FM8 in year 2017?

VST, AU, AAX, etc. plug-in Virtual Instruments discussion
bill45
KVRAF
2343 posts since 15 Jun, 2006

Post Tue Feb 12, 2019 7:51 am

CS_TBL wrote:
Sun Feb 10, 2019 12:35 pm
I can't believe it, but after many years of just being idle (had to create an account in order to download something, back then), I'm now writing my first post here. ^_^ I'm this bloke: https://www.youtube.com/user/mvsmsx/
Recently also did this one: https://www.msx.org/sites/default/files ... ntro-1.mp3 (all sounds are FM8)

Now, no-one at NI is probably reading this, nor taking it into account, but here're my points about FM8:

The pros:
- speed; it came out in 2006'ish or something, and it had to run comfortably on the computers from that era. I can run tens of instances these days, all with effects and whatnot. For larger arrangements (which is kind of my thing) FM8 is outstanding.
- vast; drums, strings, brass, woodwinds, mallets, you name it, FM8 can do it. Whenever something in my projects is synthetic, FM8 is my to-go synth. Always.
- organic; due to free-running operators, although that's actually the essence of FM itself and not unique to FM8.
- effects; there are many. Often enough I see new VST's with only a handful of effects (the usual bunch: reverb, delay, mod, distortion) with which they're missing the point. Often my basic FM-tone is rather simple and I actually spend more time in the effects section to shape my sound.
- free FM matrix; I also own a DX7, FS1r, and yes: fixed algorithms are very counter intuitive and far from spontaneous. I also have an SY77, which does a better job at that, but that one also hasn't been turned on for over a decade, I think. Long story short: I won't touch any synth with fixed FM algorithms anymore, ever.

The cons:
- bugs; there are some, like the unison on slow-attack sounds once you hit the polyphony brickwall. And there's a bug with the scaling (deleting a node). And there's a weird redraw bug that sometimes occurs in the equalizers, tweaking a parameter usually solves this.
- setting values in the mod matrix is sometimes arcane as the range of a parameter is greater than the allocated mouse range to scan. E.g. if you have a mouse range of 100 and a parameter range of 128 (just an example) then you are bound to see jumps here and there. This also happens with the FM matrix. Often, setting a value is truly pixelwork.

Wishes:
- random fluctuations on about all parameters
- more effects, not just variations of current effects, but also more effect slots. Think of bitcrushers, resampling, dynamics, vocoding, ER etc.
- operator copy
- random note-on timing fluctuations; especially with unison this gives you the impression of a section of humans playing instruments (e.g. strings, brass)
- drumkit mode, kind of how the Roland JV/XV synths do this: a complete set of parameters per key (including effects)
- make FM8 truly multi-timbral and store a set of 16 individual FM8 instances into one large bank.
- I would like more operators and additional noise operators (including true S&H noise)
- add user drawn operator waveforms
- include CC value transformers so that the effect of e.g. a pitchbend-up on operator A could be different from the effect on operator B (use a table editor)
- include a real frequency envelope per operator
- include operator waveform phasing, like in the SY77/99
- an extra LFO, I usually use one for random fluctuations on pitch, which costs me an LFO. In the Roland JV/XV there's at least the 'analog feel' parameter that doesn't cost you an LFO.
- a smooth random wave for the LFO, current steppy one isn't ideal for pitches 'n such.

Things which others have mentioned; better filter... I guess, if it makes you happy, sure. As long as they don't wipe the current one. I could imagine another character attached to a better filter (so we'd have two filters in the matrix).
Sounds great!!

ghettosynth
KVRAF
11433 posts since 13 Oct, 2009

Re: FM8 in year 2017?

Post Tue Feb 12, 2019 8:22 am

Sam Marks wrote:
Mon Feb 11, 2019 4:46 am
I think that its sound is beautiful and it is what interests me the most.

I don't need a hardware SY99 or TG77, with their cryptic interfaces and programming, to do proper RCM synthesis anymore ;)
It does sound nice enough. The 4FX unit includes SpaceMaster with simplified controls which contributes a lot to many of the presets. I also really like the complex envelopes. This is something that I find annoying with Bazille. I don't think that the workarounds suggested by others are as convenient or immediate as having proper multi-stage envelopes. I also like the samples in the sampler, I'll probably lift that, samples and all, and stuff it into my own FM instrument.

