Fathom Synth Development Thread

[doctorbob]

Thanks @Frostline ... I'm not going mad then - tried for a few hours with all sorts of combos, editing the .ini files etc etc ... going to be quite a challenge with 1 month only to find our way around this modular beast.

The bridged 32-bit mono version does work OK. Presets take around 0.25% cpu on my system, not found any yet to push the cpu!!! [ i7 8700k,32gB ram]

Maybe the native 64-bit will be even better.

dB

[FathomSynth]

Frostline, No the free version does not have an MT option (but it will).

The MD option is better for Windows 7. But I think Windows 10 people need the MT option until I create the real Win 10 builds.

Doctorbob, I sent you a private message with the MT version.
Try that and see if it fixes your problem in Windows 10 and let me know.

ETA on the real Win 10 build, 2.7 is coming out tomorrow and will include an MT mono build, but I will have to do the Win 10 build after that. So one more week at the most. I have a Windows 10 drive I created just for this but I have to install Visual Studio on it.

I'm almost positive the Reaper problems are caused by running the Mono MD version in Windows 10. I'm going to make a presence on the Reaper forum and let everyone there know I'm working on the problem. The Windows 10 build should fix this.

[Teksonik]

Teksonik, I'm glad you posted today, I've been meaning to ask you if you ever solved your problem with getting multiple MIDI signals to modulate on the same dial such as PW and AT?

No, I went to using two different mods but to be honest I haven't had much luck with AT in Fathom. I still get "stuck" AT modulation when I add an Envelope to Osc Volume. I posted an audio example a few pages back. Can you please post an INIT patch with your modulate on the same dial method ? Maybe I can see what if anything I'm doing wrong.

Has this been reported or is it a known issue ? No readout on Fathom's GUI in FL Studio 12.5.1 (both 64 Bit)

Fathom 021918-2.png

[FathomSynth]

OK, I'll post a preset with the double MIDI modulation.

In FL Studio you are not seeing the CPU bar at all?

That is caused by the FL Studio buffer setting.
You have to set FL Studio to use "Fixed size buffers".

See the Manual section 13. Host Interfaces.

[Yeager]

OK, I'll post a preset with the double MIDI modulation.

In FL Studio you are not seeing the CPU bar at all?

That is caused by the FL Studio buffer setting.
You have to set FL Studio to use "Fixed size buffers".

See the Manual section 13. Host Interfaces.

In FLStudio "use Fixed size buffers" results in NO READOUTS, No midinotes on the big screen and no cpu usage in the top right corner.
Uncheck "use Fixed size buffers" and it's all there.

[FathomSynth]

Are you using Windows 10?

[Yeager]

Are you using Windows 10?

If you're asking me then No
Windows7 64 bit Flstudio11 32bit Fathom free 32 bit.

If I want to see the CPU and midinotes in Fathom I have to Uncheck "use Fixed size buffers" and it's all there.

[Teksonik]

OK, I'll post a preset with the double MIDI modulation.

In FL Studio you are not seeing the CPU bar at all?

That is caused by the FL Studio buffer setting.
You have to set FL Studio to use "Fixed size buffers".

See the Manual section 13. Host Interfaces.

Used Fixed Size Buffers was already selected because I had read that part of the Manual

[Teksonik]

Are you using Windows 10?

If you're asking me then No
Windows7 64 bit Flstudio11 32bit Fathom free 32 bit.

If I want to see the CPU and midinotes in Fathom I have to Uncheck "use Fixed size buffers" and it's all there.

I can confirm that here. (Win 7 64 bit FL Studio 12.5.1 [build 5]). In Panel Mode "Overlay" the Waveform shows as seen in the screenshot above but the CPU meter and Notes do not. Turning off Fixed Size buffers allows all to show.

[FathomSynth]

Fathom 2.7 is out!

2.7 contains massive CPU performance improvements for dial movement and graph editing during playback.

Much of the dial movement processor code has been completely rewritten to eliminate crackling during mouse drags and fast mouse wheel throws.

Increased CPU performance allowed Bicubic Spline Sample Table Interpolation to be enabled for all oscillators further reducing the HF noise floor.

Download page now contains both poly and mono MT builds for Reaper Windows 10.

[Teksonik]

Thanks ! I'll give 2.7 a go first thing in the morning.

[Frostline]

Tried the MT x64 mono in Reaper 5.76 after deleting previous Fathom installations.
Seems to load fine.

