rob beattie

help needed on this, is it possible to have a toggle in a max application that can switch the assign midi map on/off, without running the max environment, kind of as standalone.

help needed on this, is it possible to have a toggle in a max application that can switch the assign midi map on/off, without running the ma

assign-midi-map-button-within-standalone-application

Hi Rob, have you solved this issue? Anyone here? Is there any key command to enable Max 8 Midi mapping functionality within Standalone and to save midi maps?

Found this https://cycling74.com/forums/max-8-standalone-applications-including-'assign-midi-key-map'-function 

Hello Ruslan, yeah I did, there is an object called [mappings], I just looked into the help file and it was pretty easy at the end of the day, but thanks for the link, I'll have a look, many thanks, regards.

assign midi map button within standalone application?

Shahab

First and foremost you guys rock, I've been using/learning from the community for a long time! Been working a lot with 3D cameras of various type in Max and recently got a hold of the newcomer Intel RealSense F200 and R200, unfortunately they are Windows based and I converted to Mac a while ago, still these two cameras forced me back to my PC and as it turned out at least the F200 is easy to use and comes with many end-user software to do most of the things that we've been doing with Kinect and Asus. OSCing the data from my PC to the Mac was my only working option so far and was wondering if anybody is working on a wrapper for Max, if so I will be gladly running tests. (hint: maybe Jean-Marc Pelletier or Dale Phurrough are thinking about another magical object by the name of jit.realsense?) Here is a single webpage with the drivers and SDK:

https://software.intel.com/en-us/realsense/devkit

intel-realsense-development

I briefly considered writing an object for it.  I did research into the number of people using it. So far, I haven't read about any significant number of people using it to justify the time/effort I would need to put into a quality release.
If you know instead that there is a large number of people using it, please do share with me so I can re-evaluate my previous choice.


Thanks for you reply Dale, the number of people using the tech is definitely not even remotely close to Kinect users, that is expected though knowing that the F200 recently became available to purchase (a note to friends reading this, please pay attention to the specified range of this camera it's super short range, not usable as is for performances), having said that with the publicity behind it such as DevLabs all around here in the US, also the fact that they are making laptops with the built-in F200 and tablets with built-in R200 (the long range camera) it seems it might become the only choice for a portable better resolution USB powered 3D camera. In either case I'll be happy to help in anyway necessary if you decided to work on it.


It's rather sad to see that Cycling '74 team does not take care of these things. I recently took a closer look at TouchDesigner and I saw that they have objects that come with the package for every device possible, this one included.

Here in Max we need outside people like Dale to take from their own time to build these objects. 

That is not very good. Not good for the company. And definitely not good for the users.


@YGREQ I think it's a resource issue. I suspect increasing the marketshare of Max would resolve a few of these things.


Actually I wanted to address that in another post as well. 

I actually find it very unwise for Cycling '74 to offer Max for only one month in demo mode. That is not enough time for a beginner to get their hands around Max in order to realise that buying it is worth the effort. I only became a Max user because of the fact I found a way to crack it. In the end I bought it. In case someone is suggesting it is cheap so one can afford it, 1 it is not cheap for many of us, 2 why buy it if you are not sure you are going to use it?

TouchDesigner comes with a free version that you use non-commercially indefinitely. Once you realise you want to make money out of it or you realise you want to use the the power features (that you don't have in the free version, like using a bigger resolution) you buy the app. Which costs lots more that Max by the way. In the meantime the user becomes a power user and everyone is happy.

The conclusion is that by adopting this strategy, Cycling '74 is loosing a lot of potential users from my point of view.


Good news!
Here is an interesting update, Intel realsense seems to be finally available to us Mac users! ( I haven't tested it yet though!) here is a link to the library

https://github.com/IntelRealSense/librealsense

By the way, while (to an extent) I agree to what has been discussed still I think to rely too much on Cycling74 to provide and cover for every device out there is not realistic, I myself would prefer some tutorials on how to do such a task so we could do it for every device that we want to. I think that's the point of having a community after all.

The link says:
"This project is separate from the production software stack available in the Intel® RealSense™ SDK, namely that this library only encompasses camera capture functionality without additional computer vision algorithms."

