The previous post covered the basics of the Cluster synth engine -- my best guess as to what's going on under the hood.
Now I'd like to try and provide a critique of all aspects of Cluster, from UI to sound. This is maybe a sensitive topic, and I'm going to try not to sound like a pompous ass -- it goes without saying that the folks at TE are absolutely stellar and magnificent designers and my hat would go off to them if I wore one.
Moreover I adore the OP-1, so I don't want this to come off as some sort of bashing or flame-baiting -- I just want to genuinely inspect the current design decisions in the hopes of teasing out some interesting possible future directions or alternate paths, or at least starting a discussion about the whys and hows of each synth engine.
Anyway, this is the first attempt at such a critical treatment, so please bear with me as I delve into the wonders of Cluster.
Visual System
I have no idea what the best way to demarcate the various avenues of criticism would be, so for now I'm going to start by discussing the visual aspect of Cluster.
Compared to the rest of the synths in OP-1, I think that Cluster is one of the weakest in terms of visual presentation.
For one thing -- and AFAICT; pretend I'm saying this at the start of each and every statement on this blog! -- the waveform display is more or less pure eye-candy. It doesn't seem to represent what's being generated in any real way. Specifically, a single raw osc sounds like a sawtooth, but visually looks like a random lumpy sine wave. Similarly, raising/lowering the cutoff doesn't smooth out the bumps in the waveform, it just seems to change the amplitude of the waves. So, this animation is pretty but fundamentally useless compared to some of the other beautiful synth engines.
Not to mention, this animation doesn't appear to be cheap -- TE have already dialed it down from 6 to 4 animated waves, and even then things get chuggy and slow to a crawl when Blue, White, and Orange are at high values. It's just not worth it!
The functional aspect of the visuals is a pretty straightforward metaphor -- four knobs corresponding to the four controllable parameters of the synth. Again, I think this is very weak: it's vague, it's obfuscated and abstracted, and it's not really anywhere near as useful as what's on offer in some of the other synths.
Most of all it strikes me as strange since it seems to contradict the design ethos that permeats OP-1: "instead of doing what everyone else has always done, think of new interesting approaches that offer something above the usual -- try something different!"
The Cluster graphics go against this by using the standard interface metaphor of every other sunth, and in doing so they undermine one of the biggest strengths of the OP-1: being able to visualize synthesis in useful and meaningful ways beyond abstract parameter values. Instead we're looking at a few knobs.
(It occurs to me that I should have possibly started with one of the other synths, since this sounds very negative; it's just that out of all the synths, Cluster is IMO one of the least well-visualized.)
So, how about some suggestions instead of just lots of negative complaints?
For starters -- if we must have knobs on at least one synth, why not make the knobs as functional and informative as possible? For example, rather than arbitrarily having notches at the cardinal directions (up/down/left/right), the notches could be located at meaningful positions for each knob: in the case of Blue, notches could demarcate the point at which another oscillator is added to the mix; for Green, notches could delineate the boundaries between e.g the attack region on the right, the dead zone in the center, the decay region on the left, and the weird-ass region in the bottom left.
Much better than boring old knobs would be some sort of direct "interactive diagram" type of visualization: instead of abstract and meaningless, something rich and full of beautiful information!
In the case of Blue: why not just display the actual number of oscs? It doesn't have to be a plain digit -- it could be one or more lines, sort of like the strings in the String synth engine.
For Green, why not borrow an idea from the wonderful Tremolo and Random LFOs and visually show the envelope that's modulating the filter? It would be super-useful and rewarding -- and maybe even fun -- to watch the envelope morph and change shape in response to your movements.
Green is a very complex control, and a simple visualization of exactly what you're controlling would be a wonderful solution to communicating to the user what's going on. Seriously, the visual A/D env controls in those LFOs are brilliant -- why not borrow that idea for Cluster?
For White/Orange, it would be great to directly visualize the detuning that's happening: there's some constant detune, there's some drift, and they change in complex ways as you turn those two controls. It's very subtle, and having some graphics that help reinforce what's going on aurally might make this less of a confusing/bewildering section.
Why not e.g display the current root pitch as a line or point, and a cluster of offset lines/points representing the detunings of each of the oscillators. This way we could watch them drift around and see how turning the knobs changes the detuning behaviour!
