Zebra 3 Beta Bug Reports
-
- KVRAF
- 37388 posts since 14 Sep, 2002 from In teh net
Shouldn't the ctrl knobs reset when you change each patch? They stay on the position they were in when I last tweaked them (which means the sound of each preset is changed)
-
tasmaniandevil tasmaniandevil https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=62450
- KVRAF
- 2170 posts since 22 Mar, 2005 from a planet called u-he
We are still working on this, as mentioned at the start of the first post in the beta 3 thread.aMUSEd wrote: Fri Mar 06, 2026 1:29 pm Shouldn't the ctrl knobs reset when you change each patch? They stay on the position they were in when I last tweaked them (which means the sound of each preset is changed)
And from the Special Notes section:
* Ctrl ABCD macro feature is still in flux, the parameters won't get saved with presets yet
That QA guy from planet u-he.
-
- KVRer
- 2 posts since 21 Oct, 2020
Beta 20977;
What a fantastic update to Zebra 2 this has been.
There are a few small things that seems to have slipped through, which is understandable in such a major project.
1) Like in the previous Beta, filter modules 5 and 6 in the FX grid, when clicking on the central graphic ‘WIP’ still does nothing. However the surrounding four arrows in the D-Pad work perfectly.
The other main filters 1 to 4, do call up the filter palette in the main grid as expected.
2) In the oscillator Main Editor the guide P38, says “Remove a selection, press Backspace, or right-click and select Delete”
- Pressing Backspace does not work.
Only right-click and select Delete works.
3) Again in the User Guide, P14, Module Groups section, it says ‘Use Ctrl (PC) + click to add or remove individual modules.’
I have found it is in fact Alt (PC) + click and not what’s in the guide. At least on my Windows 11 PC.
So far these are what I have noticed in Reaper v7.62 Widows 11 Pro 25H2, running on an i7-14700K
One other thing, the Hide Mouse On Drag is a great feature. I use it all the time.
But I find the parameter values while being adjusted, do not persist long enough in the Data Display window. They flick off very quickly.
On the other hand, when using the show Mouse on drag, the values remain displayed longer. Perfect in fact. Similar to the old Zebra 2, Diva etc.
Would it be possible to make the Hide Mouse option have the same display timeout?
What a fantastic update to Zebra 2 this has been.
There are a few small things that seems to have slipped through, which is understandable in such a major project.
1) Like in the previous Beta, filter modules 5 and 6 in the FX grid, when clicking on the central graphic ‘WIP’ still does nothing. However the surrounding four arrows in the D-Pad work perfectly.
The other main filters 1 to 4, do call up the filter palette in the main grid as expected.
2) In the oscillator Main Editor the guide P38, says “Remove a selection, press Backspace, or right-click and select Delete”
- Pressing Backspace does not work.
Only right-click and select Delete works.
3) Again in the User Guide, P14, Module Groups section, it says ‘Use Ctrl (PC) + click to add or remove individual modules.’
I have found it is in fact Alt (PC) + click and not what’s in the guide. At least on my Windows 11 PC.
So far these are what I have noticed in Reaper v7.62 Widows 11 Pro 25H2, running on an i7-14700K
One other thing, the Hide Mouse On Drag is a great feature. I use it all the time.
But I find the parameter values while being adjusted, do not persist long enough in the Data Display window. They flick off very quickly.
On the other hand, when using the show Mouse on drag, the values remain displayed longer. Perfect in fact. Similar to the old Zebra 2, Diva etc.
Would it be possible to make the Hide Mouse option have the same display timeout?
-
- KVRer
- 25 posts since 18 Dec, 2017
Osmose:
with mpe, Zebra does not respond to poly AT messages, you have to switch to channel aftertouch (but then it will repond per-note)
Dialing in pitchbend per note on the osmose is not great. i can only get it to do a full note, half a note, a quarter note range, etc.
when pitch bend depth is set to mpe, pitch bend wheel does 1 whole tone (2semitones)
the range you set there, never affects the per note bending on the Osmose. the range stays the same. I can change the bending ratio on the osmose itself to do 1, 1/2, 1/4 ,1/6 , /12 , /16, /24 ,/48 , /96, 192
in zebra, 1/ is always one semitone. On other software the value of the pitch bend range is used instead of 1 semitone. This allows to dial in bend range much more accurately.
The problem is this: on the Osmose my preferred pitchbend range is 2/3 of a semitone. this is the perfect depth for manually played vibrato's and sounds very natural. binary divisions do not. They are either too small or too large.