Played around a bit noticed a couple of things.
Using a basic analog waveform (whatever the first oscillator option is) playing a loop of C2 1/8th note on the beat.
Adjusting maximum partials to 2k seems to work fine. All signal above the 2k mark goes away.
But adjusting the lower values, any movement of the knob causes upper partials to return while knob is moved from the start value of 64 to 140, then when knob released the upper partials go away. Shouldn't the upper limit prevent partials above the set point all the time?

Second thing, ASDR still does not actually change until adjustment is released. I thought real-time adjustment was going to be implemented a while back but I could be confusing that with something else.

In the quick start guide, 6.4

Envelopes time is measured in time “periods” where each period is equal to one second or one
song beat depending on the Grid Units toggle setting.
Envelope periods are distinguished graphically with a slightly heavier white line than the
Divisions and Subdivisions line.

So then if I set grid units to beat and period to 1/8 does that mean the heavy white lines indicate a 1/32 note?
Also with grid units set to beat and divisions set to two and sub divisions at zero and grid snap enabled, what is it snapping to when it snaps to places other than the visible grid?
Is there any way for the user to increase the contrast between white line and slightly heavier white line?

[FathomSynth]

Frostline, By default the upper limit on the partials is the Nyquist frequency which is half the sample rate, which at 44K is 22K. The upper limit dial allows you to move this limit to a different value. For instance you want a bass patch to have a lot of high frequency content but nothing over 10K.

At run time the Nyquist limit or the Max Partials limit is strictly followed using precalculated sample buffers. If a note frequency would push the highest partials above the limit then a new sample buffer is chosen so that the highest partial in the oscillator stays under the limit. In real time, because the buffers are already calculated and can be accessed at a very high speed, the entire operation is very exact and has close to zero aliasing.

However, while the dial is being moved a lower quality version of the mechanism is used. This is done by taking the previously calculated FFT and doing a reverse FFT on it to get the new waveform with the right number of partials.

The reason this is done is because calculating the partial buffers is very real time expensive. Fathom uses oscillator buffer of 16384 samples per single cycle waveform, and includes bicubic spline interpolation between sample points, which is actually 16384 x 4 floats per buffer. This provides extremely high quality and extremely low aliasing, but takes CPU to compute. This would cause missing host blocks while the dial is moved.

So temporarily while the dial is being moved the lower quality calculation is used to prevent audio clicking. The lower quality calculation has a slight amount of distortion above the highest partial.

So if you have an accurate spectrum analyzer you will see images in the frequency spectrum above the limit determined by the max partials dial.

Then once the dial is released, Fathom kicks in the high quality calculation to get the final buffers.

This is why you will see the high frequency images while the dial is being moved and then they drop down to (almost) zero as soon as the dial is released.

Sorry I did not implement the real time ADSR envelopes yet. It will require some fancy coding. I will try to do that for the next release.

Yes, if you set grid units to beats and period to 1/8, Divisions to 4 and Subdivisions to 4, the heavy lines are (1/8)/4 = 1/32 and the light lines are ((1/8)/4)/4 = 128 of a beat.

Snapping use to be on the minor grid lines, but I got sick of having to set the Subdivisions super small so I could get very fine ADSR attacks, so I changed the snapping to one quarter of a minor line. So if you have Quantize Time enabled and you move the points, and put your glasses on and look very carefully when they snap, you will notice the point always snaps to 0, 1/4, 1/2, 3/4 of a minor line. Of course if Quantize Time is not togged on then there is no time snapping.

No, currently there is no way to change the line heaviness on the envelopes, but that is a good idea for the new settings page (coming soon). It will also have a color picker for the graph line colors. At some point you will be able to change the background color hue of the entire GUI.

[Frostline]

Awesome explanations for most everything, thank you.

So I take it we should not assign a modulator to the partial control knobs either?
Partials @ 80 Max Partials @ 1000.
SPAN shows nothing past 1k, which seems correct.
Attach a bipolar LFO to Partials without even giving any amount on the little dial on the modulator slat and SPAN showing content to 5k.

Also, admittedly most users will probably not ever do this but if one were to vigorously move the Partials dial many times and then stop, the effect of moving the dial seems to continue on for a bit like it stored the movements in a buffer and it takes a while to catch up. Eventually the dial may then display a 0 in the center and no longer move smoothly and jitters/jumps when attempting to move it. It will then also start to flash the add modulator button while mousing over seemingly random areas around the knob though when attempting to move the mouse pointer to click the add modulator button it goes back dark before getting there. Relaunching Fathom seems to fix the flashing and 0 in knob issue and I did not reattempt the vigorous knob twisting.

[Scrubbing Monkeys]

the vigorous knob twisting.

You are having way to much fun.