Changing HISE/hi_core/hi_modules/effects/fx/CurveEq.h, line 318:

sendOtherChangeMessage(dispatch::library::ProcessorChangeEvent::Preset);

to

sendOtherChangeMessage(dispatch::library::ProcessorChangeEvent::Custom);

Seems to resolve the issue. Not sure if it's a typo or if there's a missing definition for the Preset event, but I haven't found any new problems since making the change.

@d-healey I found the cause. The new HISE_BACKEND_AS_FX is enabled by default in the Projucer project (in hi_core). Switching it off resolves the issue.

It only took me compiling all release and debug builds, watching your version control video, deleting everything and converting to git to see the answer sitting right there in the commit log :grinning_face_with_sweat:

I've been digging around the ModulationMatrixTutorial snippet, and am curious about a particular use case.

Let's say we have a dual-synth setup, with a filter on each synth. A single UI slider should control the frequency of both filters, and (ideally) the slider's right-click menu should assign a modulator to both filters. We can't use a control callback for the slider, as it won't work with addModulatorTarget.

What if we duplicate the addModulatorTarget, changing the Target but keeping the same component and ID (so it shows up as a single entity in the matrix view)?

matrix.addModulatorTarget({
	"Target": "OSC 1 Filter",
	"Chain": matrix.Frequency,
	"ID": "OSC 1 Frequency",
	"Component": "Knob5",
	"Slots": [2, 2, 1]
});

matrix.addModulatorTarget({
	"Target": "OSC 2 Filter",
	"Chain": matrix.Frequency,
	"ID": "OSC 1 Frequency",
	"Component": "Knob5",
	"Slots": [2, 2, 1]
});

Chaos ensues. The slider affects Filter 2 only. Modulation activates for Filter 1 only. getModValue is lost in the slider's LAF. The mod matrix shows 2 instances of the same ID. The right-click menu doesn't reflect the currently active modulators.

I'm exploring workarounds in Broadcasters or Macros, but so far am unable to produce a solution that retains the UE of the ModMatrix system + allows (sample-accurate) modulation of multiple targets from a single component. Am I missing something obvious here?

@FatMitchell Yeah scriptnode can be daunting. I wish there was a single, comprehensive resource I could point to; the best I've found has been a combination of the docs and searching the forum internally + via google with the site:forum.hise.audio operator. The snippet browser's been great for learning structure.

We can actually replicate this in HISE's module browser faster than scriptnode, so let's start there.

Roar's waveform makes it look like some kind of modulated wavetable distortion. But running a sine wave through sounds like what Ableton describes: the signal + some white noise driven into a basic saturator. Notice the way the curve goes from smooth to almost a square shape?

That's the same waveshaping you get from ShapeFX's saturation mode. You can see the curve change in the SNEX version:

In this snippet, I've got a container with two noise generators panned in either direction (waveform generator should work too) and an instrument. The container has a ShapeFX on saturation mode, and the noise generators have filters linked to macro knobs. The single UI slider controls the distortion amount + the noise volume. The limiter at the end is just to wrangle the volume.