So i cannot get the bends on zebra to sound natural, or be in tune with the Osmose.
On other synths this is done by setting the pitch bend range, and division on Osmose, for example:
pitch bend range 32 semitones and a division of /48 gives a bend of exactly 2/3 of a whole tone note.
(!note that on Zebra 3, pitch bend depth of 32 is also not an option!)
(for reference, Win11, Ableton Live 12, vst3)
--Seaboard test to follow--
with mpe, Zebra does not respond to poly AT messages, you have to switch to channel aftertouch (but then it will repond per-note)
Dialing in pitchbend per note on the osmose is not great. i can only get it to do a full note, half a note, a quarter note range, etc.
when pitch bend depth is set to mpe, pitch bend wheel does 1 whole tone (2semitones)
the range you set there, never affects the per note bending on the Osmose. the range stays the same. I can change the bending ratio on the osmose itself to do 1, 1/2, 1/4 ,1/6 , /12 , /16, /24 ,/48 , /96, 192
in zebra, 1/ is always one semitone. On other software the value of the pitch bend range is used instead of 1 semitone. This allows to dial in bend range much more accurately.
The problem is this: on the Osmose my preferred pitchbend range is 2/3 of a semitone. this is the perfect depth for manually played vibrato's and sounds very natural. binary divisions do not. They are either too small or too large.
So i cannot get the bends on zebra to sound natural, or be in tune with the Osmose.
On other synths this is done by setting the pitch bend range, and division on Osmose, for example:
pitch bend range 32 semitones and a division of /48 gives a bend of exactly 2/3 of a whole tone note.
(!note that on Zebra 3, pitch bend depth of 32 is also not an option!)
(for reference, Win11, Ableton Live 12, vst3)
--Seaboard test to follow--
-
- KVRian
- 1044 posts since 17 Mar, 2005 from Bay Area
Very consistently, I cannot run the latest version of Z3 in cubase 15 without constant ASIO internal overloads, meanwhile Z3's internal meter shows as low as 3%. This may be that E core/P core minimum CPU usage issue, but in Bitwig retail 5 or beta 6 I see no overloads or unmanageable spikes. Im pretty much only using Z3 in BWS now, as almost every preset does this. This did not happen in the earlier beta versions of Z3.
Ive got an i9 12900KF with 32GB and a focusrite 18i20.
Ive got an i9 12900KF with 32GB and a focusrite 18i20.
-
- u-he
- Topic Starter
- 30186 posts since 8 Aug, 2002 from Berlin
MPE uses Channel Pressure. Poly Pressure is not defined in MPE. Poly Pressure may be received on the main channel (MIDI Channel 1 in our stuff), but that is only a fallback for devices that can't handle MPE. Therefore, in MPE mode we do not expect any Poly Pressure, nor should any MPE device send it.M32 wrote: Fri Mar 06, 2026 9:14 pm Osmose:
with mpe, Zebra does not respond to poly AT messages, you have to switch to channel aftertouch (but then it will repond per-note)
As for Pitch Bend: You might have to set Pitch Bend Up/Down to "MPE".
I do not know the exact details, maybe it's best to ask our support team if the user guide does not fully reflect the options yet.
-
- u-he
- Topic Starter
- 30186 posts since 8 Aug, 2002 from Berlin
That is surprising, as I could hardly play half the presets in the previous version, and with this version I can play all presets without any overloads.Milkman wrote: Fri Mar 06, 2026 10:16 pm Very consistently, I cannot run the latest version of Z3 in cubase 15 without constant ASIO internal overloads, meanwhile Z3's internal meter shows as low as 3%. This may be that E core/P core minimum CPU usage issue, but in Bitwig retail 5 or beta 6 I see no overloads or unmanageable spikes. Im pretty much only using Z3 in BWS now, as almost every preset does this. This did not happen in the earlier beta versions of Z3.
Ive got an i9 12900KF with 32GB and a focusrite 18i20.
What buffer size do you use in Cubase?
-
- KVRian
- 1044 posts since 17 Mar, 2005 from Bay Area
Im usually at 128/44, and Ive tried 128-256/44 lately. Buffer size doesnt seem to matter.Urs wrote: Sat Mar 07, 2026 12:20 amThat is surprising, as I could hardly play half the presets in the previous version, and with this version I can play all presets without any overloads.Milkman wrote: Fri Mar 06, 2026 10:16 pm Very consistently, I cannot run the latest version of Z3 in cubase 15 without constant ASIO internal overloads, meanwhile Z3's internal meter shows as low as 3%. This may be that E core/P core minimum CPU usage issue, but in Bitwig retail 5 or beta 6 I see no overloads or unmanageable spikes. Im pretty much only using Z3 in BWS now, as almost every preset does this. This did not happen in the earlier beta versions of Z3.
