Parallel Processing

Support & Discussion for Modular Mixer v2

Moderators: valis, spacef

Post Reply
scary808
Posts: 449
Joined: Mon Apr 02, 2001 4:00 pm
Location: Utah

Parallel Processing

Post by scary808 »

Sorry, I just noticed there is a thread specific to Mod Mixer V2. Just reposted my question here...

I see mention of parallel processing with the Mod Mixer V2. Is there a way to time align dry and wet signals or does it happen automatically? I do not own it yet, just researching options for when I get my system up and running again. It looks really wonderful. Are there any videos with an explanation of the architecture?
User avatar
spacef
Posts: 3234
Joined: Sun Jun 17, 2001 4:00 pm
Contact:

Re: Parallel Processing

Post by spacef »

Hi !

no there is no phase align function internally, at any level.

For the inserts, there are many different rountings of parallel, crossed , summed etc. it is made for special effects. in general there is no phase in the routing options themselves. for insert effects, it is more for stereo chorus/phaser/flangers/delays etc. you can cross channels and use modulated filters too.

For precision applications, such as eq/comps or else, you probably need to align in daw if you find it not in phase. you can record a dual mono or stereo signal through the individual outputs of each channel.

Someone corrects me if I am wrong but there is no "relative" delay compensation for insert effects in Scope (ie, Fx relative to other channels or other effects).


I will try to look for possible manuals or pictures of projects etc, later today.


I hope it helps.
plug-ins for scope
SpaceF website
SC website
User avatar
spacef
Posts: 3234
Joined: Sun Jun 17, 2001 4:00 pm
Contact:

Re: Parallel Processing

Post by spacef »

Hi,
I added the "quick start guide" to the downloads :
-->> https://spacef-devices.com/modularmixerv2-downloads/
plug-ins for scope
SpaceF website
SC website
scary808
Posts: 449
Joined: Mon Apr 02, 2001 4:00 pm
Location: Utah

Re: Parallel Processing

Post by scary808 »

spacef wrote: Sat Jul 09, 2022 12:03 pm Hi,
I added the "quick start guide" to the downloads :
-->> https://spacef-devices.com/modularmixerv2-downloads/
Thank you for uploading the manual Mehdi! Could the “Phase Fix” device be inserted in the “Dry” signal line be used to correct the phase issues?

Regardless of whether it is phase coherent with parallel dynamics processing, the effect routing you have implemented is a brilliant idea!
User avatar
Peter Drake
Posts: 214
Joined: Thu Nov 08, 2018 12:27 pm
Location: Outer Milky Way

Re: Parallel Processing

Post by Peter Drake »

Stepping in here to say yes, with some provisos. I'm an enthusiastic Modular Mixer user, and also very picky and too obsessive about phase issues. I seldom find that parallel processing in MM introduces any phase issues that are a problem. It also seems that somehow, most of the time, the insert effect routing in MM seems to minimize phase effects. I found this out when trying to duplicate a MM mixer setup in some other Scope mixers. This is not to say any other Scope mixer will have more or different phase problems than MM, but I've found some unusual signal and effect routings to be less phase trouble on MM. Maybe it's the way processing gets allocated together onto one DSP chip or something else about Mehdi's underlying design.

Here's the major precaution: once the mix has a setup that satisfies your phase needs PRINT/BOUNCE YOUR TRACKS OR STEMS. Do not assume that everything will work exactly the same way once you close and reopen the project. Once again, I think this may get into the mysteries of how Scope handles allocating resources at the DSP level. Also, phase/sample delay seems to be the only kind of inconsistency that comes up, at least for me.
I learned this lesson on a massive multitrack fixing/sweetening project that I probably should have broken into smaller submixes, but ended up re-doing a lot of compensating sample delays.

Another interesting note, if you use modular for some signal processing (and you should) you'll get more consistent phase/timing if you use the modular insert device that Red_MuZe made. (Should be easy to find by searching "modular insert" in the devices forum) Again, this seems to be more predictable and consistent than routing through a modular device in the routing window. I have no idea why, and I don't care, I'm just happy when stuff works.

As you can tell, you've hit on a mixing issue I'm very sensitive to. I'm that guy in the mix session that HAS to find where that flange/flam/ring/thinning/murk is coming from. I've also been a Scope user for over 20 years. Hey, fellow old-timers, remember Big Mixer? Great surface design, massive phase problems.

Keep in mind also, that a lot of the "magic" is in completely messing up phase coherence. Some of the sweetest and most musical EQs are a total phase disaster. I sometimes have to stop myself from "fixing" something that is better with the flaws.

Another tip: it's often faster to use another copy of a device that isn't doing anything to do phase/delay compensation. For instance, if I parallel process a track using a low pass filter I'll run the same filter set to do no filtering on the dry signal path. Much faster and easier than moving stuff with Phase Fix or nudging in a DAW later. And! Keep in mind that for many devices the latency/delay between doing nothing and bypassed is very different, so be careful.

