.

The reason we monitored off the play head was that tape had a sound and the results were somewhat unpredictable. That's why they called it "confidence" monitoring, because without it you could not be confident of the result. My setup has never given me reason to lack confidence.

Nicely put again Bos and Kurt.

bouldersound, post: 429360, member: 38959 wrote: I would have more confidence a good ME than I would in either you or myself. As I understand it, it's best to do all the processing with the most samples and the lowest noise floor, then convert it to the final format. That is the ordinary method. What you're suggesting is extraordinary, and that requires extraordinary evidence. "It sounds better to me" is not evidence no matter how many times or how loudly you say it.

This isn't a personal attack here. I'm pretty certain you are miss interpreting things, boulder. Basically what I am saying and believe, I don't trust an ME who bounces. You are far better off tracking at the destination SR and giving that to an ME. Learn how to produce the best mix you can before you give it to someone else to bounce! :notworthy:

bouldersound, post: 429335, member: 38959 wrote: By giving the mastering engineer files at a lower sample rate or word length than the project you are handicapping him

.
Assuming were are still discussing the bouncing process > No, you are handicapping yourself and passing it off to the ME.

bouldersound, post: 429369, member: 38959 wrote: Chris is now summing ITB, isn't he? So even he has decided analog summing is not worth it.

I haven't changed the way I prefer to sum and mix in years. Its always been using 2 uncoupled DAW's > mixing from Samplitude into Sequoia. I love the Neos and Dangerous Master between those two DAW's but to be honest, I can emulate all the mixing and mastering gear with Sequoia 13 loaded on the capture DAW now. So, its pointless keeping those products in the uncoupled step.

I still prefer 2 DAW's uncoupled and excellent converters. When I do my own work, I track at 96k and recapture it at 44.1.

audiokid, post: 429370, member: 1 wrote: You are far better off tracking at the destination SR and giving that to an ME.

Show me the evidence.

audiokid, post: 429371, member: 1 wrote: I haven't changed the way I prefer to sum and mix in years.

So you are summing OTB?

bouldersound, post: 429372, member: 38959 wrote: Show me the evidence.

You need to do that listening for yourself. If you can't hear its worth, then its most likely not going to change.

The OP asks, what SR I most commonly choose. I've said my value and explained why. Its not to convince you or anyone, or to "pick on you" either. I'm simply sharing a blurb on bouncing.

I track at the destination SR because its logical to me and it does sound great. I also use excellent converters and a good summing process that sounds accurate when I don't bounce. Could this be why I hear the value in that and you don't? I don't know.

My pov in this thread is based around... why bounce when you don't have too? (y)
Or do you ?

Based on my design and methods... It would be pointless to do that with my system. And it would be pointless to track at a higher SR and then pass it off to an ME to only down sample something I could do much better before hand. I am confident that I can track at the SR of choice and do a good job without bouncing.
To take it even further, if I want the better sound quality, I then choose to track at 96k and then "recapture it through an uncoupled process to once again, "avoid" down-sampling while taking advantage of the analog pass. :love:

I personally like how this not only sounds but I also like how easy it is to mix. Basically, its all to do with avoiding "bouncing" and emulating an analog mastering matrix that costs thousands.

This isn't a shootout or a personal attack to prove anything, its a preference explaining why I use 2 SR instead of one.

:)

bouldersound, post: 429373, member: 38959 wrote: So you are summing OTB?

I disable DAW1 master section completely. In a very basic way, I am changing the sum OTB. But, the sum is really done ITB when I group the stems, then assign them to either multitrack process or just keep them all in a 2 channel state. Summed ITB.

I prefer to use two converters that go through and analog pass. If I want some grit, I could put a bunch of hardware between the two converters, which is essentially what some hybrid ME do. (y) Its really the best of both worlds mastering like that anyway.

FYI, 32 channels or 2 channel captured this way sounds the same. Just as long as I am uncoupling between converters and going from example: 96k to 44.1, that is all that matters. Uncoupling is an inexpensive, and sonically excellent alternative to racks of hybrid gear. In fact, I even removed the best clocking system you can buy today. And here people making hits are still buying super clocks lol.
Note: If I wasn't going from one SR to the next, or using a two DAW step, I wouldn't go OTB in the first place. I don't Round Trip. I go OTB to avoid bouncing and a few other personal reasons which are beside the point here.