Ive got an i9 12900KF with 32GB and a focusrite 18i20.
What buffer size do you use in Cubase?
Surprising for me, also. The last few versions were mostly good in that regard in cubase or bitwig, but suddenly not.
-
eclipse_soundlab eclipse_soundlab https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=724571
- KVRist
- 79 posts since 6 Nov, 2024
high noise level from FX chorus/flanger
-
- u-he
- Topic Starter
- 30186 posts since 8 Aug, 2002 from Berlin
Only in analog modes. Switch to digital or use ModMatrix with Constant to turn noise level up or down.
-
- KVRist
- 109 posts since 28 May, 2013 from MUC
BUG REPORT
Public Beta (Rev 20977)
Module: LFO
Description:
The Slew parameter (waveform smoothing) is completely disabled when the Spice Type is set to Noise, regardless of the Spice value. This occurs even when the Spice knob is at 0.00, meaning no noise is being injected, yet the waveform smoothing provided by the Slew parameter is lost.
Expected Behavior:
The Slew limiter should remain active regardless of the Spice Type selection. If Spice (Noise) is set to 0.00, the waveform should remain smoothed by the Slew parameter. If Spice is increased, noise should be injected on top of the smoothed waveform.
Actual Behavior:
Selecting "Noise" as the Spice Type acts as a hard-bypass for the Slew limiter algorithm. The waveform immediately reverts to its raw, unsmoothed state (e.g., jagged linear segments in Rand Glide mode).
Steps to Reproduce:
1. Initialize a patch and add LFO 1 as a modulation source for any parameter to make its panel appear
2. Set the LFO 1 waveform to Rand Glide
3. Open the Scope and drag the LFO 1 source pip into it to visualize the waveform
4. Set Spice Type to Phase (or any mode other than Noise)
5. Turn the Slew knob to 100.00. Observe the Scope: the jagged random slopes become smooth curves
6. Ensure the Spice knob is set to 0.00
7. Change the Spice Type selector to Noise
8. Observe the Scope: the waveform immediately returns to jagged linear segments, as if Slew were set to 0.00, despite the knob still being at 100.00
Public Beta (Rev 20977)
Module: LFO
Description:
The Slew parameter (waveform smoothing) is completely disabled when the Spice Type is set to Noise, regardless of the Spice value. This occurs even when the Spice knob is at 0.00, meaning no noise is being injected, yet the waveform smoothing provided by the Slew parameter is lost.
Expected Behavior:
The Slew limiter should remain active regardless of the Spice Type selection. If Spice (Noise) is set to 0.00, the waveform should remain smoothed by the Slew parameter. If Spice is increased, noise should be injected on top of the smoothed waveform.
Actual Behavior:
Selecting "Noise" as the Spice Type acts as a hard-bypass for the Slew limiter algorithm. The waveform immediately reverts to its raw, unsmoothed state (e.g., jagged linear segments in Rand Glide mode).
Steps to Reproduce:
1. Initialize a patch and add LFO 1 as a modulation source for any parameter to make its panel appear
2. Set the LFO 1 waveform to Rand Glide
3. Open the Scope and drag the LFO 1 source pip into it to visualize the waveform
4. Set Spice Type to Phase (or any mode other than Noise)
5. Turn the Slew knob to 100.00. Observe the Scope: the jagged random slopes become smooth curves
6. Ensure the Spice knob is set to 0.00
7. Change the Spice Type selector to Noise
8. Observe the Scope: the waveform immediately returns to jagged linear segments, as if Slew were set to 0.00, despite the knob still being at 100.00
-
- KVRist
- 109 posts since 28 May, 2013 from MUC
Follow-up: After further testing, I’ve realized the LFO behavior isn't a bug. Sorry for the not a bug bug report! I tried to delete it.
-
- KVRer
- 14 posts since 11 Mar, 2026
Hi Guys, I don't know if this has been mentioned before but it doesn't appear with a quick search.
It seems that with some sounds which are set to use Legato or Mono, when rapid arpeggios are played (like as fast as my fingers will go) , there is a massive fluctuation in volume using MPE.
