Post by viciousbliss on Apr 11, 2017 2:20:21 GMT -6
Hello everyone,
Recently I had become frustrated with constantly running out of cpu on relatively small 96k sessions consisting sometimes of only one or two vocal tracks, 4-8 music tracks, 6 stereo aux tracks including the master fader, and 1-2 mono delay auxes. My computer is a Z800 with two 5670s. I had gotten some advice from an experienced person on how I might recreate a classic analog chain. Usually something like NLS-One Tape instance(Analog Channel/Phoenix/VTM)-Fix Doubler(sometimes Little Altar Boy too)-Bx_console-Pro Q2(sometimes)-CLA76-CLA2A-VLB902-Sknote Strip(for the balancing eq and cross talk, stereo aligned LAL on tracks and just aligned on the two main busses). I'd then add auxes for CLAEffects for throw delay, Ultrareverb or Echoboy, Primal Tap(mono), Doubler(usually DH910 or H3000), Music Reverb(ended up using Lexicon MPX to save cpu), Vocal Reverb(was last chaining T-Verb, H3000, and then MPX or Tsar-1), sometimes I'd add a second mono delay aux for another Primal Tap with a different setting, Music Bus, Vocal Bus, and Master Fader. On the busses I was using a few things. Another VLB902 to take care of leftover sibilance, Bad Buss Mojo, Bombardier, L1 or L2, Strip, maybe another AC202 or Phoenix, Satin on the master, Dh910 or Reel ADT at the top of the vocal bus, and maybe a Pro Q2. Music tracks might've just had NLS, Strip on some, Pro Q2, Bx_console, AC202 or Phoenix, and maybe a compressor like RenAxx or something not using a lot of cpu. Also used some Avid Aural Exciter and Big Bottom here and there. Mostly I'm working on Rock and Metal songs and I swear they use a lot more cpu than Rap or Pop sessions that are bigger running about the same number and type of plugs.
Then I saw a post on here mentioning the new Apogee Symphony and DSP. It seemed like adding DSP would be a cool idea. So I started researching. I've read enough bad things about HDX being really expensive and delivering way less than the cheapest Soundgrid Server. Forget that. UAD also doesn't seem to deliver a lot of value for what you're spending. I've done some emailing with Waves and Soundgrid seems interesting, but flawed. The servers will all work with a regular interface, but you need a Digigrid interface for the low latency thing and I think for the Emotion1 mixer and live sound. For $700 I can get an Impact Server that can run hundreds of their cpu efficient plugins(their plugin counts are almost all for mono instances and I was told by them that stereo counts are half the mono). But then you see that something like H-Reverb will run 8 instances on it and only I think 11 on the $2500 Extreme Server. With NLS you go from 80 mono on the Impact to 95 on Server One and 181 on the Extreme Server. There isn't a lot of improvement on most things when going from Impact to Server One. But they all have this limitation where you can't run more than 64 mono and 32 studiorack plugins per session(like Slate's VMR and that's the only way to use Soundgrid plugins, no single instances of a plug). I'm not even sure you could really utilize the full potential of the Extreme Server with that limitation. And you can't chain multiple servers for added dsp. Waves support told me multiple servers was for redundancy and some other reasons I don't know much about.
With that said I was curious what kinda result I would get by just using a bunch of Waves efficient plugins instead of what I've been using, maybe J37 instead of Satin, and minimizing use of non-Waves stuff. Frankly, the mixes got way better. Didn't even need the VLB902 anymore(sm7 is the mic into Audient ID14). Phoenix/AC202 didn't seem all that necessary either(debating about buying Phoenix after this trial runs out too). Swapped the Primal Taps for H-Delay and just left it at default settings. Put Supertap 6 in for Ultrareverb, again at default settings. Rverb for the MPX chain(also tried chaining Trueverb-Doubler 2-Rverb and using H-Reverb or Abbey Road Plates). For whatever reason I didn't need to do much eq. Took out Fix Doubler and Little Altar Boy for the most part, Doubler 4 for Dh910 Aux, swapped in SSL Channel for Bx_console when I didn't need a gate(this was the only "analog" button I turned on and only for vocal tracks, CLA76 and 2A had the 60hz on too),and simplified the bus chains. Music Bus was now API2500-L2. Vocal Bus Reel ADT or DH910(Doubler doesn't seem to work here very well)-SSL-EQ(mainly just to turn the HPF on), H-Comp(the one vocal preset with Punch then changed to 100), and maybe an L2. Master Fader was J37-API2500-L2. A side note about BX_Console. I found that it could be better if I set channels to digital and maybe left one on analog because a lot of the time the mix benefited more from the consistency of the digital setting.