HiseSnippet 1978.3oc6a8zababEen1k0UTNFwtFAo.EEDB4vpBWGsQ1NtHHPzRqVKgHYuVqx+PQgyHtyt6DycFVxgxdcfA7gdn8RABPQuTTf9In.sW5w5BzO.snWCPa9BT3a8XyaFRtbH0p0RJRNYsoNHfYdyaleu+Nu4IwVAbWRXHO.YL6NC8IHiyZ1dHSze09XJCsQCjwELuEmFRraPCE7.AkyPqLzGGFR5fLLpbS45LlsJR8ySWdErGl4RxlBg9.N0krIc.UjMaKm2i540D2grCcf1puhyFtb1pbOdDfoJlKh7wt2C2ibKrbYyXhL9Nq0gBHos.KHgHipqv6Lrce98Ywq+Cngzc8HxA0QsgMJd5lbuNRDKmEsZepWmVoxdHBYX1JSSTIVSbQysncnilOSi7pJB1YbnqOLlIO7pjCd00g2hZvaLPxPCRUigz4Ma6FP8EYTj3YNyMXBRPWLn10gR7ZQybtJlqxgUvDWd.9djlAvfQbT6ZKt3krgesv6XYAp9Pg8d3.618w9jleTc620V4Ob4dDwZc6RbE0lWQyFHNeddT9IR+gPfKqep09XbSRWgsbAyuvk1O0so85mR15mA6rEk4QYD6tQLWoemMm8dL9t0kBS.2qlKefOmAh0kfi2Khrf0mYMq0a9lgDgsnOwd.9A1c10tKOPMLQjr6AmfceR.wxi6h8jKqQ.cOB.5qsHbrG3FDGGvCcAOWLXhCU6xrYaS7BdW60X8.bKkMozzD6BKtIOnAwktKwKr1O9sthTykpiuLbd2PHBn6FIH0FMqj2DI6GkhQIaR3TiYCBQlBGDcV9sgEyeFTROc.HYnp1nsWsWKrv67HKqGAGRpCCv3pop4Zyqvv7KHOoDivpXOucgHzZ4sM.N0rZ2hKH2lUK197HK6hj51crzR1KORvXIKSbDLIFqwhFrKIHm2ArPHZJeHp4AGhpmAwMVknsPNaCFUbaeB6fxqfRzixv4DTAKUnBlOWRvbaOZGR.hBwrmwTogQJ.mjT7+7vK9eWF89az.KvoaCrivo3SfzwRAvnAYOHEabNhYMaPBumf6qVahoCj4m4g9fQYgeL1Y3nANepCZ+Yhfbf7NQpv.8DiROqDBEunPlwgEREC0usnX1xpGtrkSNY9gEtm2rEU31e73clwfWP8eZi2j6ddEy3rhYfspYyOZhWzbhCku6ATPvYU2l.iHA0OIJEv+PWJfQkSiq9Oyo0U+OWclKi89FL163Wj2LZm+4SSfmTBPRwciJ2JGDVMJXu3BhWghCkUVoqqjyopmRex0gIZAafdH1l76mape4+xArIZQgutys2iDDhG3C0h0S2+QKb2Yok1zYaRmn7Q62HRv6UHofJ3eClejPe13afzXcK5CxkJQI4ModPUqg4bJpbnsD0mnk.sMOR.R3VXnBJ3rMuUzf1PlGWB3GvXPgSvbFyHuROd7hxwRHzlv5nF7+geRHVWN1HgX8Th5NaWH1XaY1XHCOf5FlXsSGVL0pfrFCCRRGcoWN8N8CHg8Am7hDfZAgByxawDjsIdDbnlt9icjUIDnf0XNhLhi8fxHusLLV+zxHUDIug1QNF7nbQH6GLINOjwgjZOIgVtyx4Ke3uIkohGzSBzPwVv6hh70ErDt1OgoVOtyE6wAITfqtU2oqb4dM0X6ODCu.5lD3RcYh5C7h8GO768ON4uX+1tB332I.yB84g4131jAzcfZXyE0+9gvqXI+7BtbRuQbvXI8zkaBB4X4oMVDEntB5FC3QLwjr0UOb15mQAGUKK33Dr0Hylhw1T3NJHK4dDO3oQI8tpAoKNxSjNad+5s3LteeNi5paz2l.wi85QBzw9XEnh40tny9xl8FP1LFAGj+R0ijtn9QtMQi0d85lwv0VFLZ+hQghUd47QZOmu3H4gfmItQvIuALtWWp5NORWU7z+5o4a.KSIO0kR9sJSI+hTJ4Whd6dk86xeAyVbugwNvwuVUAhYM2r0DyUp9QVdZDg4lp4d7Rewm6bGcydwmlemHrWll9S3e9+9usB0m6gCNTVgiuGd0CoqxqYNRnll7ukVloG798MSL51Rk8zCtOm4c91EXeVAwqeLBhc9e+t0lTP7OnPPLBc8+bYPbYPbYP7W6+rUphOuYJPmybz+9AGX.7u8Wc8mzf3gyEv8gzNh958mbLs+ZCnX1.QKo8qX3Zkm2gqSad9xWD.p7oI.qbIld.7qZ1B+sAGhmG8s3a9di7JI8FI9u+1Qq4HKW1bjxlin0bjkJaNRYyQdAu4HVxliLwjkmHcGAg9S+cmxGVU9vpxGVcZzcDKY2QN5QwG01i7Gt9uurGmkQwkQwmzsG4rlYe+Ek8Gor+Hk8Gor+HureFCvtA765F+c6HSwXolAreL0WiXU3kK49VcfWkrky.ZG5cccS2EiYbS+Nl5bWeb.vorlIi4je6MY+gg0nDCkQSb23CatrxLPCzKp5JNCvOP+efaO980o1GxqqSN7dDE8e3u9ubU7b+ykCED0+fqe1E+i27Ke3uXYpJWcrye.A2gy7FJe6JZBxg1a3lBDjhVw0KZEQW8HZEW+EDq35SsVwYM2RN1dohlRDpfo7.X7JGWFu5wkwqcbY7sOtLd8iKi+jiJiVILVewmMmxNFK+TQFDe4KBsUq0TesiFFi9F.Lpf9J.K3zuWB