My summing process has never changed in years. I still prefer mixing from one DAW into another with an analog pass (one converter to the next) in the middle. I find I don't need a summing box or a console with Sequoia 13 on the Capture DAW now. So I have sold most of my high end mixing and mastering gear now. I've saved thousands this way and the results are better. I think its good to track at 96k and recapture that apposed to bouncing down or simply put, restricting myself to 44.1 multitrack recording when I can do better... I don't use a lot of plug-ins when multitracking like this and I think a 96k SR with good converters sounds better summed and captured to 44.1.

Cheers!

Boswell, post: 429366, member: 29034 wrote: I've told the story several times of a contract audio interface design I spent the best part of a thousand hours designing, testing and getting it to sound the way I wanted it, only for the bean-counters at the production end of things to substitute cheaper components, circuit board and power supplies. As a result, the commercial versions sounded nothing like what I had painstakingly designed, but the irony was that about the only thing they did not change was the A-D converter chip. I disowned the final product and did not allow my name to be associated with it. It really went to show that it's not just the A-D chip but everything in the design that contributes to the sound.

I would support you on this if I could and I don't think I'm alone. Is this something you could make like an example: Fred Forssell does, Bos? A high quality boutique converters you make?

What I'm doing sounds more than good enough and it's faster. I don't have time in the studio to play around so I will need some sort of actual evidence that there is any difference before experimenting. Short of that the best explanation for the difference you're hearing is confirmation bias.

To me it's like someone on the internet telling me that steering wheels are bunk and I should install steering brakes like the ones in dune buggies. Sorry, I'm not going to re-engineer my car to try it out, I'll let Road & Track or Consumer Reports try it first.

bouldersound, post: 429377, member: 38959 wrote: What I'm doing sounds more than good enough and it's faster. I don't have time in the studio to play around so I will need some sort of actual evidence that there is any difference before experimenting. Short of that the best explanation for the difference you're hearing is confirmation bias.

To me it's like someone on the internet telling me that steering wheels are bunk and I should install steering brakes like the ones in dune buggies. Sorry, I'm not going to re-engineer my car to try it out, I'll let Road & Track or Consumer Reports try it first.

So what you are saying is, you prefer to bounce then ? ;)

audiokid, post: 429378, member: 1 wrote: So what you are saying is, you prefer to bounce then? ;)

It beats steering with the brakes.

Just to clarify, this is all in fun to me.

bouldersound, post: 429379, member: 38959 wrote: It beats steering with the brakes.

:confused:

I'm not seeing the comparison when this is mostly about bouncing but I do appreciate the disbelief. I'm pretty certain however, you aren't following this process so I'll keep going in fun until one of us calls it quits lol.

bouldersound, post: 429377, member: 38959 wrote: What I'm doing sounds more than good enough and it's faster

• "Good enough" is valid but subjective.
• Faster, I doubt it. It takes many times longer to finish when you bounce apposed to retracking in the same pass as a mix. While mixing (or mastering for that matter), you are tracking the capture at the same time its being mixed.

bouldersound, post: 429369, member: 38959 wrote: I'm saying I don't see it as being a net gain. I think the benefits of decoupled transfer and the negative effects of SRC are being grossly overstated. So I'm not going to do experiments that cut into my limited time in the studio without good reason in the form of evidence.

The reason we monitored off the play head was that tape had a sound and the results were somewhat unpredictable. That's why they called it "confidence" monitoring, because without it you could not be confident of the result. My setup has never given me reason to lack confidence.

• You are also monitoring the pass on the DA side of the destination SR. How is bouncing faster or monitoring more proficient than this? :sleep: This is exactly how the Pro's in mastering monitor. You should know this.

And bouncing never sounds as good but thats because I know what bouncing sounds like. Its always smeared in comparison to direct tracking.

Chris, when you mix. Do you mix monitoring through Daw 1... Then when you mixdown you monitor through Daw 2? or do you always have both going?

Chris Perra, post: 429382, member: 48232 wrote: Chris, when you mix. Do you mix monitoring through Daw 1... Then when you mixdown you monitor through Daw 2? or do you always have both going?

both selectable.

However, when tracking , especially during overdubbing, I am monitoring on DAW1. But, I make all my decision based on what I hear on DAW2. I track at 96k but monitor and base the entire mix through DAW2 DA. Its so fast and simple and from a "Pro Audio workflow, "very very affordable". Meaning, you need very little of both hardware and bloat software . At least thats how I hear it which may also be due to the DAW's I choose.