As an example, although it happens on multiple presets, playing Eight-Bit Conan in Poly mode:
If I play a single note on my controller it will peak at around -6dB
If I play full speed arpeggios it will show a peak of around -3dB
If I change the Global setting to Legato, arpeggios will peak at +2dB
If I change the Global setting to Mono, arpeggios are now peaking at +9dB
This is on a fully updated M3 MacBook Pro, using a Roli Piano M in MPE mode via Logic Pro.
The thought just came to me, so I moved the Piano M into standard single channel mode, and now:
The same arpeggios in both Mono and Legato will peak at -7db and play as expected, but Poly mode will peak at around -2dB.
If I switch to multi-channel mode (on the Piano M) then the results are much the same as single channel, give or take 1dB either way.
Many thanks for reading all that, and totally loving this synth!! <3
EDIT/p.s
I should also mention, using MPE mode (on the Piano M) cancels out the legato and mono functions in the sense that although I can hear greater definition between notes than in poly mode, I can still actually play chords - In fact, holding a C minor and then keeping the C and D# held down whilst playing an alternating F and G rapidly, will replicate the volume problem too.
My laymans guess therefore is it may be something to do with the signals from multiple MPE channels accumulating in some way to affect the volume.
It seems that with some sounds which are set to use Legato or Mono, when rapid arpeggios are played (like as fast as my fingers will go) , there is a massive fluctuation in volume using MPE.
As an example, although it happens on multiple presets, playing Eight-Bit Conan in Poly mode:
If I play a single note on my controller it will peak at around -6dB
If I play full speed arpeggios it will show a peak of around -3dB
If I change the Global setting to Legato, arpeggios will peak at +2dB
If I change the Global setting to Mono, arpeggios are now peaking at +9dB
This is on a fully updated M3 MacBook Pro, using a Roli Piano M in MPE mode via Logic Pro.
The thought just came to me, so I moved the Piano M into standard single channel mode, and now:
The same arpeggios in both Mono and Legato will peak at -7db and play as expected, but Poly mode will peak at around -2dB.
If I switch to multi-channel mode (on the Piano M) then the results are much the same as single channel, give or take 1dB either way.
Many thanks for reading all that, and totally loving this synth!! <3
EDIT/p.s
I should also mention, using MPE mode (on the Piano M) cancels out the legato and mono functions in the sense that although I can hear greater definition between notes than in poly mode, I can still actually play chords - In fact, holding a C minor and then keeping the C and D# held down whilst playing an alternating F and G rapidly, will replicate the volume problem too.
My laymans guess therefore is it may be something to do with the signals from multiple MPE channels accumulating in some way to affect the volume.
-
- KVRist
- 109 posts since 28 May, 2013 from MUC
Bug Report: Keyboard Shortcut for Deleting Curves Fails
Zebra 3
Revision: 20977
Operating System: Windows 11
Issue: The documented keyboard shortcut (Backspace) fails to remove a selected curve from the oscillator timeline.
(even with KEY CTRL active)
--------------------------------------------------------------------------------
Steps to Recreate
1. Open Zebra 3: Load a new instance of the plugin in your DAW.
2. Access Oscillator Editor: Add an Oscillator module to the grid, then click the large pencil icon in the oscillator's lower panel to open the Main Editor
3. Add a Second Curve: Double-click anywhere along the dotted Timeline at the top of the editor to add a new wave cycle (this will default to a falling sawtooth)
4. Select the Curve: Left-click the thumbnail of the curve you just created to ensure it is highlighted
5. Attempt Deletion: Press the Backspace key on your computer keyboard.
Observe that the curve thumbnail remains in the timeline despite the documented instruction that Backspace should remove it
Zebra 3
Revision: 20977
Operating System: Windows 11
Issue: The documented keyboard shortcut (Backspace) fails to remove a selected curve from the oscillator timeline.
(even with KEY CTRL active)
--------------------------------------------------------------------------------
Steps to Recreate
1. Open Zebra 3: Load a new instance of the plugin in your DAW.
2. Access Oscillator Editor: Add an Oscillator module to the grid, then click the large pencil icon in the oscillator's lower panel to open the Main Editor
3. Add a Second Curve: Double-click anywhere along the dotted Timeline at the top of the editor to add a new wave cycle (this will default to a falling sawtooth)
4. Select the Curve: Left-click the thumbnail of the curve you just created to ensure it is highlighted
5. Attempt Deletion: Press the Backspace key on your computer keyboard.
Observe that the curve thumbnail remains in the timeline despite the documented instruction that Backspace should remove it