But, I maintain that the U/I is terrible and this is where progress needs to be made in any such instrument. Arturia's efforts are reasonable and I think that FM8 strikes a nice balance there also. SY66, like so many Reaktor ensembles, chooses a clumsy way too many controls in a tiny space approach and then adds insult to injury with a color scheme from a late eighties video game. The matrix approach is lazy and desperately needs its own presets.

Also, it's a CPU hog. Now, by itself, that doesn't bother me that much. I don't mind paying more CPU for much better sound, but, here it seems like it could be much better and, without doing much digging, my gut sense is that it's just a side effect of some elements of the design and that it most likely could be improved.

This is the kind of Reaktor instrument that I don't like very much. It's a hodge podge of other builder's efforts cobbled together with terrible taste in U/I and workflow. Reaktor's U/I tools are hard to work with, but, you can do so much better than this. This instrument needs to have a tabbed macro approach to the operators with a simplified B page/tab that allows you to edit the basics of all operators simultaneously.

That kind of work is hard to do, much more than just using the basic Reaktor tools and slapping on some crappy knobs.

root
KVRer
1 posts since 3 Apr, 2007

Re: FM8 in year 2017?

Post Tue Feb 12, 2019 2:51 pm

ghettosynth wrote:
Tue Feb 12, 2019 8:22 am
I also like the samples in the sampler, I'll probably lift that, samples and all, and stuff it into my own FM instrument.
By all means, help yourself...preparing that content took 2 weeks.

The zdf filters are those provided in the R6 library. The operators were not tabbed on purpose.

I have some things in mind for an update but not the interface. There are some custom elements but further interface work would do nothing but consume time.

I agree that there is room for a new FM synth like FM9 which is why I designed SY66. Not in Reaktor though. SY66 took about a year of solid work. Creating it's own modulation buss so you can modulate anything and then snapshot those assignments took a long time requiring a project restart as an index and scheme for all targets and sources had to be made. Presets for the FM matrix might be handy. But those matrix knobs are part of the buss meaning you can change or morph them during performance. Can you morph between algorithms in any other FM synth? Maybe you can. As a pretty advanced FM sound designer I didnt think it would help but maybe it would.

A few SY66 features showed up in the latest batch of yamaha hardware. I think Reaktor is a fabulous platform for prototyping. I'm not sure their fft spectral source is the best thing for FM and seems like an interesting way to expand options for content in the operator but in the end it just seems weird to me and not a natural extension like say...open samples plus an in op filter. The in op filter is also handy in SY66 if you want to forgo FM entirely and just make a ripping subractive synth.

Overall SY66 and FM9 could benefit from what the music project for android which was used to make DEXED showed; the importance of a really big sine table. Only one quarter of a wavelength is needed because of symmetry but the very basic FM character can be best duplicated with a significantly larger table for sine approximation. This was an oversight in FM7/8 and could be addressed in SY66 along with a bump in precision of ALL the single cycle waves. The built in sines are not sufficient. The waveforms from early DX synths were all transformations and ways of simply reading out the sine table. Increasing the internal sample rate will not help if the sine table is only 256 values.

Building a custom sine object in core containing a table of values for one quarter wavelength and including the basic readout methods would be a logical next step for SY66. Which might have happened by now if I had not also been designing and building DILITHIUM.

Nice to see someone has been using SY66 though and thanks for the kind words Sam! I would love to see or hear how SY66 is working for you with non periodic waves. I always intended this but didn't include any of my own samples beyond single cycles...I could have ripped the 7799 waves but that would have been...shitty. ;)

Jonathan Adams Leonard

ghettosynth
KVRAF
11433 posts since 13 Oct, 2009

Re: FM8 in year 2017?