Okay, so I did come up with an experiment that didn't require going to the studio or reconfiguring any hardware. This addresses the claim that streaming the audio is different from the rendered (bounced) file.

I opened a basic mastering project with a single song. I moved the master bus processing to the channel. Then I rendered the audio to a file (10 seconds for a 2:55 song) with the same word length (bit "depth") and sample rate. Then I imported the file, inverted the polarity and played the track through. The peak reading on the master bus read -Inf.

There is in fact no difference between the stream and the file on my system (Sony Vegas 6). Therefore there is no loss of quality rendering vs. real time playback.

I am curious to know how other DAWs behave when this experiment is done.

Of course this doesn't address your whole process, but it will narrow it down.

Great comeback. I'm impressed. :love:(y):cool:

My first thoughts are,
I think your DAW lacks. or better put, turns everything into the same thing no matter what you are doing. Which is what Ethan was getting at and why I don't trust those shootouts nor do I use the DAW's either of you use. :)

I did a null test between a few converters through various processes a few years back. Other than Remy's wasted attempt to derail it into broadcasting back in the 70's... it didn't get any attention here but it was essentially what Ethan does with his conversion testing approach and also falls short in this converter "importance" debate.

What I found is two things. Not all DAWs are worthy and after I did a few ADDA through any converter, high end or mid level, the mixes all ended up sounding the same. They nulled close enough that hybrid mixing and $10,000 converters with super clocks started looking like a complete joke. This is when I decided to avoid round trip processing completely and to avoid returning back through the same converters clocked to the same session I was mixing from.

If you can read between the lines why I then adapted the two DAW approach and why I avoid bouncing down, maybe you will find an answer. Why does Bos hear it like me and why do many successful mastering engineers capture audio onto DSD first.
Don't ask me to explain it all more than that. Its all so subjective but I do know one thing. Mastering gurus use Samplitude for a reason.

This is where it all becomes subjective. I do what I do because I hear differences that direct me towards what I do. If you don't hear any changes, then its obviously because your don't hear the changes. Can you explain that? The best I come up with is: I am here and you are there. Maybe Bos will explain some insight into your finding boulder.

I'm working towards using less of everything, listening in better ways so I actually hear cause and effect better and worrying only about my needs lately. Spending a lot of time and money in this business is all becoming a past life to me. .

audiokid, post: 429391, member: 1 wrote: Great comeback. I'm impressed. :love:(y):cool:

My first thoughts are,
I think your DAW lacks. or better put, turns everything into the same thing no matter what you are doing. Which is what Ethan was getting at and why I don't trust those shootouts nor do I use the DAW's either of you use. :)

Maybe my DAW sucks, maybe not. That's irrelevant. This isn't a shootout between our systems, it's a test of your premise that bouncing is different from the stream. Replicate my experiment on your system and tell us if there is a difference between the stream and the bounce. Actually, it's something everybody should know about their DAW.

I have been in the room at Airshow for at least one mastering session that used hardware processing. I can't say for sure but I think they were capturing to the same PT session as the playback.

[Edit] Of course that was probably their "C" room, but still, it's Airshow.

How did it sound, you ask? It came out as overbaked as it went in.

bouldersound, post: 429392, member: 38959 wrote: This isn't a shootout between our systems, it's a test of your premise that bouncing is different from the stream.

Premise! Are you kidding! My process is without doubt different from your stream. You are completely misunderstanding the analog uncoupling process in comparison to yours. There is nothing you are doing that can remotely compare to what I am achieving. Nor would a bounce ever null with a capture in my process. Not even close. Its impossible. Get that understood and I will get some rest . :D
Being said, we are simply trying to understand if "your DAW is in fact even accurate at this point. As a group we are more likely to be wondering why your bounce is exactly the same as any stream for that matter. I can't even believe that is possible. Something is whacked.
Bos to the rescue!

="bouldersound"Of course this doesn't address your whole process, but it will narrow it down"

Indeed, I'm glad you are acknowledging that. Thanks for your participating in this.!!

bouldersound, post: 429392, member: 38959 wrote: Replicate my experiment on your system and tell us if there is a difference between the stream and the bounce. Actually, it's something everybody should know about their DAW.

