Scene Controller discussion

GuyJames

Inspired
I’m on a quest to reduce cpu and it seems like the Scene Controllers and control switches are silent killers.

Cooper Carter has a great video out but it’s still got me confused so I’m hoping we can talk about it more here.



My questions are:

1: does programming one scene controller to do all the same things that two or three scene controllers could do with less programming really reduce cpu usage?

2: coopers video only demonstrates using 3 scenes. How does this work when you’re using more than 3 scenes in a preset?

3: lastly, is there any evidence or answer that setting the scene controller update rate from slow to fast strains cpu anymore?

Im only really using SC’s to switch my compression mix and amp gain parameters across scenes so not super involved. However if I can wrap my head around this maybe I can take more advantage of it and apply it to the reverb/delay blocks because at the moment I just toggle the channels to get the different settings I want but as you know there is a gap when switching block channels.
 
I watched that video and it’s confusing. Maybe if I watched it 10 more times. Tag CC on your post maybe he’ll answer some of your questions. Best luck man.
 
1. Yes. Fewer modifiers and controllers will consume less CPU overall. The trade off is possible added complexity in the modifier graph shape to get the specific parameter values you need for each setting. With one scene controller per parameter, you can set the parameter to any value for each scene very easily. When you double up on parameters, you're forced to use the modifier settings for each to make up any difference in the values you need, like if you need one parameter to go up while the other goes down for a given scene.

2. You can set the controller value for all 8 scenes. The tricky part is translating those 8 settings to the different parameter values you might need. Unless all 8 needed values follow a relatively simple modifier graph shape from Start to End, your options will be more limited the more discrete settings you need. Three values is easy because we have separate Start, Mid, and End values that we can adjust in the modifier. The more settings you need between those, the harder it is to dial in those exact values.

3. Nothing is free. Setting it to a faster update rate means the process has to calculate the value more often. More operations means more CPU load.
 
Last edited:
I don't know how much it will help, but I talk about how I used to use Scene Controllers on this Axe-FX III thread:

https://forum.fractalaudio.com/threads/my-construction-of-gapless-presets.176655/

Personally, I would set Update Rate to Fast and Attack and Release to 0 ms, and only change those parameters if you found you were having trouble with CPU. I used to use Scene Controllers quite a bit, and you can do a lot with them. I know you're dealing with power limitations, but I know with Scene Controllers I was able to get great gapless presets in the pre-gapless days of the Axe-FX III. They work the same in the FM3; it's just CPU you have to worry about.
 
Changing update rate is really only needed if you are getting artifacts during the transition between values. If you set your Attack and Release to zero, the value change happens almost instantly, so there's no need to calculate any additional values in the time between those values.
 
Man, I watched the YouTube video; I can see how you're dancing on the edge of your CPU there. Shit, if gapless makes it over to your unit, you can break this up to several presets and not tax the CPU like that, but I see you're getting the hang of it. Using just one Scene Controller is brutal for exactly the reason you ran into: the wacky modifier curves you're forced to experiment with.

It was killing me watch you suffer through that, because all this would be simple with at least one more Scene Controller. I would copy your preset to another slot and add a goddamn secund Scene Controller anyway just to see how far it pushes the CPU. Like, with dedicated Scene Controllers, you can use the spreadsheet I posted in the opening post of that thread I linked, and it'll tell you the exact Scene Controller percentage to program for each scene. You don't mess with the curves that way, and it's easy.

But if this is the most the CPU on your modern will handle, what you may want to do is create a spreadsheet of the few parameters you need to program, then try to map out the most straightforward curve t and controller page settings to get you there with less pain.

One thing to keep in mind is you can have curves that look way more f**ked up than that too. Just spend a few minutes using the other curve parameters of the modifier screen, like offset and scale, and you can easily set, e.g., the first part of your curve to beat straight horizontal line that then goes up. The curve can change in really interesting ways when you get deeper with it. If you try it, I recommend to get your curve close and remember that you can readjust your min and max settings to compensate.

E.g., your curve looks finally right, with the start and end looking good, but your scene 1 value is a little high. Well, maybe lower your min value below the value you need and get the start value a little higher to hit the target dead on. So many nonintuitive adjustments like that you can make.

But I think you're on the right track. The thing is, you can use Scene Controllers in many different interesting ways, and when you get into it deeply, you start to see more possibilities.