I was trying to do as little tweaking as possible to each plug, only adjusting when I felt a need. What I found was that a lot of these analog plugins do things that are hard to qualify. They can throw things outta whack with the touch of a button. Then you have to load another plug and try to balance it out. From what I read in some VTM thread I think it was, is that a lot of classic recordings are the result of countless bounces and passes through the board and whatever electronics. How could that possibly be emulated on a computer? In reading about the Aural Exciter, it seems a lot of recordings got so dulled by the constant use of the tape that exciters were used to bring them back to life. I've always thought the Waves Aural Exciter was a bit harsh, but the Avid one is pretty good. I did this Waves workflow test on some older songs with the rock band tracks, a Soul Asylum soundchoice karaoke wav from my disc, and Madagascar from Guns N Roses using the lossless wavs. My thinking was that if I was gonna record vocals to go with music tracks recorded decades ago, it would make sense to use plugins that emulate what was used then. But that doesn't seem to be the case. In fact, it seems the analog plugins can end up making things sound harsh and whatnot, even if I'm being very conservative and barely driving them. I turned the drive off on NLS and didn't use the Buss component or put the NLS Channel on the music tracks. J37 I changed the track and tape type sometimes, but left everything else at default. CPU usage total was usually around 40% instead of 100% in this Waves-centric workflow where I'm using just enough analog plugs to get a good mix instead of trying to set everything up in hopes that I'm emulating a classic analog workflow.
As far as Waves goes, they have some serious holes in their product line. It doesn't appear that they have a gate anywhere as good as Bx_console. The tape plugs are cpu monsters. You're basically stuck with supertap or H-delay for a delay aside from the sig series(I get the impression that their business model revolves around getting hobbyists to buy up sig series/abbey road/ssl). Doubler can't touch Eventide(but Reel ADT does unique stuff that's awesome when it works). Rverb and Trueverb seem quite mediocre and I don't find cpu monsters Abbey Road Plates or H-Reverb as easy to use as Lexicon MPX, Tsar-1, and a few others(nothing really does what Tverb does either). They don't have much in the way of great modulation stuff when compared to Eventide and Soundtoys. I'm pretty sure most of the sig series plugs are cpu heavy. Waves support couldn't give me plugin counts for anything not on their site already.
The studiorack plugin limits seem to be the biggest problem. It forces you into a Waves-Centric workflow. Seems the reality of mixing at 96k is that you have to use cpu efficient stuff for the most part. I'm not aware of any other solution for added DSP other than buying a server of some kind. Anyone know of anything else? I'm not sure how much higher the plugin counts would be on some modern workstation costing thousands of dollars. Probably not enough to justify the cost.
Recently I had become frustrated with constantly running out of cpu on relatively small 96k sessions consisting sometimes of only one or two vocal tracks, 4-8 music tracks, 6 stereo aux tracks including the master fader, and 1-2 mono delay auxes. My computer is a Z800 with two 5670s. I had gotten some advice from an experienced person on how I might recreate a classic analog chain. Usually something like NLS-One Tape instance(Analog Channel/Phoenix/VTM)-Fix Doubler(sometimes Little Altar Boy too)-Bx_console-Pro Q2(sometimes)-CLA76-CLA2A-VLB902-Sknote Strip(for the balancing eq and cross talk, stereo aligned LAL on tracks and just aligned on the two main busses). I'd then add auxes for CLAEffects for throw delay, Ultrareverb or Echoboy, Primal Tap(mono), Doubler(usually DH910 or H3000), Music Reverb(ended up using Lexicon MPX to save cpu), Vocal Reverb(was last chaining T-Verb, H3000, and then MPX or Tsar-1), sometimes I'd add a second mono delay aux for another Primal Tap with a different setting, Music Bus, Vocal Bus, and Master Fader. On the busses I was using a few things. Another VLB902 to take care of leftover sibilance, Bad Buss Mojo, Bombardier, L1 or L2, Strip, maybe another AC202 or Phoenix, Satin on the master, Dh910 or Reel ADT at the top of the vocal bus, and maybe a Pro Q2. Music tracks might've just had NLS, Strip on some, Pro Q2, Bx_console, AC202 or Phoenix, and maybe a compressor like RenAxx or something not using a lot of cpu. Also used some Avid Aural Exciter and Big Bottom here and there. Mostly I'm working on Rock and Metal songs and I swear they use a lot more cpu than Rap or Pop sessions that are bigger running about the same number and type of plugs.