Again, you would need my system and DAW's to learn this is impossible. But just from a scientific understanding of how there is change between an analog pass ( not one is ever the same), its impossible that a 44.1 bounce would ever null to a 96k to 44.1 analog pass.

Bouncing like you are doing, I don't know. Its not something I've ever thought about but it is a good question.

audiokid, post: 429393, member: 1 wrote: Premise! Are you kidding! My process is without doubt different from your stream.

To quote the animated version of Saddam Hussein, relax, guy! I was talking about comparing your audio stream to your bounced file.

At this point I'm only trying to address one question: Is the bounce different from the stream? I answered that to my satisfaction for Vegas 6 and I kind of don't care if anyone believes me. But I am curious about other DAWs. Do they behave like mine in this respect? Will anyone give it a try?

I also need to repeat my experiment with a reverb. All I had on my track was a mastering limiter. It's possible a mix with time based effects won't null, but I'm betting it will. And of course I should do it with a bigger project.

Of course a capture won't quite null. That's a definite difference that could be good or bad. If it's good then it's good.

bouldersound, post: 429395, member: 38959 wrote: To quote the animated version of Saddam Hussein, relax, guy! I was talking about comparing your audio stream to your bounced file.

I know and I'm also very relaxed. I'm poking fun at you all through this. I found your comment, Premise, formal, yet sort of blind which made it fun to exclaim back at you. Sorry if you don't get my rash humor, pal!;)

I hope others participate.

Cheers!

I'm kinda kickin' myself in the you-know-what over this SR thing. I really should have been tracking this current project at 96 from the get-go. There's not much I can do about it now.

But I definitely have a case of the "shoulda-coulda-woulda's".

Chris, So in Daw 2 do you run mastering plug ins on the input. or do you run mastering plugs on the output of Daw 1?

How would approach mastering an entire album? Do you wind up taking the mixdowns from Daw 1 that are captured in Daw 2 44.1/16 bit stereo and then reassemble them back in Daw 1 with correct levels/fades/tonal settings song order etc, and then recapture in Daw 2 for final renders?.

Or seeing as you capture in Seqouia do you take Daw 44.1/16 bit mixdowns and do a standard master effects process internally rendering the finals in Seqouia?

Chris Perra, post: 429407, member: 48232 wrote: Chris, So in Daw 2 do you run mastering plug ins on the input. or do you run mastering plugs on the output of Daw 1?

Both, but this is a great question. (time providing) I plan an taking time to share this section so I'll keep you in mind on that.
The skinny is, one of the greatest assets to DAW2 allows you to capture endless mixes of the same sessions and compare. One might ask, why do I need to compare so much? This workflow is a self teaching dream.
To answer this question,
I will use plug-ins on the channels and on the master section.
The master section usually has the same settings throughout a session that I don't change. These are the stock Sequoia spacial plugs, EQ, limiter, and a master reverb. I keep these settings the same for most of my mixes as the are fool proof and the emulation of my high end analog mixing and mastering gear. I bought as much hardware as I could over the last decade, studied what it was good for, then emulated it all around with Sequoia. So, it is a mirror of what my (now sold off :) analog mastering matrix was. More on that later.

The channels usually stay flat but I may succumb using a comp or minor eq tweaks from track comparison to the next. I do this to learn how a comp or eq change effects the final mix. I use this to learn how my DAW 1 mixing effects the 2-bus and capture.
For educational purposes, I may end up with 20 versions of a mix on DAW2 and will compare all the changes I made while mixing. Its astonishing to refer back, (min, days, hours etc) to hear what you did over a course of a session or year to learn what you do. In the end, this entire way to capture and compare is very powerful.

Chris Perra, post: 429407, member: 48232 wrote: How would approach mastering an entire album? Do you wind up taking the mixdowns from Daw 1 that are captured in Daw 2 44.1/16 bit stereo and then reassemble them back in Daw 1 with correct levels/fades/tonal settings song order etc, and then recapture in Daw 2 for final renders?.

Not usually. But its also very easy to transfer back and forth between both DAW's as they are connected in a home sharing network. The times I will transfer captures back into DAW1 would be more when I am sound designing or building a library that I capture first, master it and then use it in a mix.
The standard procedure, DAW 2 is for all 2 channel mixdowns that can be mastered or simply prepared to be mastered by someone else. I capture all the audio at around -10 so if I choose to master the best track, I simply enable the one I want and finish it. DAw 2 is basically DAW 1's mastering section with a greater way to process and monitor. Its a very deep topic that I'd love to keep discussing later.

