Rossum Control Forge and Satellite

[kxx303]

Is it possible to create some “turing machine” algorithm woth Control Forge? Like in “TM” module by 2hp.
Control Forge can be up to 8 steps of quantized/unquantized sample and hold generator. Next we need to teach it how to lock a pattern and how to unlock it gradually, by cv 1/2/3/4 parameter inputs.
Looks easy on paper but how to do it?

[mdoudoroff]

Nope, sorry.

[mritenburg]

Is it possible to create some “turing machine” algorithm woth Control Forge? Like in “TM” module by 2hp.
Control Forge can be up to 8 steps of quantized/unquantized sample and hold generator. Next we need to teach it how to lock a pattern and how to unlock it gradually, by cv 1/2/3/4 parameter inputs.
Looks easy on paper but how to do it?

There is one way to simulate a TM, but it is extremely labor intensive. You start with an 8 step sequence that you like, make multiple copies of it, adjust each copy to output the variations to the original sequence that you would normally expect the TM to provide, and then chain the sequences so that that they play one after another. You would have to modify each successive sequence's parameters to simulate the various stages of unlocking and locking patterns. Really what you are doing is that you are planning out, say 128 steps (8x16), and programming each section of 8 to simulate what you would expect a TM to do. Again, only simulating a TM, super labor intensive, but possible.

[kxx303]

There is one way to simulate a TM, but it is extremely labor intensive...

This is very interesting but yes requires some effort. Thank you.

Your advice gave me an idea that Control Forge’s interactive and playable preset system is like a tree. You can grow any shape you want, but this process is very slow. Maybe even a lifetime.

[mritenburg]

Your advice gave me an idea that Control Forge’s interactive and playable preset system is like a tree.

Indeed! If you view each 8 segment sequence as a branch on a tree, you can move sequentially between the branches starting at the root, and with conditional jumps, you can jump non-sequentially between branches of the tree, or even between trees, with varying degrees of predictability.

[mdoudoroff]

I don’t think the above is simulating a Turing Machine; rather, it’s merely programming a series of deliberate 8 step sequences. You could call it simulating one potential outcome of a Turing Machine, I suppose. All fine if it meets a particular, immediate, utilitarian need, but maybe not playing to the strengths of Control Forge.

Control Forge is very good at precision and performance, but the provided tools aren’t rich for integrating random or facilitating diverse outcomes over time. You can step forward and backward, and you can use boolean tests, but only to pre-ordained results. Perhaps peppering your presets with enough conditional jumps, you can sort of wring an impression of random behavior?

[RGB]

Thanks for the replay! Yes, that is essentially the idea. Thanks for your suggestion. I guess that might be a lot of programming time to accomplish something that you would think would be a pretty simple task with such a robust module. Maybe there is another way outside of jump targets..?

I both love this module and hate it at the same time....

Maybe we can get a bunch of wigglers to ask Rossum if it would be possible to update the firmware? Rossum has been very responsive to requests in the Assimil8or discussion.

I don't know what goes into programming the firmware for the Control Forge, but it seems like the programming logic would be simple, something like "if the condition is met, pick a random number 1 through 8, then jump to that segment number, otherwise proceed to the next segment."

Yes, I think everyone in this thread would love to see a firmware update but it seems the Rossum crew has been absent for quite a while when it comes to Control Forge. It is to bad because just one firmware update implementing the various suggestions from this thread would take this already very special and amazing module to another level.

Until then, I will use your recommendation as a work around. Thanks again for the note. I will also use as many empty segments available to add additional jump conditions per preset. Rather than only being limited to one jump condition per segment, using the remaining empty segments to program more possible jump options based on incoming cv1 and cv2 values.

+1 FIRMWARE UPDATE PLEASE!!!

[kxx303]

I don’t think the above is simulating a Turing Machine...

By the way Control Forge is almost there with that “Randomness/Randomize” button. AFAIR in “Play mode” it flicks segment’s values momentarily. If it only could remember values untill the next press.

Edit:
+1 for the update of course

[RGB]

Control Forge is very good at precision and performance, but the provided tools aren’t rich for integrating random or facilitating diverse outcomes over time.

Very True... would love to see a firmware update that might help provide more - integrating random or facilitating diverse outcomes over time.

[mritenburg]

Control Forge is very good at precision and performance, but the provided tools aren’t rich for integrating random or facilitating diverse outcomes over time.

Very True... would love to see a firmware update that might help provide more - integrating random or facilitating diverse outcomes over time.

As a group, we should come up with a list of feature requests then engage Marco to see if they are feasible/likely to be implemented. I'll start the list:

1. Random segment within the current sequence as jump targets
2. Random presets as jump targets
3.

[Marco Alpert]

As a group, we should come up with a list of feature requests then engage Marco to see if they are feasible/likely to be implemented.

I certainly can't make any promises at this point, but I'll be following along.

