Omnisphere 3

[Alexander_D]

Just got my hands on Omnisphere 3 and I'm hugely impressed by the new libraries. Appreciate being able to find things much more easily, but the presets are also a big step up and more usable — in particular the club/electronica orientated sets are much better than Omni 2. The effects rack is the icing on the cake!

How does the quality of effect rack BTW? Still nothing about them so much on YouTube...

[billinder33]

How does the quality of effect rack BTW? Still nothing about them so much on YouTube...

Really nice. Way more variety than Serum FX. Lots of verb/delay and distortion options. Nice vintage compressors and a multiband (but it doesn't resemble an OTT). Some other unique things you won't find anywhere else. All parameters can be exposed to the DAW for automation.

Main downsides are the potential lack of oversampling in some units that should have it and also stupidly a 4 unit limit per rack which makes no sense in 2025.

Overall a really useful toolbox though. I've used it a few times already and I've got all the usual suspects to choose from (Infiltrator, Shaperbox, HexFX, Transit, Minimal FX bundle, etc). The verb and delay options are probably its biggest strength.

[db3]

Agreed, the effects rack has a lot of variety and sounds really good to me. Makes sense if Overloud are involved as has been reported.

What I don't like about Omni 3, is that it appears many presets are duplicated across libraries — that's a big annoyance.

[Alexander_D]

How does the quality of effect rack BTW? Still nothing about them so much on YouTube...

Really nice. Way more variety than Serum FX. Lots of verb/delay and distortion options. Nice vintage compressors and a multiband (but it doesn't resemble an OTT). Some other unique things you won't find anywhere else. All parameters can be exposed to the DAW for automation.

Main downsides are the potential lack of oversampling in some units that should have it and also stupidly a 4 unit limit per rack which makes no sense in 2025.

Overall a really useful toolbox though. I've used it a few times already and I've got all the usual suspects to choose from (Infiltrator, Shaperbox, HexFX, Transit, Minimal FX bundle, etc). The verb and delay options are probably its biggest strength.

Thank you very much!

[Neon Breath]

New Omnisphere’s effect are top notch, very high quality sounding, especially the reverb ones and the creative ones. Love them!

[zvenx]

Maybe I am not looking hard enough.
I have seen a few complaints/comments here and vi-control that there are some patches that are in more than one of the 18 libraries (which to me kind of make sense).
But when I do All Omnisphere as my search path and randomly select a category and have them alphabetized, I haven't seen any patch name repeated in the last 3 or so minutes of doing so.. Is it clever filtering that they don't show the duplicate patch name when browsing this way?
rsp

[db3]

I'm not sure what's going on, but it's making it even harder to audition libraries and select favourites!
Boards of Burbank is an example. I have it starred in Analog Vibes, but it also exists unstarred in Ambient Dreams.
If I select all libraries only one version is shown which is unstarred. Spectrasonics need to remove all duplicates and recount the number of unique presets.

[zvenx]

Here is why I don't think removing duplicates is a good idea.
I am sure there are a few patches that fit more than one library.
so if you remove duplicates, someone who works specifically in one library, that that patch has been removed from may never ever know that patch exists which may be an inspiration for them for who knows what.
Not that it matters to me, but if they of course give a count of unique patches instead, sure.. no problem with that.
rsp

[IvyBirds]

Here is why I don't think removing duplicates is a good idea.
I am sure there are a few patches that fit more than one library.
so if you remove duplicates, someone who works specifically in one library, that that patch has been removed from may never ever know that patch exists which may be an inspiration for them for who knows what.
Not that it matters to me, but if they of course give a count of unique patches instead, sure.. no problem with that.
rsp

I totally agree with you

Sadly Spectrasonics is really caught between a rock and a hard place

Either they do as they decided to do or they have to decide to list a patch one place but not another. No matter what they do it's going to result in some one not being happy and will end up with online criticism

The #1 critical if Omnisphere 2 was that it was to big and you couldn't find anything and that you could end up spending hours working your way through presets

Eric called that an existential crisis so they went a different direction with Omnisphere 3 and tried to break the library up into collections only then you have the issue of how to divide everything up into collections and what to do with presets that will be at home on multiple collections

If they just went with not duplicating things between libraries then people would complain on why a certain patch is in one when it really should be in another and why that makes it harder to find again