Chris Perra, post: 429407, member: 48232 wrote: Or seeing as you capture in Seqouia do you take Daw 44.1/16 bit mixdowns and do a standard master effects process internally rendering the finals in Seqouia?

yes. all the above plus...
DAW2 is a libray of mixdowns that can be mastered or prepared for mastering. I also use it to capture audio from the web for quick references which is another great topic.

Did this help answer your questions?

Ok so it's more about the Mixdown being a physical playback than worrying about rendering/exporting once all tracks are summed to stereo. Or do you prefer to do a physical playback with the mastering plugs as well?

Chris Perra, post: 429448, member: 48232 wrote: Ok so it's more about the Mixdown being a physical playback than worrying about rendering/exporting once all tracks are summed to stereo.

Partly.
I don't bounce which is a big reason and I also find mixes sound more accurate when I am monitoring the mixdown on DAW2.

Chris Perra, post: 429448, member: 48232 wrote: Or do you prefer to do a physical playback with the mastering plugs as well

Yes, and because I find the mixdowns sound better via this process, I use less plugs, need no hardware and it simply sounds better.

In a nut shell but not necessarily in this order;

I love the improved sound that happens between the analog pass of two uncoupled daws. This is something that cannot be achieved any other way other than uncoupling;

I prefer mixing into a master or capture;

I love being able to compare mixes in a separate Daw;

I like how much faster it is to finish projects that are in higher sample rates;

I love how cost effective it is in comparison to other types of more expensive hybrid mixing and summing workflows. To my ears, and to my clients , The sound quality is better when you use less gear. I've done blind tests and less is more, always wins;

Mastering is a simple process, especially when the mix is done well so this set up actually helps my mixing, thus less heavy handed mastering is needed.. ITB always sounds better for mastering. The uncoupled pass is really all that is needed when it comes down to what analog adds.

Basically all i do is use a limiter, reverb/ spacial m/s effects and either subtle eq and very minor compression.
In a master , digital tools win every-time so its the obvious step to master or prepare your master on the capture DAW.
I use very basic plugs but that's me. Some people or some songs require a more savage compression that pumps to the beat which still rocks on the capture DAW.

The inetersting area of this is will higher sample/bit rates and more powerful CPU processing one day allow the summing in digital to not have audible bottlenecking.

Chris, I break your process down very simply. It's identical to having a mix down deck in tape. But the processes most important objective I belive, is simply in the bypassing of the master channel, and its associated effects/artifacts.

I'd be curious myself one day to simply add new recorded tracks one at a time, and do a typical bounce to stereo, and listen for the point where it bottlenecks. Perhaps it's not digital summing, just it has a limited range. There is certainly a threshold for plugins, where that one more puts it over the edge, into flatland.

The other thing to consider when hearing those differences, is the gain staging and playback of the file. It would have to be a daw that ran multiple instances or projects at once, so you could easily a/b the multi track and bounced versions.

I'm hear this stuff for myself, and it's frustrating. I belive the two daw method is accomplishing something. I'm interested in exactly what problems it is fixing, with hopes that software designers focus more on that type of thing.

I've been recording for a long time at 48K, or at least thought I had until I discovered cubase 8 has been recording at 96, and I didn't even notice until last week. I only realised on a big multi track, multi effects project when it glitches, and I was sure I'd had this quantity of tracks happily working before!

DonnyThompson, post: 429276, member: 46114 wrote: Which sampling rate do you most commonly use when recording?

Please don't include mix projects which come to you where the SR is set by the client's project/files...

I'm talking about when you begin recording a new project.

along with your vote, comments -like bit resolution choices - are also more than welcome.

;)

I've been using 88.2kHz at 24bit the best choice for a couple of years, soundwise and daw wise! It has the perfect sound and doesn't puch our PC too much!

I have not ventured out of 44/16. I record in 44/16, mix to 44/16. I've read all of the reasons for higher, and I am sure they are valid, but for an HR'er, I think going higher at my level yield very little.

For those in the pro studios, with better ears, better equipment, better rooms, I can fully understand the higher rates.

DM60, post: 430606, member: 49090 wrote: I have not ventured out of 44/16. I record in 44/16, mix to 44/16. I've read all of the reasons for higher, and I am sure they are valid, but for an HR'er, I think going higher at my level yield very little.