Since this technique relies on adding frequencies to the signal (via noise) before driving all of it into a saturator, the range and amplitude of those frequencies makes all the difference. Mess with the filters, and try tweaking the max values in the slider callback to see how much the shape of the sound changes.

Faust can handle it, but I'm not aware of any demo or resource covering this kind of routing. It has noise generators, envelope followers and oscillators, but you'd need to script the signal chain and roll your own saturator (or just use a SNEX shaper after the Faust node). I'm still green when it comes to Faust and the topology, but I'd recommend checking out Faust projects on GitHub to see how similar effects are structured.

@FatMitchell What's your path looking like so far? I got pretty close in Live with a white noise layer (eq'd/ bandpassed to taste) + a soft/hard clipper, with the noise volume & clipper shape mapped to one knob.

You can do it in scriptnode with a noise osc layer + soft/hard clipper (the saturation snippet from the new browser should work for this).

Ableton's manual mentions

Noise Injection applies stereo noise followed by smooth distortion to the input signal.

I think a container.multi with a noise osc for each channel, followed by the saturation should reproduce the stereo effect. If you stick with a noise sample just make sure to add a short delay to one of them.

Re: Erosion, based on Ableton's description it sounds like a whole different beast, but probably doable in scriptnode.

The Erosion effect degrades the input signal by modulating a short delay with filtered noise or a sine wave. This adds noisy artifacts or aliasing/downsampling-like distortions that sound very ”digital.”

Idk how dependent this technique is on the delay algorithm, but scriptnode and faust have plenty of options, and they can all be modulated by an oscillator set to sine or noise.

The Frequency control determines the color, or quality, of the distortion. If the Mode control is set to Noise, this works in conjunction with the Width control, which defines the noise bandwidth. Lower values lead to more selective distortion frequencies, while higher values affect the entire input signal. Width has no effect in Sine Mode.

Add a bandpass filter and when the sine osc is active, disable the user's width/resonance control and set a static value.

Noise and Sine use a single modulation generator. However, Wide Noise has independent noise generators for the left and right channels, which creates a subtle stereo enhancement.

So two noise oscillators in a container.multi = Wide Noise mode.

