strange behaviour with vst~ (and DSP)
anybody experienced a "nan" (not a number) coming from the vst object?
My vst~ has Kontakt plugged in…: how come can it send a not a number signal? never happened before…
What is it sending a nan out of? A signal outlet? A number outlet?
Where are you seeing the message? In a float/int box? The Max window?
I’ve run into nans in a float box when using Alex Harkers descriptors objects and ended up using a [split 0. 20000.] or something like that to pass only actual numbers.
ah, well, the nan is from a vst~ so it is a signal nan.
i see it probing the msp chord. and when i print it on max window as well…
btw, VERY important: as soon as i quit and restart the dsp it goes back to normality
I mean, where are you seeing the (nan) message?
coming out from the [vst~] object’s patchcord…
A signal outlet doesn’t display thing. What you plug it into does. What do you have vst~ plugged into?
Or can you post your patch?
first of all thanks for your interest… below is the scheme of the patch.
nan as said is coming out from the vst~ object (plugged to Kontakt), i see it through number~ and it is printed as well on the max window.
----------begin_max5_patcher---------- 2414.3oc6bs0jahiE9Y2+JTw9ZOcgDHtr67Rddqp+CjIUJrQ1MSvfG.2IYmJ 429pKfMBK.AFiwolToraDf7QemKR56bf+9oUFqS+FI2.7uAeDrZ0e+zpU7lX Mrp73UF6C91l3fb9kYjP9Z55+z3YwoJHeqf27APvwvnzp12llTjG8+HryAQu XV1bxw8QIwjBdWUqwziEUsBKa8PPwl2hR184LxlBg.BsgzdBfvrNDfM4GP+D 7ox6IJjKKT462fv5hRRvdtnX7grnf3pyH9UK99Ahn+MLN0S7ecRVINTBDz9J Jl7NIKOJMolrtxH3vgZMup1svPu+Lk2Q3mO0TThnI3olxHuGUc+9hF+wSUc+ YHPLn8sYe5XyOv1qFDP0d6hS27ERXMDdkQHY6P5izCjjyWO+Jj9P9JiRNjQx IIEAEkCf5+tAGiK9rZyA4yuMXCo0aVoBbkwtrnvzDlPHcmrlq94nlMXlXy97 rbyuhjfCJt4hzz30AYL0w5XhjZl5YDjDsOnfTDIjGj4o6KZ+grnjBo9hjDP6 i2x2jkFGK0Uhy7thyDRsD1P9ZTXwa795LZI4qV2FqtOqT6c46J6+FmFDtmjm Cf0OcK5sKbkgxm3R241boQ9baQtCsKpoCcSmZ+lxlZyh17sqN6Od5op+34oG IK.Pf4zBhntAQHxhagawMzsr6CFsuBXjYg+r3q4BPy.zgZJM95X.05mX8tMo wh.urnZ1HKLxmEPyB55X6V+ufMPu1Mpk5RKOaKnMqiPt1XHuGQPri+k8XG5Q SdbVrWe5Qr804Oz.rqMRfxg74x9DprEBiwvcDL0IZBxQDHAW8Ya3GTxMcS59 8DQ7aCCEC1ySdDGkzVbXtPwNuZTHO8X1lJMPkJDHKfgj7hnjSyk9wS9rMtt2 hBCkm5RrzhvCoTuyR4SS01PEaeMkZFBufDa6GWwFN2hcYiUKB0HO3cR3mo+L TusOGTTjEs9XgvMPZEx8sRNUKCrq1aYkeW8JGiSWGDWtL+SWiwSmQA9mB0mN 