For those in the pro studios, with better ears, better equipment, better rooms, I can fully understand the higher rates.

If you are happy like that, it's fine.. The main reason to at least go to 24bit is the headroom and noise level ratio..
I'm at 24/96 since about a year and I'm loving it !! ;)

For the important projects, it's 24-bit 96KHz for me. As I have posted previously, at that rate, the top-octave phase issues due to anti-aliaising filters plus things like digitally-generated EQ and effects largely fall outside the audio band. To avoid digital sample-rate conversion (SRC), I replay the tracks in analog at the original sampling rate, mix on an analog desk and then digitize the 2-track mix at the target rate (44.1KHz for CD, 48KHz for video). Only the two tracks of anti-aliasing filtering are in operation here.

For demos and other non-pro work, I mostly track at 44.1KHz to avoid SRC when burning the target CDs.

I am sure it is better, not sure how much of that can actually be heard.

As stated, if someone is working with higher grade equipment, that extra might be noticeable. I don't think I have the ears to appreciate the difference and really doubt anyone listening to my stuff would either. (Assuming someone besides me is listening)

I use the highest sample rate I can, depending on how many tracks and the computer ect. My argument is that the higher sample rates won't sound audibly worse (in general) and with sooo many different formats for delivery I want the source tracks to be as 'pure' as possible. Also, for future compatibility. That said it's usually 24/44 or 24/96. I've always found it dumb that movies use higher fidelity audio than records, but the 'mixdown' daw system , that I'm setting up for my new setup, will avoid, any odd numbers in the division's by avoiding the src internally.

Two DAWs avoid all the bounce SRC issues for me. If I wasn't using two DAW's, however, I'd be tracking at the destination SR to avoid bouncing. I'd also be investing in the best AD I could afford. The two area's go hand in hand. Better converters sound better at lower SR to me.

audiokid, post: 430617, member: 1 wrote: Two DAWs avoid all the bounce SRC issues for me. If I wasn't using two DAW's, however, I'd be tracking at the destination SR to avoid bouncing. I'd also be investing in the best AD I could afford. The two area's go hand in hand. Better converters sound better at lower SR to me.

The destination SR is what I was getting at. Do you do anything different for say a tracks iTunes version versus the album? I don't know that I've even heard .wav music much in the past few years via all the internet radio I listen too. Not being a jerk chris, rather, I know how tuned in you are to phases issues, and am curious how you deal w the myriad of destination formats? Particularly because you use the mixdown daw method.

DM60, post: 430612, member: 49090 wrote: I am sure it is better, not sure how much of that can actually be heard.

As stated, if someone is working with higher grade equipment, that extra might be noticeable. I don't think I have the ears to appreciate the difference and really doubt anyone listening to my stuff would either. (Assuming someone besides me is listening)

I assure you, record 24 tracks at 16bit and 24 tracks at 24bit, you will hear the difference.. At least I do, specially in the silence, you get more noise with 16bit.
Other than that, 44 vs 96 was a big thing when all plugins were following the resolution of the DAW project. But many are doing oversampling now so, even at 16, many are processed at 32bit. But for me the I don't take any chances.. I want all the processing to be done at higher resolution when mixing and mastering.. and then I convert to lower quality.

48/24 simply because I want to. I can go to 192 if needed but I don't really do anything that would warrant this. And I NEVER 'bounce' unless its a quick demo of a part of the project for the client to take home and decide on something. There's aliasing that takes place when you render inside the same machine that will affect SOMETHING in the process. Is it going to effect the project to a point of default? Maybe not. Maybe. Can I prove this other than my ears....Maybe not. Maybe......

I think that the way each manufacturer deals with the code necessary for this, makes a difference in comparison between programs. My process is much like Chris' only I'm not the mastering engineer. So my projects all get to the mix stage at which time everything becomes 'groups' each separated into their own bus and each with their own VCA master into a master bus also with its own VCA master. This gets assigned to a PRINT BUS when its all balanced and salt and peppered and the entire mix is then recorded to this new two track which then gets exported to whatever media the mastering engineer wants WITHOUT any conversion.

So, like Chris, all SR conversion happens at the mastering level from an unprocessed source. I guess "unprocessed" is a little deceiving...the mix is processed as a mix, just no internal SR conversion....

There is a BIG difference. Audible.

Dave what are you using for your 2ch converters on the mix down daw?