Fathom Synth Development Thread

[Scrubbing Monkeys]

I finally got a chance to fire up the latest versions or Fathom 2.29 30. Fahom is gaining ground on becoming my goto. It has been my goto for a while for complex evolving pads and anything wavetable. But lately the obstacles that were in the way of being a bread and butter synth are getting smaller and smaller. I can now load up a template for a basic 2/3 Oscillator synth with a couple of filters and I can quickly get where I want to go. The new filter is quite good. Juicy is what comes to mind. it handles extreme resonance very well. It is for now my favorite so far.

I didn't really pay attention to cpu improvement. My longer pad patches normally stay below 25% Which is fine for me. I use 3 detune voices with 2 bar envelope. My machine s a 2013 Dell laptop. 2.26 dual with 8 gb ram on Win7 using Reaper or Mulab.
Maybe a good way to test is to load up last years One synth challenge. 29 instances were just about to choke my machine. I will see what it does now.

[Amram]

I finally got a chance to fire up the latest versions or Fathom 2.29 30. Fahom is gaining ground on becoming my goto. It has been my goto for a while for complex evolving pads and anything wavetable. But lately the obstacles that were in the way of being a bread and butter synth are getting smaller and smaller. I can now load up a template for a basic 2/3 Oscillator synth with a couple of filters and I can quickly get where I want to go. The new filter is quite good. Juicy is what comes to mind. it handles extreme resonance very well. It is for now my favorite so far.

I didn't really pay attention to cpu improvement. My longer pad patches normally stay below 25% Which is fine for me. I use 3 detune voices with 2 bar envelope. My machine s a 2013 Dell laptop. 2.26 dual with 8 gb ram on Win7 using Reaper or Mulab.
Maybe a good way to test is to load up last years One synth challenge. 29 instances were just about to choke my machine. I will see what it does now.

29 instances of fathom , fathom is cpu ugh.
I am using the same laptop like you and there is alot of synths like zebra 2 and synthmaster , rapid , that dont use cpu like that.
I thought to shut up with untill you post this.

[zzz00m]

If I can stick to my plan of attacking the Vector SIMD for the next couple months then the CPU problem will be solved but I will have to start saying no to feature requests and all but the most serious bug reports.

I'll vote for Vector SIMD. Fathom has a lot of features as it stands, and more can be added later. Honestly in my opinion, Fathom has so many detailed features that if you can't get the sound your looking for, your not delving into the depths, and taking the time to understand synthesis combined with this extremely precise and detailed instrument. Just my opinion, but my opinion after delving into the depths.

I will 2nd that vote!

I think that the 8x performance boost offered by implementing vector SIMD would be a worthwhile effort.

[FathomSynth]

OK, your voices have been heard, I will focus on the CPU now exclusively and get it resolved.

Sorry, if I sounded defensive before. The CPU issue is really the only thing preventing Fathom from being a world class synth in the top 10, so I'll attack this now.

[grandmasterbird]

Another +1 for cpu optimisation, running a 2011 MacBook Pro i5 over here!

I realise of course that my machine is on the way to becoming obsolete (if not already) but gotta work with what I've got. Not sure how many other people run older machines but any cpu optimisation is much appreciated.

Yet to try the latest update

[glokraw]

OK, your voices have been heard, I will focus on the CPU now exclusively and get it resolved.

Sorry, if I sounded defensive before. The CPU issue is really the only thing preventing Fathom from being a world class synth in the top 10, so I'll attack this now.

Is it possible to have a global setting that determines how many
voices are used? This would mean the ablilty to look into
presets at startup, and overide the voice count, if different
than the chosen setting.

Or a utility that would parse a bank or folder of presets,
and do the same thing. This would help when you only need 2 or 4 voices, or want a certain larger number for xyz reasons. You could have the same bank set up with different voice counts.
edit: I think I would set my voice count at 4, and boost it
in some situations.
Cheers

[FathomSynth]

No, sorry, I'm concentrating on CPU now, please read back a few posts.

Plus there is already a global setting for the polyphony that is saved per preset.

All Fathom preset files are in XML format, you could easily do a global replace if you want using something like notepad++.

[glokraw]

There was a cool sound, that was using a goodly
amount of cpu. I noticed the polyphony was at 32
I changed it far lower, and could not hear
any sound degradation. Seems like it might be
one of the blades helping to skin the cpu kitty.
But I can barely code a postage stamp
Cheers

[AmbieticA]

There was a cool sound, that was using a goodly
amount of cpu. I noticed the polyphony was at 32
I changed it far lower, and could not hear
any sound degradation. Seems like it might be
one of the blades helping to skin the cpu kitty.

Absolutely Correct. That technique combined with the Version 2.30 Filter optimization makes that "skinned" kitty purr I've since adjusted accordingly

OK, your voices have been heard, I will focus on the CPU now exclusively and get it resolved.

MANY many Thank you's

[Kwurqx]

I know that all attention will go to CPU/Performance for now.

So.....I just added a feature request for Surge. But in another timeframe it could also help Fathomsyth stand out (it will no doubt not improve CPU use/performance).

(Kurzweil V.A.S.T. inspired) FUN Equations.

Text below quoted from:
http://users.skynet.be/aemit/kurz/funintro.html

What are FUNs?
FUNs are basically mathematical functions performed on one or two inputs (controllers, ENVs, LFOs etc) The output of these functions can be assigned to VAST paramaters just like any other controller.