Also, also, mid-side mixing techniques can save a lot of time and trouble with phase. I find I can get away with a lot of adjustments solely on the mid channel and it doesn't cause the "air" or "space" on the side channel to get weird or collapse or peculiar or just get less clear.

I hope this helps, thanks for reading. You will find modular mixer has amazing power and possibilities for you.
User avatar
astroman
Posts: 8406
Joined: Fri Feb 08, 2002 4:00 pm
Location: Germany

Re: Parallel Processing

Post by astroman »

Here's the major precaution: once the mix has a setup that satisfies your phase needs PRINT/BOUNCE YOUR TRACKS OR STEMS. Do not assume that everything will work exactly the same way once you close and reopen the project. Once again, I think this may get into the mysteries of how Scope handles allocating resources at the DSP level. Also, phase/sample delay seems to be the only kind of inconsistency that comes up, at least for me.
I can confirm the same experience, in particular with a repeatedly loaded „default project“.
For whatever reason the content may change in mysterious ways. Memories from the days of phase cancelling tests... :D

That dummy assignment of a plugin to keep timing in line is a great advice 8)
User avatar
spacef
Posts: 3234
Joined: Sun Jun 17, 2001 4:00 pm
Contact:

Re: Parallel Processing

Post by spacef »

scary808 wrote: Sun Jul 10, 2022 4:19 pm
spacef wrote: Sat Jul 09, 2022 12:03 pm Hi,
I added the "quick start guide" to the downloads :
-->> https://spacef-devices.com/modularmixerv2-downloads/
Thank you for uploading the manual Mehdi! Could the “Phase Fix” device be inserted in the “Dry” signal line be used to correct the phase issues?

Regardless of whether it is phase coherent with parallel dynamics processing, the effect routing you have implemented is a brilliant idea!
I could do a device that is an insert and does nothing. it should take the micro latency of the insert and may align with another parallel insert (in the same channel probably). In the past I did some micro-delays that used dsp delays (not pc ram) but had too short max delay time for all situations. The dummy device is probably good to have in the tool kit as an alternative that takes less DSP resources. never really thought about that before (or had the time to do it if someone asked...). what effect do you already use ? I suppose any effect with dry/wet path is useable ?
plug-ins for scope
SpaceF website
SC website
scary808
Posts: 449
Joined: Mon Apr 02, 2001 4:00 pm
Location: Utah

Re: Parallel Processing

Post by scary808 »

For the record I do not own this device (yet) I'm simply doing some research for my requirements. I think some of the ideas thrown around in this thread are very helpful.

Thank all of you for your insights. This is why Scope is so great!

It looks like there are some dynamics plugins that have wet/dry mix that can achieve parallel processing so there are a lot of ways to approach this. I was just curious if I could do parallel compression with Vinco and not cause problems. Maybe a dummy Vinco in parallel set to not compress?

I'm excited to add MM2 to my workflow
User avatar
astroman
Posts: 8406
Joined: Fri Feb 08, 2002 4:00 pm
Location: Germany

Re: Parallel Processing

Post by astroman »

I‘m only into parallel compression if it‘s the Motown style:
squash 1 signal to the max (it would sound horrible on it‘s own) and mix that in a small dose with the source.
I noticed that years ago in a record shop on a vinyl that I handed back to the seller and commented „no, it‘s remastered“.
He pointed to the label and replied „look, original pressing...“ :oops:
Later I read how and why they did that kind of processing.

But since I use Saw Studio my use for compression with level as it’s main target is close to zero.
So much easier to just set the threshold and target loudness on the Levelizer plugin (VST equivalents Anwida L1V or FabFilter Pro L2 and possibly more today).

The process may seem simple, but obviously it‘s rather tricky. FabFilter needed years to get to ProL2 from ProL1, which had a significant amount of distortion.
Today all 3 can do enormous gains in loudness with almost no THD increase at all... IF the signal contains peaks. It doesn‘t work on a „sausage waveform“.
Maybe Mehdi can have a look into this case of processing. No big monetary gains to be expected, tho... L1V is just 50 bucks.
User avatar
spacef
Posts: 3234
Joined: Sun Jun 17, 2001 4:00 pm
Contact:

Re: Parallel Processing

Post by spacef »

scary808 wrote: Mon Jul 11, 2022 5:59 pm For the record I do not own this device (yet) I'm simply doing some research for my requirements. I think some of the ideas thrown around in this thread are very helpful.

Thank all of you for your insights. This is why Scope is so great!

It looks like there are some dynamics plugins that have wet/dry mix that can achieve parallel processing so there are a lot of ways to approach this. I was just curious if I could do parallel compression with Vinco and not cause problems. Maybe a dummy Vinco in parallel set to not compress?

I'm excited to add MM2 to my workflow
I don't think there is a difference between types of plugins, as long as they are inserts. That's because you want to use the dry channel and no effect parameters. The inserts functions places the effect in "another dimension", I believe it uses the PC ram too, hence a little bit of latency/micro-latency.
But the "Use Vinco with parameters that do nothing" is a good idea, because it goes through the same sound path, and it has an output volume knob.