I think that means we get the camera stream inputs but don't get our hands on all the tasty tracking, gesture and scanning stuff sadly :(


you can just rent the software if it is interesting you? its quite cheap / month


There is already an external made for it , it gives you acces to the depth image but thats it https://github.com/blueyeti/jit.realsense 

Hi all, 
do you know if with this external is it possible to use multiple devices on a single computer?

if the usb driver is multiclient? which i wouldnt expect...

In theory it's possible with the SDK
https://dev.intelrealsense.com/docs/multiple-depth-cameras-configuration 
It's just a matter now to check if the external is capable of handling multiple devices

So, did somebody try and succeeded in using multiple Intel Realsense cameras with a single computer?

I believe it is possible to use multiple devices using the "jit.realsense" external programmed by BlueYeti, since in the code it states a device index to be used, which can also be changed through an attribute. 

The problem is that it doesn't work, returns an invalid pointer error in the Max console. 

- in jit.realsense.cpp there is a variable called "device", which can be assigned to an integer (the device index),
by default is set to "0".  This variable can be modified by the attribute "rs_index" inside Max. 
- When modifying the attribute from Max I get the error "invalid pointer", which I think could originate from this function: 

// Try to get the new device
      auto devs = ctx.query_devices();
      auto n_dev = devs.size();
      post("There are %d connected RealSense devices.\n", n_dev);

     if(n_dev <= device)
      {
        post("Device %d is not connected.", device);
        return;
      }

     dev = devs[device];
........
try
{
  // Get and check the data.
  if(!x || !x->dev)
  {
    error("No device");
    return JIT_ERR_INVALID_PTR;
  }

So, apparently  ctx.query_devices(); doesn't return any valid device. 

Looking more attentively I can see another error thrown out in the Max console:
realsense: info Firmware Version not supported by the device!

Which originates from: 
catch(const std::exception & e)
    {
      error("realsense: %s\n", e.what());
      if(dev)
        dev = nullptr;
    }

But this is very strange, since the device are exactly the same. 
I'll keep posting updates here which could possibly be helpful to someone in the future. 

It seems I'm able to change device by using the "rs_device" attribute, and it seems to switch between them. 
But, it will always use the stream of only one of the cameras, so even if it says "using device 1" it actually uses device 0. 
You can see they are actually different devices from the serial number. 

Using device 1, an Intel RealSense D435
    Serial number: 923322072590
    Firmware version: 05.11.01.100
There are 3 connected RealSense devices.

Using device 0, an Intel RealSense D435
    Serial number: 919122071639
    Firmware version: 05.11.01.100

Hi! I know this thread is quite old now but I am getting crazy trying to compile the jit.realsense object for macOS.
Did anyone succeeded in this? Please let me know, there is couple of things that sadly I do not understand :-(


I've started to take a look at this project and will be evaluating if I can commit some time to it. For now I've got the mac version of the external built and functioning as expected (as far as I can tell with some very basic tests).

Please give it a try: 
https://drive.google.com/file/d/1ZY1-XqI2a6QwyNfmmB_gimsovvR2Js_4 

Hi Rob, thanks a lot! Wonderful, really really thanks!


rob - I have a real sense 430 and made my own object to get the depth image but would prefer to use this if it works out. so far it seems to be working really well on the Mac. let me know if you need any testers in the future

Hi Rob, nice job. I just tested it with a D435 and it works as expected. 

I also tested it with the Lidar L515 and while the docs say it uses the same SDK, the external doesn't like it: 

realsense: Couldn't resolve requests.

Hi ROB,
I have installed on mac (mojave) Max 8 latest version, where should I put the jit.realsense.mxo file? I tried in the library and it doesn't recognize the object.

it makes me a strange thing: as soon as I create a new object in a new patch and write "jit.realsense" the MAX8 application unexpectedly closes

create a package folder with an externals folder inside and drop the mxo in there:

~Documents/Max 8/Packages/yourPackage/externals/jit.realsense.mxo

restart max

now it should work. I assume once the object is production ready it will be distributed this way..