(It occurs to me that it would probably be great if I could knock up some simple animated mockups of different UI ideas to better communicate; a picture is worth a thousand words and all that. Sadly this would require time and effort far beyond what I currently have to spare, however it's definitely an interesting idea in the future!)
Anyway, basically I think that Cluster could stand a complete visual overhaul -- currently the graphics aren't at all informative or even very interesting, especially in comparison to some of the other synth engines.
Parameters: Green
This control seems a bit haphazard and confused. For one thing: values between 7-8 o'clock sound identical to values around 2 o'clock. Is this a bug?
The granularity of Green is far too low: there are literally only 6 different options for decay rate. It seems like possibly the visual style is crippling things, because those tiny dials don't have a very fine resolution visually, and TE seem to be doing a commendable job of ensuring that any audible change corresponds to a visual change. (This is assuming that rather than no reason for making Green so coarse, the reason was to make it better match the coarseness of the visuals. But still: is this coarseness necessary or beneficial? I think not.)
This criticism applies equally well to all parameters here: by my reckoning there are only 32 unique values for each knob. Obviously there is a tradeoff between coarseness and speed of tweaking, but in this case I think TE have erred too far on the side of coarseness; a bit more fine control would be really appreciated!
The huge dead zone between 11 and 1 o'clock isn't helping matters either; why not make straight 12 the dead zone and free up more space for finer-grained control of decay/attack times? It seems like wasted space.
Finally, it would be nice to have some direct control over the cutoff. The whole "ganged attack/decay/cutoff" idea behind Green is amazing, but the current range of controls is just a bit limited compared to what it could be.
What if Green was divided into three segments: 7-10 o'clock is the "decay" region, with cutoff at max and CW movement gradually increasing decay time to infinity; 10-2 is the "cutoff" region, with CW movement gradually moving the fixed not-modulated cutoff from max to min; 2-5 is the "attack" region, with cutoff at min and CW movement gradually decreasing the attack time. This way the regions seamlessly blend with each other, and we get both the A/D control we currently have and some nice direct cutoff control that we current lack
.
Parameters: Blue
Can anyone hear the difference between 3+ waves? Frankly I can't tell 4 from 5 from 6 -- it's a very subtle change. This seems like sort of a waste of parameter space: what if we coupled it with something else?
Drawing inspiration from the wonderful and magnificent Dr. Wave engine (seriously, I can't *wait* to get to that one, it's my fave), one idea would be to introduce continuous control over the wave shape -- say, smoothly morphing from sawtooth to square.
From min to max, the knob would sweep through 1x saw -> 2x saw -> 3x saw-> 4x saw -> 4x saw/square morph -> 4x square -> 3x square -> 2x square -> 1x square. If TE are feeling super-fancy, the control could even wrap around, so that users could morph from 1x square back to 1x saw!
Finally, it sounds like some sort of attenuation is applied so that e.g 3 oscs isn't 3x as loud as a single osc. This is good, but it's maybe a bit too much currently; since detuning oscs tends to diffuse the sound and lower the overall level -- what with the waves partially cancelling each other here and there -- the current setup means that when tweaking the knob CCW, the sound gets markedly louder when it jumps to a single osc.
Since there's no direct volume control on the Synth page -- the user has to navigate to the tape page and adjust the input gain -- it would be better to ensure that all values of Blue produce an even level of output. Currently turning Blue up from one to many is basically bit of a "bury this in the mix" control :)
Parameters: White/Orange
I don't really have many thoughts on this beyond what was mentioned in reference to the visuals -- it would be great to have a more useful visualization of what's happening here. Drawing the current pitch offsets of each oscillator in some way would be great.
One consideration is that these two parameters together don't offer a super-broad palette of sound. Don't get me wrong, it's a wonderful and useful range, but it seems like a range that's small enough to be encompassed by a single knob. There must be a way to gang the two controls together in a way that preserves most of the original breadth, right?
I have no idea what wonderful possibilities the resulting freed-up knob should be put towards -- maybe continuously variable control over the filter, e.g lowpass<->bandpass<->highpass? I'm sure TE would come up with something good.
The End... for now!
This concludes my look at Cluster. Stay tuned for the next exciting installment of OP-101, in which I tackle the mysteries of the Digital synth engine!
Thoughts, comments, criticisms? The discussion thread for this post can be found here, on the wonderful unofficial OP-1 forums: http://ohpeewon.com/discussion/226/op-101-cluster