But like I said, hopefully the engineers will be able to port over gapless scenes and presets soon and you can forget we had this conversation. :)
 
I'm using a similar system for a gapless switch on the FM3 and in my tests, regarding this:
2: coopers video only demonstrates using 3 scenes. How does this work when you’re using more than 3 scenes in a preset?
this system works for 3 values (start - mid - end), so you need to have a consistent 3 possible values only in your scenes...

e.g.
Assuming I want to use only 1 scene controller for everything and having a gain of 2 - 5 - 8

My parameter range should go from 0 to 10
Screenshot 2023-11-30 at 15.32.14.png
Then my start - mid - end should be 20% - 50% - 80%
Screenshot 2023-11-30 at 15.33.25.png

So I can have now "only" 3 different values (in multiple scenes of course)
Screenshot 2023-11-30 at 15.36.15.png

In my case I have clean on 1 and 4, crunch on 2 and 5 and leads on 3 and 6...


--
If now I want to have the compressor with "opposite" values (higher for clean)
Assuming I want a mix of 75% - 30% - 25% (clean - crunch - lead)
My parameter range should go from 0 to 10
Then my start - mid - end should be 75% - 30% - 25%
Screenshot 2023-11-30 at 15.41.25.png
 
Forgot to mention... assigning a specific start - mid - end value, is reflected by placing the controller at 0 - 50 - 100%
You can achieve this result in a lot of combinations but I think this is the simpler
 
And just because I discovered some love for the Dirty Shirley, there you have a 6 scenes patch from clean to mean with 1 scene controller only 😂
Heavily influenced by your settings @GuyJames

I need to dig into how the scene controller is set up in this preset. It sounds good to me. Usually if I try a downloaded preset, it doesn't sound great on my rig and etc. The Shirley was my favorite amp pre Cygnus, and then it didn't sound the same. I tried it again last night and got happy with it. And saw this one so I figured I'd give it a try. It's cool to see how other people go about building presets. I'll use as a study in scene controllers. Thanks for posting.
 
For the sake of conversation:
I saved the preset posted in another location but using 4 different scene controllers + 1 control switch (all blocks exactly the same, all values the same) and
1: does programming one scene controller to do all the same things that two or three scene controllers could do with less programming really reduce cpu usage?
I noticed at least 1% increase in cpu.
Assuming that the value in FM3 edit is "real" every scene change was firing the cpu over 72%, while in the "1 scene controller only" it reaches barely 71%.
So this is not a scientific test... but maybe there's a reduction overall by using only 1
 
I need to dig into how the scene controller is set up in this preset. It sounds good to me. Usually if I try a downloaded preset, it doesn't sound great on my rig and etc. The Shirley was my favorite amp pre Cygnus, and then it didn't sound the same. I tried it again last night and got happy with it. And saw this one so I figured I'd give it a try. It's cool to see how other people go about building presets. I'll use as a study in scene controllers. Thanks for posting.
Glad if this can help!
By the way if you improve something let's share the results... I mean... maybe a Recto cab on this is overkill 😂
 
Glad if this can help!
By the way if you improve something let's share the results... I mean... maybe a Recto cab on this is overkill 😂

I like the Recto cab. The delay is set wrong, it gets loud when bypassed. I changed to mute FX in. The PEQ is just a fixed high cut yes? Compared to the Shirley I made last night, mine is brighter but still nice. I put your PEQ and comp in it too. Having a FAS modeler is a lot of fun.
 
I like the Recto cab. The delay is set wrong, it gets loud when bypassed. I changed to mute FX in. The PEQ is just a fixed high cut yes? Compared to the Shirley I made last night, mine is brighter but still nice. I put your PEQ and comp in it too. Having a FAS modeler is a lot of fun.
Good point on the delay, I forgot to set the mute.
The peq removes some spikes I don't like and add a high cut depending on the gain (~8k on the clean and ~5.2k on the overdrive)
The preset is intended for live use!

About the brightness I should have mentioned that I just restrung the guitar :tearsofjoy:
 
For the sake of conversation:
I saved the preset posted in another location but using 4 different scene controllers + 1 control switch (all blocks exactly the same, all values the same) and

I noticed at least 1% increase in cpu.
Assuming that the value in FM3 edit is "real" every scene change was firing the cpu over 72%, while in the "1 scene controller only" it reaches barely 71%.
So this is not a scientific test... but maybe there's a reduction overall by using only 1
Anything small helps cause it's a lot of those those little things that save the FM3 from freezing lol. Another user recommended reducing the reverb block echo density which really helped and honestly it's not a difference that is bad.
 
Anything small helps cause it's a lot of those those little things that save the FM3 from freezing lol. Another user recommended reducing the reverb block echo density which really helped and honestly it's not a difference that is bad.
that's for sure, especially live you cannot hear a difference but saves a lot of cpu!
If not already taken into account @Marco Fanton made this
 
Back
Top Bottom