Direct Offline Processing Sounds Different to Inserts

Does anyone know if direct offline processing applies more than just the plugins & processes to the direct audio clip as I am getting a very different sound once using direct offline processing vs. using inserts.

For example, if I duplicate a DI guitar track (with exact same routing, and settings) and apply a guitar amp sim using direct offline processing, and then have that same amp sim with the same settings on the duplicate track just as an insert, they actually different, very different.

The direct offline version is mushy and loses a lot of clarity.

It should sound exactly the same so I’m wondering if something else is being applied in the render, maybe the ‘pre’ section settings?

Is anyone else getting this, its driving me mad and makes direct offline processing unusable as it destroys the sound. I’ll have to go back to render in place otherwise.

Hi,

Could you attach some files, please? Maybe some screenshots of the settings, or a snipped of the project? Or an Audio with the different results?

Place it maybe on Dropbox (or similar) and share the link here.

The project would be the best.

Yes, I will do this over the next couple of days, but one question that might be the answer is does direct offline processing also include plugins through the whole signal chain, and does it include the channel pan and volume (as well as pre, does it take into account pre-gain?)

I would assume it only applies the chosen effects and processes to the original event waveform as chosen in the direct offline process box, and not take into account other signal path elements outside but I’m not sure.

If it doesn’t work this way it could be that it also applied group and master effects further down the chain, so then the processed version would be running through those plugins twice (like what you need to be careful with for render in place as it does pick up inserts across the whole chain if you don’t turn them from before you render)

Hi,

Your assumption is correct. The effect is applied to the Audio event only. Pan, Inserts, Sends, etc is processed afterwords.

If you are using a mono audio files on a stereo track, that could be the difference as well. DOP on a mono file will be a mono effect even if it’s on a stereo track.

I’m definitely just using mono audio tracks and mono plugins.

If DOP only applies to the audio (no additional processing) I do have a problem. I’ll make some files

Yes, DOP processes the ‘naked’ event. RIP has options: include channel fx, signal chain, signal chain + master).

Sounds like RIP with included channel inserts could be what you’re aiming for. Does it? Sadly there are no further options yet (rendering without fader/pan, maintaining the file format are what I wish for).

So, on a mono track you’d have to set the fader to unity and 32 bit float as format (just to be safe from clipping at unity). That’s what I do…

Thanks for the feedback but I am getting differences with both RIP and DOP unfortunately and I’m not sure why.

Here is a link to a file I made on some cymbals for comparison using RIP - https://1drv.ms/u/s!Anv8bLn34focgvF8VfpqOzn_IATk2A

You can hear the first 20 sec is the original track where I have 2 inserts (i) Studer A800 taper saturator, (ii) EQ with low pass filter @ 35Hz. I then RIP with just the pre and inserts activated, but all sends, group effects, master effects (everything else in the chain disabled). You can hear from 23 sec onwards the rendered version. It is significantly thinner in the mid to low range and I don’t know why.

I thought it was just DOP but seems like I’m getting differences with RIP too.

All routing on the original and RIP channels are exactly the same (see link for screenshot - https://1drv.ms/u/s!Anv8bLn34focgvF9xX1D593pTpbF9w but they sound different. I’ve triple checked the channels paths multiple times and I can’t see anything that would scoop out more mids?

Indeed the difference in the soundfiles is more than obvious.

All I can say is that DOP/RIP do both produce expected results here, but that isn’t very helpful I guess. There must be something causing this… did you test it with a single simple file in an empty project already?

have you tried saving the insert preset and loading it into the DOP presets?

make sure you aren’t accidentally double stacking the same processing twice somehow, like rendering the master output.

it could potentially also have to do with how you have things panned, and what your pan law is set as… potentially.

I confirm, that Direct Offline Processing (DOP) processes audio differently from Inserts, Render in Place (RiP) and Freeze, when the plugins have exactly the same settings. Inserts, RiP and Freeze produce identical results between themselves, but DOP differs from them.

Initially I thought it was just some sort of a delay compensation issue, but, after doing sample accurate alignments, the waveforms always differ between RiP and DOP.

I will do some blind test comparisons regarding, how noticeable the sonic differences are, but for now it looks like I am going to stay away from the DOP simply because the Inserts and DOP don’t produce exactly the same results and the Inserts are more useful. My guess is, that there will be meaningful perceptible differences with layered sounds, where phase between layers tends to be critical. Who knows, maybe DOP sounds better for layering… maybe not…


Just in case, here’s a quick description of the rough test I did:

  1. Use one audio file - duplicate the track to have an identical source audio for the DOP track. Basically, the 1st track is for Insert processing and the 2nd is for DOP. Route both to the same bus.
  2. Insert a plugin (or several plugins) on the 1st track and drag the same plugin(s) to the DOP window on the 2nd track, so the plugin settings would be identical. Try to use plugins, that don’t have random/non-linear processing components (usually emulations add some noise and other non-linearities, that hardware has).
  3. RiP (channel settings only) the 1st track to create the 3rd (rendered) track. Invert polarity (phase) on the rendered track (3rd) to compare it to the Inserts (1st track).
    Result - signals cancel out, because they are identical, so all is good and life is beautiful.
  4. Compare the 2nd (DOP) track to the 3rd (rendered) track - nothing cancels out. If necessary, align RiP and DOP tracks by adding the amount of samples, that certain plugins require for processing (can be checked in the Channel Latency Overview).
    Result - waveforms are different. Hmmm… Interesting… Fiddle around some more with the alignment… Hmmm… Interesting… Try other plugins… Hmmm… Interesting… Fiddle some more… Hmmm… Not interesting anymore…
  5. Derive conclusions.

I played some more with DOP, after I wrote this:

What happens is that DOP processes the audio files before anything else, which includes the time stretching (Musical Mode). Time stretching happens after DOP, but before the Inserts - that’s where the difference comes from. The moment you turn off the Musical Mode, polarity (phase) cancellation tests work fine, so I take my words back - DOP processes plugins the same way as the Inserts do, only before the time stretching happens.

So, if you use audio loops (or any other audio files with Musical Mode enabled), that stretch with your tempo changes, be aware, that there will be differences between DOP and Insert results due to this sort of “intermediate” layer of processing, that Musical Mode introduces.
Also clip fades, crossfades and gain happen after DOP, but before the Inserts. Same goes for AudioWarp and VariAudio, but I don’t use those very often, so don’t take my word for it. I think VariAudio actually gave you some sort of warning, when you applied DOP after you did some pitch corrections.

Unfortunately, this thing with Musical Mode time stretching between DOP and Inserts is not very intuitive and doesn’t seem to be pointed out anywhere. But now we know. Hopefully, future manuals will include this info.

I hope this helps.

1 Like

Makes sense as the time stretching algorithm will behave differently depending on how certain frequencies and such are being stretched.

Think of DOP as being a printed audio file, because it essentially is… so yes, it does come before everything.