Post Tue Feb 12, 2019 4:25 pm

root wrote:
Tue Feb 12, 2019 2:51 pm
ghettosynth wrote:
Tue Feb 12, 2019 8:22 am
I also like the samples in the sampler, I'll probably lift that, samples and all, and stuff it into my own FM instrument.
By all means, help yourself...preparing that content took 2 weeks.
Yes, I know that's work and it's the kind of tedious work, like certain U/I work, that I don't like doing either. To be clear though, in case there was confusion, I'm not talking about anything being shared in the library or another product, I know that Reaktor users do that often without asking permission. If you've read my posts about Reaktor you'd know that I seldom share my work in the library and that I'm a stickler about copyright issues and distribution.

I have no qualms, however, with using bits of anything I have in other instruments in anything else that I make for use in my own music or live performance, without permission. That IS what Reaktor is to me, a toolkit of synth bits and if you put something out there then I'm not going to feel obligated to ask to use bits in my own music. In particular, I have an ongoing FM instrument that works for me, largely for the same reasons that you are expressing, that FM8 has not been developed further, but it is not, and it never will be, available anywhere. If it were to develop in that direction, I would not distribute it with anything that is not distributed with an explicit license that permitted redistribution, e.g., creative commons, BSD, etc.
I have some things in mind for an update but not the interface. There are some custom elements but further interface work would do nothing but consume time.
Agreed, but it's the thing that separates good instruments from weak instruments. This is a thread about FM8 and I think of all synthesis forms, FM is one where the U/I matters a lot. You can say that the operators are un-tabbed for a reason or by design, and sure, that's fine, but it also leads to an instrument that, in many ways, is far less usable than FM8, which this thread is about. Any instrument that is going to replace FM8 needs FAR more consideration to the U/I than what you've put into your instrument.
I agree that there is room for a new FM synth like FM9 which is why I designed SY66. Not in Reaktor though.
I see no issue with it being in Reaktor, but I think that the design has to be elevated, like, for example, Kontour. With all due respect, I don't think that you can project your own experience as an example of what's possible in Reaktor. I'd put money on NI being able to deliver a much more refined instrument, both in overall design, and in U/I and workflow.

I also get that a lot of that kind of U/I work is time consuming and I don't often do it either. However, I do think that you can leverage the basic tools in Reaktor to come up with a much more usable instrument.
SY66 took about a year of solid work. Creating it's own modulation buss so you can modulate anything and then snapshot those assignments took a long time requiring a project restart as an index and scheme for all targets and sources had to be made. Presets for the FM matrix might be handy. But those matrix knobs are part of the buss meaning you can change or morph them during performance. Can you morph between algorithms in any other FM synth? Maybe you can. As a pretty advanced FM sound designer I didnt think it would help but maybe it would.
You have your own tastes here, so whatever works for you, but, for me, you are expressing exactly the problem that I think plagues many Reaktor designs, namely, the embrace of excessive and unnecessary complexity without a commensurate effort in workflow and U/I. Sure, it's nice to be able to modulate/morph everything, but, how much value does that really provide? Is that where your CPU overhead is going?

To answer you question, FM8 allows you to morph all matrix settings from automation. Now, AFAIK, they (the matrix levels) are not modulation destinations for anything. I don't know if your instrument does that, I didn't pay that much attention.

However, I think that a good FM instrument must put a lot more effort into how the user interacts with the matrix and Reaktor provides decent tools for this. You can even improve it dramatically by ditching the physical knobs. A scrolling numeric value, i.e., just a knob/fader without the image showing, makes for a compact and readable display that is not strange to users of FM.
The in op filter is also handy in SY66 if you want to forgo FM entirely and just make a ripping subractive synth.
Yes, I've argued often for filters in FM synths. I think in this thread even. I don't even think that you need to make this really strong distinction between FM and subtractive, sometimes a filter is the right tool to get just the sound that you want in the larger context of an FM architecture.

The bottom line for me is that there is not a Reaktor ensemble that comes close to replacing FM8 and that, beyond basic features, limited effort in the U/I is one of the main reasons. SUKfm was a good effort in a 4-op synth, but there I think that there were some aspects that went too far in terms of being essential to a good U/I, for example, the retro looking gas florescent display elements. Sure, they look nice, but that could have been time better spent on creating a 6-op instrument instead.

Return to “Instruments”