@DabDab Because selling Software doesn’t involve the paper trails and delivery confirmations you get with selling physical goods, you’ll want to collect as much information as possible to support 2 points:

1. Your product is clearly and accurately described, as is your return policy (links to site/documentation, screenshots of the product, show that the refund process is reasonably accessible, etc)
2. If the customer experienced an issue, they did not seek customer support/refund (this is where you’d show that they purchased and downloaded from a specific IP. If there was any communication to/from the customer, report it. If not, report that there was no communication)

If you have a demo version available for your plugin, consider logging IPs for those downloads. If you can connect that to the bad-faith purchase, that's evidence that the customer used the software before deciding to purchase it (which conflicts with the "not as described" claim)

If you don’t have a demo version, consider offering one. Some people see a lack of demo as justification for acquiring the product illicitly (to "try it out"), and that can include filing a chargeback. Offering a demo also provides you with one more point of evidence that you are not misrepresenting/obscuring your product’s capabilities.

Time is a big factor in addressing these disputes, so it’s a good idea to have a template stating your case so you can submit your info quickly. For example, all of the info about your product will be the same across cases, so you can have that ready to copy and paste at a moment’s notice.

Also, it’s important to make it clear throughout the process that you have been and are still willing to issue a refund. I’ve had holds lifted in the past to allow me to return the payment myself.

@DabDab IME there doesn’t seem to be a bulletproof solution for fraudulent PayPal disputes. In the past I’ve been able to reduce the frequency of them by:

• using a platform that focuses on digital goods. In my case, that meant migrating from woocommerce to easy digital downloads. Besides eliminating all of the variables associated with physical goods (item size/weight/shipping/inventory), it allowed for a streamlined way to log IP/location at time-of-purchase as well as download. That info alone was enough to swing some cases my way.

• applying a “no questions asked” refund policy and making sure all potential customers are aware of the policy. Of course, you’ll get some bad-faith purchase/return loops, but that’s less costly than chargebacks in multiple ways.

I tend to see chargebacks as falling into 3 types

• legit customers that turn to disputes to avoid the communication involved in a normal refund (sometimes not realizing the damage done by chargebacks, sometimes simply not caring). I’ve seen users turn to chargebacks if their support ticket isn’t answered within 15 minutes.

• “Customers” buying with the sole intent of seeking a refund/chargeback for free product

• “Customers” actually using stolen card info to purchase the product (often with VPN/proxy/false info, I traced more than a few of these to folks sending the software to crack teams)

The first type can be largely mitigated by good documentation/extra friendly customer support/heavy emphasis on easy refunds for unsatisfied users. Thorough purchase/download records (IP logging, etc) can control the second type pretty well. The third type is a hopeless case; best course of action seems to be using a platform/processor that flags/refuses suspicious payments.

Some sellers may let the first type slide/try to win them over, but types 2 and 3 should get an instant, site-wide block. They will 100% pull the same trick again if given the opportunity.

Unfortunately PayPal seems to side with customers if they press the issue enough, regardless of the evidence against them. It ultimately comes down to whether they’re going to have the time/gumption to complain their way up the customer service chain until someone folds.

Can't edit my previous post but confirmed on my end as well. x86_64 + ARM64 are in the plugin build compiled directly from HISE. No additional steps required.

Occurring on M4 running Sequoia 15.2, Xcode 16.2, current develop build (happens on older builds, too).