6KpHc2N5xEUsQG3f2niuHPlmS0T.suQGmV2My44nG3PoJRXy83wVpLX+ow31 rcqUtCgZS0o0FAUiOlsiOBbAB4nD1sK3wpisANgPyuCd8CuBhxAzKfJq.53. rMKcOfZcs..LqRnRrSHWbWHFZZQrV3KHLXyOoA0PsQX.zDC8TM1KWKbyEy0J dpFiz6dUsDCpv4P2trKcwcVPmW7s7rvzU7gfdcBpd8SGgp0fI8wnblmx3Rt7 sV.875OrD9FDWpESol6OXPjOMt8DzBjMh8BzAL6Xqda.RvL9pn4Ryk8OM5oC iQOM3Pcth8PI30xA14bCZ3TtTIBTr6FGdXcKA+IVroEGBQf82G+CQf+ZRDn7 xXjcSo1ueGrN6XQPLH+64Ej8fhTPHofZG.RBR.AIg.pVNDD.9XdkO8mXWT9W indKfzsa4WTZBn3MBHL+P8eH1Fe1jdT.PN0IvPikGMQDp0COBPA43HwDulbO CO+1IRvpGdXlPpSjlNeBIh0RDvDgaYt8KVPCpWZBmTJitUiaD1dXia3uFiaH 1eXiay4cb2AevgjXfko4n3XGMZ6DOTM75x8bdA2h3qfa10AI6lU3DM2vIzGO L3z5QBNgyMb56OHzz6QAL2QWgE.NtbRNdayR7iO6Ot2ro43tXxJIKAtA6HyL ZIBLh3bx4HXlyqijV48HjD27qImivaQNGMuM4brLxgOOxgi4Tlxw4MmhcnLS SHrnVScXDjVHqHmtN8FSFCu1fxOOyAm46xEjGsKQV5t8aOiClVkHqeuqcvb9 1dVWvUvlu.hA+FX6X.KqwFeVTrbPDGyvN8AVVtOBgmO.37tbk0CxXvQStQms Uu6WZ5pMiFTPJCYsREYazQ131UQKYKTS1F8jRTTpjlRagCqeIECMJIC0TU1F ck80WZSYoFzVpC0k8ReolTX1AMl5QkYOzY1Ckl8RqYaTa1M8lcSwY2zbplpy l1uMilbw46KphF0AYOQX5bpstizz5Tb0x7u55hr0BdVufN8F.WJHtbf7qUEb Q4eMT.zTC.rLdsqc60BVWkWZ6kD1MDX5w1rn4rd2K6RQthrM0wtzaBLKEQi9 z8WCPCiSxhS+5BPG3zwB5ZpCbeXBMzmGPZ922uNM99h+BB1f1h7y6qiFv4WE MPHcwAch9vWv9Xnm63UAVZnBDZ.QB5P9ZoBPSgNHNJmW+6UeKm2i6mZYOoHK sIuyiS4fll3S1hhKvSq0tXMAplFbenPWHuT4Kph6KUQpJTWEJpKJXWml0gaG EsquxqUYc61cs6NHiuwNxXqxT6gFSs9.M1XAHjqy5NFaNOTCMuAn0POVCM2A Lz7dr70rFvPy8FNzpcx577z2iDvEDUoC4HswvRemqCRUlLxY53QFPF0Vx08g XkSNH8J6CW3horOtK4Bt5IvP2bA+PjbSE6h51+j9BkplNbuYsww9gHQD7bf8 5GT8uWm0jhIrTKSJVYVG67g.1dQjUrbRbYckNyOxzvZoEyA2aZwblfryNeUu vABILNZ+7WWWdb+ZwCXjW+Yl0Y43ieb+ZR1Oq+KtOMjzDLt4UMff1ZaG8xV6 0Ty.UItGXrMNMnPNmizSVlLlIDi4viwsJyrP8dnusr0AUjt66xSFtk1OrxKp mUYWnlhsi6hRtUHNODOa3VtCAugKF4F5nobyGfnkichmlxMdY4Wh00ujUeli wNAZ58Blpor8D4c0TwQkY0F4c5n6sVxdY8d0PauBmQJ29bnWTsmPKypCtQCG j8PbxWPSFnsb67fN4q23jajCyhA5yWqkO9zA2J2AcewmrrTCVNCQM.e7liva QA2ZOi7HiYVMylX+r0O5VY1aciGQHS64cDoskkhg9fzQXH+qxwG6n69HBOIi He3sdDoPPmXqtxwvrY0oafAjtaoq82ZLM4nYlVtntAJPvG0QntFkrmG86vH7 edKoMIuZXdOu3mzM4i.+gw+kJMAeo.X+GFJeOwXp98DCR46IF2tdQNY8Bqdo 78N8dwR7d2r9CunT8iYMhW0NmIY87e0r9wJa57akGlQT48+udkcJF7vuqzfv iIQ+0QBUtnFrzvm7+eFzXW0+Q9sqDuakYlTnoZ5KVpu5olezodXtz0SsyzEF O5JRNZHQHuYUjvXMjI77BSHKMjIqAhSHD5EmRGGHjuPKODemfrClCIdl0rbY BpiLAG.JZ4bJ7iGeW01H3Kvx2rbrVZ3D+zOd5+eFEHvG -----------end_max5_patcher-----------
That makes sense now.