I tried to build your updated repo but I failed with the cmake step:

maybites@maybitessMBP172 build % cmake /Users/maybites/Arbeiten/02_code/library/max/jit.realsense -DCMAKE_BUILD_TYPE=Release
-- The C compiler identification is AppleClang 11.0.3.11030032
-- The CXX compiler identification is AppleClang 11.0.3.11030032
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/c++ - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
CMake Warning (dev) at CMakeLists.txt:27 (set):
  implicitly converting '' to 'STRING' type.
This warning is for project developers.  Use -Wno-dev to suppress it.
CMake Warning (dev) at librealsense/CMake/lrs_options.cmake:12 (option):
  Policy CMP0077 is not set: option() honors normal variables.  Run "cmake
  --help-policy CMP0077" for policy details.  Use the cmake_policy command to
  set the policy and suppress this warning.
  
  For compatibility with older versions of CMake, option is clearing the
  normal variable 'BUILD_UNIT_TESTS'.
Call Stack (most recent call first):
  librealsense/CMakeLists.txt:5 (include)
This warning is for project developers.  Use -Wno-dev to suppress it

CMake Warning (dev) at librealsense/CMake/lrs_options.cmake:14 (option):
  Policy CMP0077 is not set: option() honors normal variables.  Run "cmake
  --help-policy CMP0077" for policy details.  Use the cmake_policy command to
  set the policy and suppress this warning.
  For compatibility with older versions of CMake, option is clearing the
  normal variable 'BUILD_EXAMPLES'.
Call Stack (most recent call first):
  librealsense/CMakeLists.txt:5 (include)
This warning is for project developers.  Use -Wno-dev to suppress it.
-- Internet connection identified
-- Info: REALSENSE_VERSION_STRING=2.38.1
-- Performing Test COMPILER_SUPPORTS_CXX11
-- Performing Test COMPILER_SUPPORTS_CXX11 - Success
-- Performing Test COMPILER_SUPPORTS_CXX0X
-- Performing Test COMPILER_SUPPORTS_CXX0X - Success
-- Setting Unix configurations
-- using RS2_USE_LIBUVC_BACKEND
-- Found usb: /usr/local/lib/libusb-1.0.dylib  
CMake Error at librealsense/src/gl/CMakeLists.txt:43 (include):
  include could not find load file:
  /Users/maybites/Arbeiten/02_code/library/max/jit.realsense/CMake/opengl_config.cmake
  CMake Error at librealsense/examples/CMakeLists.txt:14 (include):
  include could not find load file:
  /Users/maybites/Arbeiten/02_code/library/max/jit.realsense/CMake/opengl_config.cmake
-- Looking for pthread.h
-- Looking for pthread.h - found
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD
-- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
-- Found Threads: TRUE  
-- Could NOT find apriltag (missing: APRILTAG_INC APRILTAG_LIB) 
-- Unable to find apriltag library, skipping pose-apriltag example
CMake Error at librealsense/tools/CMakeLists.txt:28 (include):
  include could not find load file:
  /Users/maybites/Arbeiten/02_code/library/max/jit.realsense/CMake/opengl_config.cmake
-- Building with TM2
-- Fetching recommended firmwares:
-- D4XX_FW_VERSION: 5.12.7.100
-- SR3XX_FW_VERSION: 3.26.1.0
-- T26X_FW_VERSION: 0.2.0.951
-- L5XX_FW_VERSION: 1.5.0.0
-- http://realsense-hw-public.s3.amazonaws.com/Releases/RS4xx/FW/D4XX_FW_Image-5.12.7.100.bin
-- Download firmware 0;"No error" for D4XX_FW_Image-5.12.7.100.bin
-- http://realsense-hw-public.s3.amazonaws.com/Releases/SR300/FW/SR3XX_FW_Image-3.26.1.0.bin
-- Download firmware 0;"No error" for SR3XX_FW_Image-3.26.1.0.bin
-- http://realsense-hw-public.s3.amazonaws.com/Releases/TM2/FW/target/0.2.0.951/target-0.2.0.951.mvcmd
-- Download firmware 0;"No error" for target-0.2.0.951.mvcmd
-- http://realsense-hw-public.s3.amazonaws.com/Releases/L5xx/FW/L5XX_FW_Image-1.5.0.0.bin
-- Download firmware 0;"No error" for L5XX_FW_Image-1.5.0.0.bin
-- Configuring incomplete, errors occurred!