HiseSnippet 850.3ocsU0sSTDCEtEnnr9STiO.S7JzXL6BqKPTCHKftQA23fF8JRYZG1lsS6jocPVMdiwmGeE7Yw2.uv60S6LvrqrB5l3bwjb94qmud9qcyzQbiQmgv01cPJGguLIbfx1qcOpPg5rABeUx1TikmETnZ8AoTigyPX7zOwo.O2LH+22Wccpjph3UpPnWqEQ7mKRD1JscW6YBobKJiuqHYHuatVmHspsVpyA9LMoNJkF0md.eGpysoHnmRM8P36Ppe+Vb9RKFyXzEWj0pUqUVdklqzfFWuIc+FsVo4BMWZgkWJFgmcSlvpyBsTK2fvyrtlMHrm9cph.7ZgQruj6DZfBgHWndKsj4thNsn18DRV2iSTFDbJcqRaSWj1tIYaASbh9pz207FBpPLbBDO0nza5QnWi+D8FCkvCQoYJnz0IgQYhTakEGetDoiBplwTnNMLUJ7Eg+AosFbPYuWBsOeqLP3D.y2pd86F.+t8CpUCJUFavgzrfmoz62H3QAGC7.tssNIUq.g4uk25sbHhyUQVgVEnU6ns7Wnl+109Ps4p8wZA+to33wZyEgLsTxyFqYW6T1YAbdUdx97r6BzVlyOwQHkMZcX1+t5PTwEdHG0pNJg8Eo7R4ytQBUlwb0uRFBtZ8UuqVV8BkBFOCIfC4BDetD4Ie0XyQOD8pNaPsziOF3DgnjxyrB2kAuA+PXHrnoXNxFbSeqN06aYQBgux4FzipBnXsAUBucMicfK0Lq2QThfwj7tZivk5GdU.rDIMT7dOy+vM+xS916+7pIir.4ql73XwQ9Q8XgLwXAxzIA1.fv2f7JCOfwio4RafoO.TJT84rc0d+emfYgcCD7mfSpGWbPOqS5g9ENmZ1.lJ0rbI0N5npagVoAG4Gd9vMDnfK0fgW38OL+V+Lme+ao30IcE1ndimiSMFNB8D+O3X4VuqP1LNlGYqH3Ljsdyjth6bB+K04Vg5fsoPWAziP1IOIDdoHhCQWo3RiqdOkappPttS1kAB4JlW3mvWowFNYbowFGaDkPixz6EULK51qdQuFfSJ+aPyAOFBxAmZHD4Z7E6EEM5QcJfKLo.WbRA1bRAd+IEXqIE3RSJvkOeftWgebtUmTL1fPa2cS+RSLdSEE5.8cqneAC18cCB

2. double click the slider to reset the value to default. While the UI value reads 0, the actual component value is off, like a floating point value is getting lost in translation.

It doesn't occur at a stepsize of 1, only .01 and .001.

Not seeing this problem on an Intel machine running the same HISE version. Is anyone else encountering this on Apple Silicon, or did I just forget to set something in the native ARM build?

Sorry to dig this one up. The Effect Slot issue was resolved quietly months ago, and I incorrectly assumed the fix carried over to HardcodedMasterFX.

