Online vs. Offline rendering in reaper

[John_C]

I did a little test using stock reaper plugins:
i recorded a short clip, then rendered it online, then 1x offline, then full speed offline.

When the phase was swapped (on the mixer channel) between any 2 of them, complete digital silence (as judged by the peak value thingy on the master bus not coming off - infinity)

i added heavy compression (with automatic gain compensation, and anti alaising), then heavy eq, then heavy limiting (reacomp, ratio infinite)
i did the same method of rendering and phase swapping and again:
DIGITAL SILENCE

I then recorded another short clip on a seperate track to the original (directly underneath, so they played at the same time), and added heavy eq, delay and reverb.

same again and got:

guess what.............DIGITAL FRICKIN' SILENCE

which shows that in reaper, using reacomp, reaeq, reaverbate and readelay and using up to 2 tracks
offline and online rendering makes precisely no difference

 

[TheDriller]

Good test.

i'll do one in Logic 9 this evening.

can you do the same test in reaper but with 3rd party plugins? preferably some that might induce latency, and see if they cancel to silence ?

and it would be awesome to get a few other hosts and plugins tested too.

 

[Torniojaws]

I've been under the impression that 1x rendering is just for hardware effect compatibility? Otherwise they should be 1:1 identical in the actual render, offline just being faster since it can use the full CPU power without limits.

 

[John_C]

well the thing is murphy was saying otherwise, so i thought i'd do a test that could confirm that it was identical.
However, the details of the pro-tools HD mix engine are completely unknown to me so it could well be completely different there

 

[TheDriller]

well the thing is murphy was saying otherwise, so i thought i'd do a test that could confirm that it was identical.
However, the details of the pro-tools HD mix engine are completely unknown to me so it could well be completely different there

ProTools can only render in real-time anyway, so it wouldn't make much difference.

 

[John_C]

ProTools can only render in real-time anyway, so it wouldn't make much difference.

True. I wonder what he was comparing then, coz he's def. a HD man

 

[dani]

as expected.

i think the difference comes from DSP based effects, which are sort of midway between digital and outboard. so, for users of UAD and HD.

thanks for doing this - i'm not going crazy! aha.

thanks,

 

[Soundlurker]

True. I wonder what he was comparing then, coz he's def. a HD man

He used DP

i bounced mixes (off-line) in DP for years... and always compared them to real-time (on-line) printing to a new track and exporting... the latter always won

I did a test some time ago and the result wasn't silence but rather just drum reverb. This happens with all plugins that affect the signal differently with time and those that are in some way random, ie reverb, flanger, chorus and so on. Still, I don't see a reason they would sound better in real-time. Of course, real-time is useful for a limited number of purposes but in most cases it is not necessary.

 

[Metaltastic]

SWEET, thanks a ton John, another myth debunked! (at least in Reaper and with no external-DSP enabled plugs, which is all I care about )

 

[TheDriller]

So here's my test for this myth :

- DAW = Logic 9

-I loaded up an old Ambient Dubstep project i had, it includes a good several soft synths on MIDI tracks and some audio tracks. sends to two different IR Reverbs, and some third party plugins like IK Multimedia T-Racks compressors and such. and of course, a load of stock Logic plugins.

-I bounced the same section both in realtime and offline, and then did a second realtime bounce for good measure.

-loaded up the bounced files into a new project, set faders to unity and first compared realtime vs. offline bounces, with one phase inverted.

-some elements canceled out completely such as the drums, IR reverbs and the Rhodes piano track. it just so happend that the first half of the section was just these instruments, so total digital silence.

-the synths however, didn't cancel. they were clearly different between the two bounces.

-however, when i checked the two realtime bounces, i found the same thing happening. the synths didn't cancel, so it is not the difference between offline and realtime, it is simply the behaviour of the synths.


-next i tried a different project. i loaded up a clean guitar recording and set up a send to an algorithmic reverb. i put some Logic Channel EQ and the T-Racks compressor on the guitar track for good measure.

-i coppied the guitar region so that it would play twice, with a gap before playing a second time. i then automated the reverb send so that there would be no reverb on the second repeat.

-i bounced in realtime twice and offline twice, and imported all 4 bounces into a new project.

-all 4 bounces (realtime1, realtime2, offline1, offline2) cancel perfectly with each other. i compared each file to every other file and all resulted in total digital silence.



so the conclusion is, that Realtime and Offline bounces are identical , even with reverbs (both IR and Algorithmic) and third party effects. the only catch is that some soft-synths and VIs will never bounce the same way twice, regardless of whether it is Realtime or Offline.

Myth debunked i think.

 

[Metaltastic]

Thanks a ton for doing this guys, nice to know I don't have to deal with real-time bouncing! (WHO HAS THAT KIND OF TIME )

 

[slashvanyoung]

I would have been VERY surprised if the outcome would have benn different...

I'll probably give it a go with Cubase too, later on.

 

[LydonB]

Do this test with outboard gear and I am willing to bet that the results are different.

 

[Metaltastic]

Of course, it wouldn't work at all if outboard gear was involved since it wouldn't get any signal But I doubt that was the point of this test, since I'd wager most on here are ITB guys...

 

[TheDriller]

Do this test with outboard gear and I am willing to bet that the results are different.

Well, of course it would.
because you can't do a render faster than realtime if there is outboard compressors involved.
same with reverb units.

or any outboard that affects the time domain.
so whats your point?



its like saying "sure that roast chicken tastes good, but cook it in the core of a volcano and im willing to bet that the results are different".

 

[Metaltastic]

its like saying "sure that roast chicken tastes good, but cook it in the core of a volcano and im willing to bet that the results are different".

Hahahaha, yeah, that's pretty much what I was getting at

 

[LydonB]

Of course, it wouldn't work at all if outboard gear was involved since it wouldn't get any signal But I doubt that was the point of this test, since I'd wager most on here are ITB guys...

And that is completely fine if everyone does things ITB.

its like saying "sure that roast chicken tastes good, but cook it in the core of a volcano and im willing to bet that the results are different".

 

[TheDriller]

care to present an actual counter-arguement?

 

[LydonB]

care to present an actual counter-arguement?

There isn't a need to.

 

[egan.]

Can someone do a Cubase test? That is actually the DAW that I have heard of the greatest issues between offline and real time renders.