Then I saw a post on here mentioning the new Apogee Symphony and DSP. It seemed like adding DSP would be a cool idea. So I started researching. I've read enough bad things about HDX being really expensive and delivering way less than the cheapest Soundgrid Server. Forget that. UAD also doesn't seem to deliver a lot of value for what you're spending. I've done some emailing with Waves and Soundgrid seems interesting, but flawed. The servers will all work with a regular interface, but you need a Digigrid interface for the low latency thing and I think for the Emotion1 mixer and live sound. For $700 I can get an Impact Server that can run hundreds of their cpu efficient plugins(their plugin counts are almost all for mono instances and I was told by them that stereo counts are half the mono). But then you see that something like H-Reverb will run 8 instances on it and only I think 11 on the $2500 Extreme Server. With NLS you go from 80 mono on the Impact to 95 on Server One and 181 on the Extreme Server. There isn't a lot of improvement on most things when going from Impact to Server One. But they all have this limitation where you can't run more than 64 mono and 32 studiorack plugins per session(like Slate's VMR and that's the only way to use Soundgrid plugins, no single instances of a plug). I'm not even sure you could really utilize the full potential of the Extreme Server with that limitation. And you can't chain multiple servers for added dsp. Waves support told me multiple servers was for redundancy and some other reasons I don't know much about.
With that said I was curious what kinda result I would get by just using a bunch of Waves efficient plugins instead of what I've been using, maybe J37 instead of Satin, and minimizing use of non-Waves stuff. Frankly, the mixes got way better. Didn't even need the VLB902 anymore(sm7 is the mic into Audient ID14). Phoenix/AC202 didn't seem all that necessary either(debating about buying Phoenix after this trial runs out too). Swapped the Primal Taps for H-Delay and just left it at default settings. Put Supertap 6 in for Ultrareverb, again at default settings. Rverb for the MPX chain(also tried chaining Trueverb-Doubler 2-Rverb and using H-Reverb or Abbey Road Plates). For whatever reason I didn't need to do much eq. Took out Fix Doubler and Little Altar Boy for the most part, Doubler 4 for Dh910 Aux, swapped in SSL Channel for Bx_console when I didn't need a gate(this was the only "analog" button I turned on and only for vocal tracks, CLA76 and 2A had the 60hz on too),and simplified the bus chains. Music Bus was now API2500-L2. Vocal Bus Reel ADT or DH910(Doubler doesn't seem to work here very well)-SSL-EQ(mainly just to turn the HPF on), H-Comp(the one vocal preset with Punch then changed to 100), and maybe an L2. Master Fader was J37-API2500-L2. A side note about BX_Console. I found that it could be better if I set channels to digital and maybe left one on analog because a lot of the time the mix benefited more from the consistency of the digital setting.
I was trying to do as little tweaking as possible to each plug, only adjusting when I felt a need. What I found was that a lot of these analog plugins do things that are hard to qualify. They can throw things outta whack with the touch of a button. Then you have to load another plug and try to balance it out. From what I read in some VTM thread I think it was, is that a lot of classic recordings are the result of countless bounces and passes through the board and whatever electronics. How could that possibly be emulated on a computer? In reading about the Aural Exciter, it seems a lot of recordings got so dulled by the constant use of the tape that exciters were used to bring them back to life. I've always thought the Waves Aural Exciter was a bit harsh, but the Avid one is pretty good. I did this Waves workflow test on some older songs with the rock band tracks, a Soul Asylum soundchoice karaoke wav from my disc, and Madagascar from Guns N Roses using the lossless wavs. My thinking was that if I was gonna record vocals to go with music tracks recorded decades ago, it would make sense to use plugins that emulate what was used then. But that doesn't seem to be the case. In fact, it seems the analog plugins can end up making things sound harsh and whatnot, even if I'm being very conservative and barely driving them. I turned the drive off on NLS and didn't use the Buss component or put the NLS Channel on the music tracks. J37 I changed the track and tape type sometimes, but left everything else at default. CPU usage total was usually around 40% instead of 100% in this Waves-centric workflow where I'm using just enough analog plugs to get a good mix instead of trying to set everything up in hopes that I'm emulating a classic analog workflow.
As far as Waves goes, they have some serious holes in their product line. It doesn't appear that they have a gate anywhere as good as Bx_console. The tape plugs are cpu monsters. You're basically stuck with supertap or H-delay for a delay aside from the sig series(I get the impression that their business model revolves around getting hobbyists to buy up sig series/abbey road/ssl). Doubler can't touch Eventide(but Reel ADT does unique stuff that's awesome when it works). Rverb and Trueverb seem quite mediocre and I don't find cpu monsters Abbey Road Plates or H-Reverb as easy to use as Lexicon MPX, Tsar-1, and a few others(nothing really does what Tverb does either). They don't have much in the way of great modulation stuff when compared to Eventide and Soundtoys. I'm pretty sure most of the sig series plugs are cpu heavy. Waves support couldn't give me plugin counts for anything not on their site already.
The studiorack plugin limits seem to be the biggest problem. It forces you into a Waves-Centric workflow. Seems the reality of mixing at 96k is that you have to use cpu efficient stuff for the most part. I'm not aware of any other solution for added DSP other than buying a server of some kind. Anyone know of anything else? I'm not sure how much higher the plugin counts would be on some modern workstation costing thousands of dollars. Probably not enough to justify the cost.