HiseSnippet 2420.3oc4Y0sbaabEFT1HMDwo0YlzK5cq4EYHiooIonjrrqmJJRPKNg+MjT1ViphmU.KIVIP.V.PKooQuQ8gnOR8Mn8b1Ef.fhR9mIIWTCMiD1cOmc+N+tGbz.OWCluuqmRFswWMmoj4ApitxIvpgEk6nztoRl+hpE0yzv0jYRlbIw21MvmDv7CT1+p4TeeloRlL26UH4YxdeEwy+4usO0l5XvhmRQ40tbCVG9LdP7rC16m311snlrw7YInt1dsMbcZ3Z6t.f18TKqLmZbNcJqGEIaCUkCn9VJY9Q0xasMisylSLMoato41au8tOa2Z6VgNobM5oU1d2ZUqsS0msyDkLektIOv0aT.EPuRl6uuq4UirbuvQd.ul6yO0lgCpnLBNY4zsbsMQQDmUogE21bPjNyWA1kAwZv6I0feuZWtIe47wZxGJVfDyQREXlMRCu6kBdURBuxIf2ZfTlDP59RH8cpiL73yChWAwy2n11If4MgB1ojPQRqxFa7P0Ft.ENAklQOm0xCFrji7aWtbQxVkKW3EZZO8oY6UuqNoeKx3CzIsdKYbeR6dizGNljWu6fwGQdIomKrPgrOYcOZf01Of7dpGx7KIGqkSe17fqxUTKWiEdumo+OvWGwmM2lMj8dl2owiQeObTSlMUvgvIi4ImapG3cI1GKWuE93aCrn9rVuUrCzfEdTPuKH9JG5LtgflQVz4HMZmHkO.Vi5ze7HxP8V5C060PO6sIAiDgHnPHBkJMkEfS05s4yoOYByHPPQkbEJdmDT8CQvleHBp8gHXKjfPAbP8d5cP6Vy1iFzo9QQVxF86bX2d2zrkEdHhekx3Mf5vrAgOxyAN3Ftyl65.CxmKb8bnSCdnidS8A3gz5vdMF2teu04cnwcr4NLxjENFAbWGh+Ez444lWVoHA9c0BZ+Ssr1tFTaH+TE3nE5+iQB.Qa4JUStRUbEsr.8kD6FrbAwLg.rjGaN3UEjGl85kN.RUzf5s6MlLr+gia2KkSfVDy9rfAH2CcWD.HOeDxyOUf0okl.I8paamOvh6iZn74NcpLYWtBHNlhaQKWTiYxlPWXGjqHoRMbITKSAQIh0Nn3su6BGS+7kkx.RBllFnhd7lmPdpTtKYyblFXsjjIWJbRQMApJb8H48IbGIwEzxBHMjNfLwYAQhdfYT5.gpFIAbSffIWhTz1LdZHVhAKvMK4u3T+.OtyTvpUh6Xxtr+j74dWtBjGSDFwPnsj0SCyFI2WteT1I4lGRj6omQdI7Nhyr4lbYtmuVTVLbYDNqkDDyRhBIA+ibhHb.SF8JL+0gf.Tfkluv2JOfDDXWGoYo1ysPaT4RaIz1nittNFaM9f5iEAVMGV+UuRuYXNk18Zp+VRaL4B71PX9N0GMVxHjHcLodm2T+nQj2ztSGx9GQdc6Qs2uit7zL.ie.0yKxXFZKOCAvKHmQ9qoM+jyd7iAqaV9DjlGE5IYRCnkf.CbkrQ6nT5NS5UkdxzL8BDpQROOFCmKwv42.CmKv.Z43.IQa8wmKbFIgnCzuGyOozRmge4WHgSIcsdII7RBAniU56faBy1mkb5JkJGZ1DAWxnsDAfvEFyRFBlMDDoDTTcwKD4yIiiGhwdLygfmD0YpMK+wUKRpBd1hPvejvKBQgUOg7DRshx4f2NoHYSwojBMx+3W5BdfUcD2wyXA3qnzwpPDildzKpaym5vLGytLHReIc0KRNtbwUvfbL7mbFf2uG3WK1pqkRanF6SPzRqadbhw.zldzuqBdJrH8PRoB9nv5sqftV65BqbOP29GNRmzndmN6WuwOs5MWIuLnq6BeVCps8oPQrw2F.kw3DHtQ.CarRjT+.FepkLCahj4VqKSN5iJ1nRVtPYQje3GHORN1i4CWbfdqwxpjlWR5RgZBlX655Ex7UvdiGSgU1zvMAyD0oe8lIuqF1X7Vp2leksuHQxJFQglmU2RCatw4xvY4DnAHJavpTIBsSa69.nOapsIbuStGB6MrKwDbTHaBZLr.e8vqZtNMxmg1wCmi5XteS1D3tcyUxDhqciTFOoRn3kDDtymKtf6tEFQ0Io1vhjarKBJWUI8jJuX8hTji7HVfrJsKCV.UHMi4rH6cW2EFuFQedHeM8pD0dglaYUuYy8DxJ+.UTmMmqyScmLQ7pGaF3p.g01hgqmd7MtiOyKHZpDWugGGVsRHLjdOHDj2OguJE23Z57w77ycmuXdWPVaiDCkUIY5LWtS9b+cGQxerhuzBK7gKRkVBEVjdY8eQyspBg.FgUTfPT84CY8LrjRTnfX.eeBYMZkmmZsD5Fbgrm5wnm+hkjDptEKIMQIJqZYkvmrlRltQEWYuaxAoZI8OJ5PjLtBlRX2E35lUcBOfCqMi5kes6f7F+muBn9PL.eH2GEKfE9.n9KxazIhzcXoZ0aL9v5cvLePc+8i9rHrvMfgvxyWcyAMRn1I1Q71+5Bsk9GtN8bCX8cxK7LztVir5RSlr10P+ROWaal2ZWF6th2cwXdmEyNEyc+dp8B1RBUxb+zskP81aKQxtlXH+5uDD55z1gGzeNy415khR3mLBucX6lPhLrWFgyAzMG704HDxzj8dtAS1YirpMY9mG3NWPa32Ypj4AAhU+1n9dfZcEtIRenMP4xkca5eau2UwsdZ7dWvMCrhm3m2yRbcbL8d6g4.h5N0CTqUc2Z6t8NU2cKEvy18hnq5Av9mTguyiDOQ5zOJY9Wp2HJWSF3pEGqncSZRG+qIbx0B6ThVx1jnE2iDMQCRzB6NhVXqQzj8EQKroHZK6HhVT6PzB6EhRbMxRo9Ykqt41aWsrRzWuJmdmmUcypU2sbB5qFsRkcKW9YkUTtYOpdnZWWyE13ImrkYHvCW.bXS0mJrWTN97fqR13we05i1GKD+N0AX960iwMVCFA25eKvXX2G+VUYhmX.devk+2nVMpdSczeV8fksKtK0GbzfrkBb7GUS18qTHpGK3BWuyEM1My89nQWk6V4Ha+xTn.KONDsq1awrQfunACzMNPJ.LcVlMvDOxwkwwn8YDywTL3+BOgKVAGmIbwJQKp7Yq.p9ktBXyuzU.09RWAr0+mn.983LlQM7bemgrXML06WKlAjaGw+Qr+fZmV8IUTDkuEWox61aF2j+NCizazJrU8yisM+7Xq1mJappuVuymJSekZu9i0WgK34t45qUGVuWy9c+TOsuQsa+lj2bf9M.5ZNR7+BX8EAtyjkKnnzcftnd2LYzcnfesHdP4+UVDtQA