So why do you need a number~ from your vst~ output? And are you only seeing it when you use the right outlet of number~? (have you tried snapshot~ instead?)
So the nan you’re getting (I didn’t see any in my system, but I didn’t load a vst) is happening in the max world (and no the msp world).
Whenever I need a non-signal float from the signal world, I use snapshot~, but that may just be habit.
No idea why it would only happen the once though.
well, yes, you know already that it is coming from the plug-in, so you know there is nothing you can do.
as long as your IO device driver doesnt chrash from them, nans aren´t dangerous. :)
i am not sure if thre is a msp external which can filter nans out, but there are vst plug-ins which can do it.
thank you for all your replies…
@ Rodrigo: the number is for displaying reasons only. in this way I can see the "nan" and use some ways to trigger action accordingly.
"So the nan you’re getting (I didn’t see any in my system, but I didn’t load a vst) is happening in the max world (and no the msp world)"
well, to me it looks like it is happening in the MSP (vst) world, the max objet i use as said is for displaying reasons only. I could have used snapshot as well, as you suggest…
"well, yes, you know already that it is coming from the plug-in, so you know there is nothing you can do.
as long as your IO device driver doesnt chrash from them, nans aren´t dangerous. :)"
you are right, but the DSP stops working at all! so i unless i manually send a 0 and a 1 DSP just stops working… no sounds…
@Andrzej thanks, i tried with that, but the problem is I want to eradicate the problem at the origin. Once i get nans they will never stop unless as said i close and reopen the dsp. By using bitsafe~ I only transform nans to 0s, to i lose the only thing i have to detect them and in case do some trick accordingly… ;-)
BTW, nan is such a new thing to me, and looks like i have to figure out why the hell they are coming from my VST.
Here are some thoughs…
A) I did not upgrade or modify anything in Kontakt. So it should not be a bug or something I guess. It has been working flawlessy for years
B) I have changed something in my max/msp world, so I guess this is responsible for that, in a way or another. This points to point…
…D) is it possible that some MIDI informations sent from Max to the VST are directly responsible for a nan output from the vst as well?
This i guess is the point I may need to look at… unless…
..E) by changing some instrument variable inside Kontakt itself (like gain, filters etc…) that sampler is completely responsible for the nan output. Even though i doubt that because I even tried to send super-distorted signals for hours from it, without getting any crash or nan in the VST…
So… a long story short, does anybody of you guys have any idea? ;-)
Thanks A LOT!
I have run into this issue with Omnisphere and also just found out about bitsafe~
My question is, how expensive an operation is bitsafe~?
I’m tempted to use it not only after VST~ objects, but also following adc~ objects and (for absolute safety) just before dac~ objects. I had a very unpleasant experience (extremely LOUD bang in a very powerful PA system) due to NANs a few weeks ago.
Actually they are very dangerous to your hearing —- when Omnisphere produced (what I later discovered was) NAN values, the very powerful PA system to which my rig was connected produced an incredibly loud transient bang. Trust me, you don’t want that happening more than once, ever!
nans aren´t dangerous. :)
Raja — thanks for the info — I have bitsafe~ objects all over the place now in my Max 5 system and I have seen no detrimental effects. This is totally defensive as I have never actually run into a problem with NAN on my Max 5 system but you never know.