Firstly, many thanks to Jim for all his efforts on v1.4 and to John for his continued persistence in pushing to make the Solaris even better than it is now. On the topic of future enhancements here's my thoughts (quite long, sorry):
For me the most pressing issue is fixing Linear FM to work correctly with audio rate sources. See this thread where I provided an explanation of why I believe the current implementation deviates from what one would expect from experience with other synths that provide the feature: https://forums.johnbowen.com/viewtopic.php?f=21&t=16350
Since I posted in that thread, Dave Smith has added Linear FM alongside the existing Exponential FM to the Prophet 12, an instrument which I believe takes some inspiration from the Solaris, and I think it's a real shame to see the Solaris surpassed in this way. IMHO the current "Linear FM" feature is pretty much useless in it's current form for use with audio rate sources since you can't make anything remotely tonal with it, and fixing this would open up a vast area of sound design for the Solaris that would make it pretty much unique when combined with the existing feature set.
Multi-timbrality: is this a DSP coding task rather than, say, voice allocation? I'm not hugely desperate for splits and layers given that 10-notes isn't a large number of voices to begin with, but I can see it being useful provided that the implementation allows voice allocation to be flexible enough. From what John has said with regard to voice allocation, it doesn't appear that voices would be able to be dynamically allocated between parts, so for me I would want at minimum to be able to choose how many voices from 1 to 10 (or however many remain from other parts) to allocate to each part. The way DSI instruments implement bi-timbrality by just dividing the available voices by the number of parts (two) has always bugged the ---- out of me, so it would need to be more sophisticated than that. A bonus would be able to use unison per part and choose how many unison voices the part would get separately from the number of voices allocated to the part (e.g. 2-voice unison detuned part with 2 voices for monophonic bass, plus an 8-voice, non-unison, pad).
One thing I would really
like to see in the Solaris is MPE support so that we can use MPE controllers like the Roli Seaboard and Linnstrument. Making the Solaris 10-part multi-timbral with each part on a different MIDI channel would be a way to achieve this. If you wanted to make life a lot easier for us a simple "MPE Mode" switch could work like a setup macro: when switched on it would take the current preset, make all ten parts of a 10-part multi use the same preset, allocate 1-voice to each part and set the parts' MIDI receive channel to be from 2-11 consecutively. Making channel 1 accept global controller values to be routed to all parts would be a bonus.
There's a number of "more of everything" type requests which is okay and all, but I'll echo some of the responses here: not at the cost of reduced polyphony.
Finally, on effects such as reverb: I can understand that live players might want a reverb in the Solaris, but to be honest, as a studio-only user I wouldn't want to pay extra for effects in the Solaris personally. If there was the prospect of integrating the FX with the Solaris's modulation sources then it might be different, but that doesn't look like it's on the cards. I would think that most of us are already pretty well served by plug-ins and external reverb units, and if I we're going to put money towards a reverb then I would rather spend it on a plug-in that would give me better utility overall because I could use it with every instrument in my studio rather than just on the Solaris. I'd also be able to choose the type of reverb that suited me from the many that are available on the market for pretty low prices. I don't think we can reasonably expect anything that competes with a Lexicon or an Eventide given the development time and the limitations of the Solaris hardware, and to be brutally frank, if what we end up with is just another so-so hall algorithm then I think that I'd just switch it off in favour of using something better that I already have, which would seem a waste, really. I understand that others may feel differently, though.