[kxx303]

As a group, we should come up with a list of feature requests then engage Marco to see if they are feasible/likely to be implemented. I'll start the list:

1. Random segment within the current sequence as jump targets
2. Random presets as jump targets
3.

3. Turing Machine mode - it was described earlier here. Somewhat similar to 2hp “TM”. Maybe it can be released as a preset.
4. Randomize settings aka randomized preset - like some of Intellijel modules can (“Rainmaker” an “Shapeshifter” for example)
5. Manually adjusted triggers - ability to set triggers on chosen segments. For example trig1 on segments 2,3,4,5,6 and trig2 on 4,5,6,8.

[dBVelocity]

A feature that could be of true help would be to just allow a clock in to the time scale cv for tempo sync that could be toggled from the global but has the mult/div options to sweep across from the data input. I prefer note values for durations and a clock could be set to push it along that mode without having to change the tempo on the preset should I use it in a different tempo.

[wavejockey]

i keep pointing to the EMU Command stations in that regard
every EG/LFO value has a zero value, up does numerical changes, down (below zero) goes into musical divisions

[dBVelocity]

Another small mod that I'd like to see: a different value display tweek. The negative value for Relative Quantized level to read e.g.. -1 Oct -2 semi instead of: -2 Oct +10 semi. Another example is just -2 semi and not -1 Oct +10 semi.
If you're doing a scale or mode based arp on a CV pass through and use relative steps in succession, it gets awkward as you move along steps in relation to the root.
Perhaps it's only more useful for the Relative Quantized as Absolute is fine the current way in relation to note value.
Hope that makes sense.. How do you all feel about this?

[kxx303]

Another small mod that I'd like to see: a different value display tweek. The negative value for Relative Quantized level to read e.g.. -1 Oct -2 semi instead of: -2 Oct +10 semi. Another example is just -2 semi and not -1 Oct +10 semi.
If you're doing a scale or mode based arp on a CV pass through and use relative steps in succession, it gets awkward as you move along steps in relation to the root.
Perhaps it's only more useful for the Relative Quantized as Absolute is fine the current way in relation to note value.
Hope that makes sense.. How do you all feel about this?

Nice idea. -1 Oct +10 Semi is a confusing thing indeed.

[dumbeat]

Cant believe you can't output from each segment separately. Its the basic property of any analog multi stage function generator, why not from a Digital very expnsive one?

[dBVelocity]

Yeah, how dare they be different from all the same old modules. Lol.

You can do plenty more than you see at first glance than you may realize. It challenges you to think a bit more where almost anything is possible with the proper approach, exponentialy with Satellites.

Switch behavior from the pass thru shapes, triggers on jumps, evaluation segments and chaining, quantizing, random etc. All of the tools you can't get on a simple multi stage.

[ignatius]

Cant believe you can't output from each segment separately. Its the basic property of any analog multi stage function generator, why not from a Digital very expnsive one?

really? what module does this? i'm curious. the encore UEG doesn't. i guess maybe verbos multistage? that's more a sequencer though right?

[dumbeat]

Verbos Multi Stage for one. And the CF is also a sequencer if you wanna get down to it, I don't see one reason not to have a CV or Gate or both output at any of the 8 segments. Really don't. Looping the segment would give you 8 different "LFO"'s or, if at audio rate, an VCO. Not to mention combinations of several segments. Totally different beast on a different level.

[hlmm]

... I don't see one reason not to have a CV or Gate or both output at any of the 8 segments. Really don't. Looping the segment would give you 8 different "LFO"'s or, if at audio rate, an VCO. Not to mention combinations of several segments...

You can do that with Assimil8or — you can cycle through channels using a single trigger input as well as access them separately. Worth mentioning is that each channel can have 8 different samples assigned to its zones which can be cycled through, accessed randomly or by CV.

[wavejockey]

and you can sample CV in it too

[dumbeat]

and you can sample CV in it too

Im talking about the Forge,

[BrokenBo]

Verbos Multi Stage for one. And the CF is also a sequencer if you wanna get down to it, I don't see one reason not to have a CV or Gate or both output at any of the 8 segments. Really don't. Looping the segment would give you 8 different "LFO"'s or, if at audio rate, an VCO. Not to mention combinations of several segments. Totally different beast on a different level.

VMS has only a gate output per stage.

you can loop segments in CF and switch between multiple looped segments that are at diff speeds (and therefore have diff LFOs/VCOs). you can process external CV, you can have random output per step etc. etc. there is literally nothing the CF can´t do that the VMS can except maybe choosing multiple arbitrary gate outs.

you might want to try reading the manual which describes diff uses very well.

[dumbeat]

Since you can't do any programming on Sattellite other than change duration of a contour and control 2 things via CV, isn't it by far more economical to get an Assimilator and have in essence 8 Satellites that sample your your contours in real time and frees Control forge on the fly?

8 Satellites is like 2k and takes up a ton of HP. Cant see the benefit in doing that if there is no programming possible.