The issues seem to have been on the develop branch since November/December.

1. Loading an effect slot and setting an effect from the module browser popup = instant crash. This doesn't happen with hardcoded slots.

2. The "swap" function breaks parameter values of swapped effects, instead of restoring them. This snippet can be used to recreate the issue, but it also affects hardcoded fx.

Just want to confirm it's not a rogue file or something on my end before submitting an issue on GitHub

@mehmethand

HISE/hi_components/midi_overlays/MidiOverlayFactory.h

Line 40:

class MidiOverlayFactory: public DeletedAtShutdown

gets changed to

class MidiOverlayFactory

@Soundavid Can confirm this worked for me in the plugin. Also tried recompiling HISE with the change, and it's more stable in handling the project now :folded_hands:

Thoroughly crash tested the midi note display and drag&drop, all good. The other MidiOverlayPanels seem to work smoothly from cursory checks, as well.

This bug's been present for a while, but will likely become more noticeable as the new pitch_shift node is implemented.

@clevername27 Sorry I missed that bit, sounds like it could be a missing file pool issue in that case. Have you tried Export > Export Pooled Files To Binary Resource before compiling?

@clevername27 said in Where are MIDI Files Stored in Compiled Plugins?:

midiPlayer.setFile("{PROJECT_FOLDER}" + [MIDI File Name] + ".mid", true, true);

Have you checked the result of this against the respective file/value from midiPlayer.getMidiFileList()? I went through something similar recently and it was from a misplaced text.replace on the [MIDI File Name] variable.