In the end there is no solution that will make everyone happy

No matter how you want to count them there are many thousands upon thousands of presets with Omnisphere. You have to organize them somehow I don't think there is a way to do it that would make everyone happy

[Touch The Universe]

Put those multiple use presets into kitchen sink categories. If possible, collect ratings. Pay 100 people to go through them all and staring the favorites, sources and presets and find commonalities. It's an interesting problem to have and possibly to solve. There simply are too much - how one intelligently reduces that is the key - bring ai into the mix to train based on a pool of data outside, vs learning the individual user, after awhile, patterns could unfold that might be shared across different genres, age groups, skill level, etc. That problem will never go away - it must be solved eventually The end goal is to tailor a huge pool of resources to serve more and more specific uses and tastes.

[IvyBirds]

Put those multiple use presets into kitchen sink categories. If possible, collect ratings. Pay 100 people to go through them all and staring the favorites, sources and presets and find commonalities. It's an interesting problem to have and possibly to solve. There simply are too much - how one intelligently reduces that is the key - bring ai into the mix to train based on a pool of data outside, vs learning the individual user, after awhile, patterns could unfold that might be shared across different genres, age groups, skill level, etc. That problem will never go away - it must be solved eventually The end goal is to tailor a huge pool of resources to serve more and more specific uses and tastes.

But 100 random people will never rank the same patches the same way as you do. Everyone has different tastes and is working on different music

Unless everyone is making the exact same cookie cutter music as you are why bother?

If you find a patch you like, save it as a preset in your user library and then organize them there. You are probably going to tweak it a bit anyway

Then organize your user Library. Then it's simple just to see the list of patches you liked and tweaked. Rather than have then all buried inside of everything else

[Touch The Universe]

Train an ai to identify so called "cookie cutter" patches - that will vary by user.
100 random people and 100 specialized people - the goal is not only to find the ones most rate highly - that is a good start - but the more interesting solution will be perhaps finding multiple veins of interest contigent upon various user personalities. I could have fun with something like that and definetliy see the potential for experimentation.

AI needs to be trained on the 1) User and 2) Collective appeal and taste and find as many useful statistical patterns as you can - get creative in the patterns - that is the fun part, just as creative as making music.

Or, go twiddle one by one in 40k patches and have 500,000 people do the same tedious, unintelligent unispired thing - decisions decisions. But, to each there own.

[IvyBirds]

Train an ai to identify so called "cookie cutter" patches - that will vary by user.
100 random people and 100 specialized people - the goal is not only to find the ones most rate highly - that is a good start - but the more interesting solution will be perhaps finding multiple veins of interest contigent upon various user personalities. I could have fun with something like that and definetliy see the potential for experimentation.

AI needs to be trained on the 1) User and 2) Collective appeal and taste and find as many useful statistical patterns as you can - get creative in the patterns - that is the fun part, just as creative as making music.

Or, go twiddle one by one in 40k patches and have 500,000 people do the same tedious, unintelligent unispired thing - decisions decisions. But, to each there own.

The way people lost their sh*t in this thread over an AI image of a guitar, would seem to indicate that AI is not the answer

And the issue is the same. You will end up with the patches that the AI thinks sound awesome but they may or may not work for you, and more than likely they will skip right over patches that you would think are awesome

If you are going to use AI to pick your patches for you, you may as well have an AI to design your patches for you. That way you would at least have some unique patches

Google Gemini 2.5 Pro can already design Omnisphere 3 patches FWIW. It won't give you a preset file you can import, but it will walk you through ever detail of patch creation like a recipe

Gemini also thinks the Standout Patch is"In the Year 2049" It says it's a massive, Vangelis-inspired CS-80-style pad that perfectly demonstrates the new vintage glide modes and rich filter sound.
​
It also likes "Epic Analog Feelings," "Juno Resologue Lead," and "Warm Thoughts of the 80s."

[zvenx]

Personally I trust Eric Persing's sound aesthetic more than any other human or AI currently avaialble on Planet Earth.
rsp

[billinder33]

If you are going to use AI to pick your patches for you, you may as well have an AI to design your patches for you. That way you would at least have some unique patches

A categorization AI could apply a clustering algorithm to group presets together and generate some categories that would satisfy a majority of people. XLN XO and Atlas do a similar thing today. Would be awesome if Omni had a visual preset mapper similar to XO.

A built-in prompt based preset generator AI would be pretty sweet as well.

Fun to imagine, but wouldn't be at the top of my feature request list.