For phase issues, at designer level, there are not a lot of solutions:

- first it concerns big modules that need more than 1 DSP. Keeping plugins simple and small avoids phase issues in 99.999% of the cases.

- designer places the plugin or a part of the internal circuit in a group with attribute flag "load on same DSP=Yes" . This is the method used in Scope since old days. That's how Red Muze's modular modules are made. The limit of this method is that the particular module is supposed to be loadable on a single DSP, so only small modules can do this. This is one of the reasons why BB3 oscillators are monophonic. I have done a mixer (FP104) which attempts to load the whole device on same dsp as well as internal parts (there are multiple levels of "OnSameDSP", but we know the mixer itself cannot load on a single Scope-PCI DSP). Nobody knows why, but it works in terms of compact loading. Not certain in terms of phase.

- Manual alignment: Use micro-delays: manually adding small DSP delays on each channel, and let user the delay time that he thinks fit, manually, on every channel, with or without linking functions etc. This is very DSP hungry, and not very practical because it is done manually by ear.

- Automatic phase align: like on STM mixer(s). This is a script that is not extractable from the STM mixers and not useable by designers. Does not fix any latency due to inserts, only channel latencies. I am not fond of this method, and personally do not like the sound of it. That's one of the many reasons why I began doing mixers, because I needed alternatives (other reasons are: I like doing mixers, needed other effect methods, there are better mixer to do for scope that simulation of hardware etc etc

- Send the device to Sonic Core for the ultimate phase fix dsp align placement. This is how it is done ultimately, for any device, and it could be done for modular mixer. This fixes internal channel latencies and phase across multiple DSPs, but (if I am correct) it does not allow user to place on different DSP. It is kind of "hard-wired" as placement is forced. It would not help for "insert latencies" though, so it would not change a lot for parallel FXs. Note that Holger proposed to me to do it once, but in that time, the Modular Mixer was far from finished. Then I forgot, but at the same time, I do not know if it could change something with big modules like the S12. It could be nice for smaller modules such as 4 channels, routers etc. A said, it does not change anything for inserts latency.

- The way the mixer is built changes a lot. For example, you could build a single channel module , and then import that channel into your mixer. So internally, the channels are grouped together and a single channel can be instructed to load on same DSP. or smaller parts of that channel. I've done this, but also built mixers where the groups are the horizontal sections: all mutes, all inserts, all volume faders, all those are in a large groups that are inserted one after the other inside the mixer. There is no "channel" visible in the circuit, just a succession of modules which are each line of the mixer. It means that you can flag OnSameDSP "all mutes", all insert 1" "all insert 2" , "all solo switches" etc. You could also mix both method. None is better that the other, at the end, it all depends on how big is the device. All my FP mixers are build in that "serial" way and it works very well, could even save a little bit of DSP in comparison the parallel building. May change phase too but not necessarily. I have not used this method in Xite/Modular Mixers which are classic "channel-per-channel" builds (otherwise it would be hell to make).

- Choice of internal modules is important too. For example, the summing of all channels could use dynamic "Add32" module, the Normalized adders, or a network of "mix2" modules (makes use of dozens of mix stages). Soundwise, it is all the same, but in terms of DSP resources, placement, etc , it is not the same. I used the micro-mix method in the routers of Modular Mixer v1, and it takes a lot of DSP. In modular Mixer v2, I reverted to 32Adders and it is much better in terms of DSP resource, and indirectly it makes phase issues less likely to happen, as modules are more compact and more likely to be loadable on a single DSP.

Mixing all these possibilities is how you make a mixer that is also "new inside" and it has a direct or indirect influence on phase.

May be others do it differently, but I think that the only possibilities available in scope.
plug-ins for scope
SpaceF website
SC website
User avatar
Peter Drake
Posts: 214
Joined: Thu Nov 08, 2018 12:27 pm
Location: Outer Milky Way

Re: Parallel Processing

Post by Peter Drake »

Very informative, and might explain why I keep reaching for MMv2 and Black Box devices. Sometimes the ears make decisions I can't explain, even to myself.

Also, outside of Scope, Sound Radix Auto Align can do amazing things to bring a big mess back into alignment and phase. It is just scary and disturbing just how well it lines things up.

I used to make some good money doing technical tweaks on movie and television production audio. A lot of this work was fixing timing/phase like we've been discussing. One of the most difficult was re-synch/alignment between lavalier microphones on talent to audio from a semi-fixed boom microphone at camera position. Imagine two actors with wireless lavaliers walking by a camera position with a boom microphone that tracks them from camera. For even a short segment the editors wanted to use, it could be four hours of very intense editing to get everything aligned. With Sound Radix it takes just a few minutes.
Not perfect every time, but a huge time-saver in most cases.
User avatar
dante
Posts: 5043
Joined: Sat Nov 24, 2001 4:00 pm
Location: Melbourne Australia
Contact:

Re: Parallel Processing

Post by dante »

Check tests 1,2 and 3 which address cross farm latency in the XITE-1.

https://www.scopeusers.com/ScopeRise/is ... temast.htm
Post Reply