Two things you must know:

All FUNs output values in the range -1 to +1. Controller inputs are converted to this range so while Modwheel goes from 0 to 127, as an input to a FUN it is converted to 0 to +1. Pitchwheel gives values -1 to +1.
Which brings us to the second vital bit of info. Controllers can be unipolar, i.e. only have positive values (like Modwheel) or bipolar, i.e. have positive and negative values (like Pitchwheel, some LFOs, ENVs).
It is important you understand these terms.

Some of the FUNs are very simple, others can be quite mystifying. These tutorials aim to shed some light on the FUNs. Those new to the K2000/2500/vx need not be intimidated. You can follow the text and play the programs as you go.

FUNs allow you to do a lot that you cannot do with simple controllers. They are in my opinion the most unique feature of the VAST system. The manual has a good section on FUNs. You should at least read the introduction to that chapter. As we go through the FUNs you should read the manual entries for them.

I am no expert on this stuff. I wrote this series of tutorials as a means of learning more about the FUNs myself. There will be many applications of FUNs which I have not touched on here, but I hope this gets you started.

And some "examples" (the names are just specific example KRZ programs, not relevant in this context)

FUNTUTE1 : simple additions and division
a+b, a-b, a/2+b, (a+b)/2, a/4 + b/2, (a+2b)/3, b/(1-a)
FUNTUTE2 : adds depth control to Src1, exponential controllers
ab, -ab, a*10^b
FUNTUTE3 : controlled chaos and stepped controllers
Sample B On A, Sample B On ~A, Quantize B To A
FUNTUTE4 : limiting output ranges
|a+b|, |a-b|, min(a,b), max(a,b)
FUNTUTE5 : simple switches
a AND b, a OR b, Track b While a, Track b While ~a
FUNTUTE6 : vary response times
lowpass (f=a,b), hipass (f=a,b)
FUNTUTE7 : Extra LFOs and how to modify their shapes
ramp(various types), sin(a+b), cos(a+b), tri(a+b), warp1(a,b) warp2(a,b), warp3(a,b)
FUNTUTE8 : a simple way to limit and weight control output
diode (a-b), diode (a-b +.5), diode (a-b +.25)
FUNTUTE9 : feedback equations which give unusual results
a(b-y), a(y+b), ay + b, (a+1)y +b, y +a(y+b), a |y| +b
FUNTUTE10 : making a curved response, fixing discontinuous signals, making discontinuous signals
(a+b)^2, warp4(a,b), warp8(a,b)

These mathematical calculations and/or complex and/or logical functions open up a world of modulation and intermodulation options. I don't know of (m)any synths that offer this. Apart from Kurzweil V.A.S.T engine based synths...

[FathomSynth]

Very cool, I will put this in the feature list.

I've been planning on doing a polynomial oscillator for a long time but these functions seem to be a more general case which includes many other mathematical options as well

[willsavino]

Here's a niche request:

When making custom wavetables, I usually edit wave 1 and then wave 16, and then use the "fill" command. Sometimes, I'll edit a wave in the middle as well, and then any time I use a fill from then on out, it never seems to react the way I want.

What I propose is a "fill from wave x to wave y" option or something. That way, if I edit wave 1, 9, and 16, I can fill from 1 to 9 and then 9 to 16, but then I can also decide "actually, wave 9 sucks" and REPLACE that with a "fill from 1 to 16" command. I feel like a blanket "fill" command does not give me the flexibility I'd like, and I often end up starting from scratch more often than I'd like.

If there is already an easy way to do this that I'm missing, please let me know!

[FathomSynth]

It already does this, I tested it just to make sure I'm telling the truth.

Any wave position which has been initialized is internally considered "clean".
Any wave position untouched since a table was loaded is considered "clean".
Any wave position touched in the editor is considered "edited".

When you hit fill it fills in all the clean positions between edited positions.

I just tried it by loading a wave table from file and editing five waves at positions 1, 5, 9, 12 and 16.

Then I hit fill and it filled in smoothly positions 2,3,4 between 1 and 5, positions 6,7,8 between 5 and 9, positions 10,11 between 9 and 12 and positions 13,14,15 between 12 and 16.

[Scrubbing Monkeys]

Wow. I thought being able to use 29 instances of Fathom was pretty good considering it was known that optimizations had not been made. Who needs or uses 29 instances of anything other than eq in a normal production.

Also with something like Fathom there is no need to use multiple instances to layer. The only reason that count was that high was for drums and perc other wise it would have been very managable.

Now If I were using a model d emulation and needed 3 instances for one kick sound. I would have probably used more and ended up with the same result.

Now....If it were an ambient piece. It is possible that 10 instances of many high quality synths would result the same.

The biggest cpu drains for me are ...thick phasor, fractal rev. and detune.

[willsavino]

Any wave position which has been initialized is internally considered "clean".

Ahh, this is what I'm having trouble with. Even after initializing the waves I want to "fill over," it ends up doing a very short fill to the now initialized wave (i.e. it morphs from a regular wave to a flat wave over the course of "1" wave position.

This seems to only happen after I've already successfully filled once. In other words: it works fine at first, but if I've already filled, waves that would normally be considered clear stop acting clean.