One thing that’s worth checking is whether you actually get much more CPU with more instances of the plugin. I’ve done some very unscientific testing in the past, and often a single copy of a plugin looks like it uses e.g. 10% of the CPU, but if you add 4-5 or even 10-20 copies it doesn’t necessarily multiply the CPU very much at all.
I may have a test with Serum 2 just to see what happens.
Did it ever go for 99 though? Never saw that myself
Original serum yes. Never had issue with it. This version tho, spikes easily to 135% with a push of a button. Reducing voice counts from 16 to 5 on some presets does help.
But still not as good as OG.
And dropping quality to Good(or whatever is lowest valuel
I am on win11, ryzen7 3700x with 32gb memory.
Though from interview with Mr. Bill, Steve was mentioning 2.1 update.
Maybe he can address these issues.
Anyone else having this issues?
I’ll play with it more these days. But if i can’t use these engines, then not worth it for me.
Ok i see on Gearspace people have exactly same issues.
Got an answer from a man himself
This is one point about it to make - the new engines are powerful but also hungry. I’m on M1 standard, and I have a patch with 2x WT and 1x granular, plus a few LFOs. This runs 25-30% on the Ableton CPU meter at 256 buffer, which is a fair bit. Certainly as much as my more hungry analogue emulations.
Perhaps it’s been shipped and then the power consumption will come later?
When i am at home I’ll try ypur setup to see percentages on my machine.
So far Intel ones don’t have this problem
I am expecting it to be more cpu intensive, but usable
There is a modulation source: relative velocity , what does it actually evaluate? If the velocity is higher or lower than the last note? And then do some modulation based on that?
Is there any possibility to modulate with note length? I.e. longer notes get different modulations?
I searched the PDF, but didnt find a suitable paragraph.
A very nice tool if you intend to you SFZ into SERUM 2 (Free)
https://www.mossgrabers.de/Software/ConvertWithMoss/ConvertWithMoss.html
The following multisample formats are supported:
I try and it work very nice (I test SF2 to SFZ… you may need to normalize the sample if the level is a bit too low)
Notes: If you intend to embed your SFZ layers and Sell your Presets packs don’t forget to ask authorizations to the owner of the SFZ (proper contract where the owner provide you the authorization so then you are protected), or be sure to own the rights. I saw plenty of DAW Templates recently of Producers that selling their Templates, but if they can release the song to a label and use sample collections (bought so they have a license) usually in the license agreement you can’t resell in mass. you can only release your music to people to listen to. Not embed Sample(s), even if you bought the Sample Collection. (Be very careful with that)
Yeah, optimizing per platform often comes after MVP. Sometimes the first patch builds even. But always hard to guess without direct talk of prioritization from the devs.
Well I have to say I’m having a great time with this thing. I have some very minor feedback that I will try to send in the right direction for future updates, but basically I think it’s really pretty solid and it retains everything that was good about V1. I think it comfortably stands toe-to-toe with other premium VST wavetable/granular synths.
Just to make sure it isn’t my old eyes - does anyone else find the text a bit fuzzy if they zoom the interface to make it bigger? It’s looks sharp at 100% but a bit fuzzy at 110% and 125% (which is about right on my 27-inch QHD monitor).
man this update is amazing. it’s everything i wished for, and not too much more so it still all feels very familiar.
the inbuilt sequencer is amazing for jamming, especially because you can instantly switch between patterns via midi notes. super fun to sequence from elektrons btw.
one slightly annoying thing is that there’s a bug where the sequencer stops sending midi note off messages so you’ll end up with tons of stuck notes until you delete and re-insert the plugin. did anyone else encounter that?
Yep. All of ‘Native’ plugins were half price over the holidays. Just checked my account. Modwave Native at 99 on Dec 4th. I got Opsix for 99 the previous year (2023).
I’ve decided I’m getting Serum 2 after work today. Too good to pass up.
Ahh I see! But that’s Korg, Serum is from another developer called Xfer, so I wouldn’t get my hopes up 
Don’t have Serum anymore, but I might buy Serum 2.
Relative velocity sounds interesting. You sure it’s not release velocity? If not then maybe it evaluates velocity over a given time or within a chord, although that would probably be average velocity…
i notice this on a bunch of VSTs, some dont use vector/svg/math based graphics so it’s all just scaled rasterized images. i believe some have different resolution graphics that switch when you set the scale to avoid this but i’m not too sure. Not sure why in 2025 they arent using vectors but i don’t know enough about the performance differences. Using GPU to do math based drawing is usually better
It’s mostly explained by laziness, I’m afraid. Making a good scalable UI takes work and time. Slapping the UI your designer sent you as a bitmap and call it a day is much quicker. Reality is the lack of a scalable UI does not translate to many, if any, lost sales, so I can hardly blame them.
i used to write all my UI graphics code using math when i worked on iOS dev because of all the varying screen sizes. It wasn’t that hard (granted it wasnt anything even remotely complex as Serum) but if you don’t do it from the beginning it’s not gonna happen. And i agree, i don’t think it will translate to crazy sales, but with the iPad and mobile devices having gained so much power it would be in their interest to do it at some point
this stuff would look really really bad on a high res screen when maximized
I just realised there is a Serum 2 FX version like with the original.
Free multi-band multi-FX plugin with piles of modulation built into it - yes please!
In serum’s case this must be it. Who knows what gremlins are lurking in the code base….
I did discuss it in another forum, its relative to the velocity send before, i.e. it allows for dynamic patches /playing - if you play crescendo you can utilize this to alter the sound.
Also it can FM from the filter output - the filter output is the wavebundle (oscillator mix) so its a feedback fm system. Quiete intresting.
I made a patch yesterday night, also with the new filter designer, and if you modulate this with a lorenz x,y path modulation - it made a very nice acid patch, and that with only one LFO.
I saw a recent interview with Steve Duda (recorded a couple of weeks before the release of Serum 2) where he mentioned that one of the first things the team did when starting this project is ‘remake’ Serum 1, so I assume it’s been rebuilt from the ground up.
Also apparently the UI is very modular now so it’ll be a lot easier to add features or do more creative reskins.