I wanted to look into the possibility to output the pointcloud as a matrix to be used with [jit.gl.mesh]. or maybe even one step further: the possibility of [jit.gl.realsense], where the object renders the mesh directly like a 3d model.


the readme in the repo is incorrect for building Mac. I won't be providing any support on this anytime soon, but it should be pretty straight forward to get going. steps are something like:

git clone, including all submodules

cd /path/to/jit.realsense
mkdir build && cd build
cmake -G Xcode ..

open the generated xcode project inside the build folder and build

you may need to run the cmake step twice (i did, errors the first time)

one thing i noticed, for some reason the xcode project Debug configuration has optimizations enabled, so if you want to debug you'll have to manually disable the optimizations setting (Optimization Level setting) or figure out why cmake is generating with those enabled.

Thanks for the hints. I indeed had to redo the first cmake command.

the readme in the repo is incorrect for building Mac.

I was able to build the external with the original readme command. both ways seem to work. 

I have forked the repo and will update the readme with your suggestions plus a fix to the 'CMakeLists.txt' 

-> https://github.com/tecartlab/jit.realsense

TNX @MAYBITES 
I tried to make the folder with externals in PACKAGEs but it doesn't work it quits unexpectedly as soon as i try to create an object with "jit.realsense". I reinstalled max8. Can someone help me? 

Please, Can you post your external here: "jit.realsense.mxo"? I will copy it to my externals folder 
thnx!


https://github.com/tecartlab/jit.realsense/releases/tag/v5 

THNX MAYBITES
I copied the folder in Package/jit.realsense/externals but this happens. WHY IT?
What do you think about it?


Yes I restarted max.
Should I install  Intel® RealSense™ SDK 2.0 (v2.39.0) first of all this?

I think thats not needed. the brown color indicates max is not finding the object-external.

restart Max, create a blank patch, add a  jit.realsense object, go to the max console, select all, copy, and paste the contents here 

I would like to do what you tell me but does not allow it closes immediately as soon as I create the object jit.realsense and press return.
Is it 32bit?
I use max8
tnhx

please, can you create another build version for MAC OS of  jit.realsense?

you have a screenshot posted above. there is a jit.realsense object that is tinged orange. I just want to know what the Max console says in that screenshot. Help us to help you!

yes, that is the situation in OCT 08 2020 | 10:56 AM. but now  MAX8 closes immediately as soon as I create the object jit.realsense and press return.

hi I build the external by cmake but this is

remove the quarantine attribute via xattr - https://cycling74.com/articles/using-unsigned-max-externals-on-mac-os-10-15-catalina 

also your folder structure in the Packages folder should be Packages / realsense / externals / jit.realsense.mxo 

i would just use the -d option and remove it on the entire folder.

 unfortunately it doesn't work...,
I tried it but terminal display returns this:


xattr: /Users/admin/Documents/Max 8/Packages/externals: No such xattr: com.apple.quarantine

hi,
I had trouble too. This is what helped for me:
1. You need Catalina (OS X 10.15.7)
2. Use xattr as described above (Type in terminal: xattr -d -r 
com.apple.quarantine then drag the folder containing 
jit.realsense.mxo into the terminal window and hit enter)
3. install homebrew https://brew.sh/index_de
4. install libusb with brew install libusb
5. not sure if the installing the xCode command line tools is necessary too?

My understanding is, that libusb is required to use jit.realsense.mxo.  


(i think xcode command line tools is necessary to install homebrew anyway)

thank youThank you rob! I have been debugging in the terminal tool for 1 day, but I can’t use  sdk in maxos normally. Fortunately, I searched the forum and found jit.realsense here. It works very well and is perfect! grateful!

Hi all,
did someone succesfully use the external with 2 devices connected to the same computer? i 'm on Windows10 64bit and use the compiled version from blueyeti..(eventually it's not the newest version)  

Federico raised this question too some time ago... i can switch rs_device but the output stays the same.
maybe someone can provide a already built version for Win10 64bit? 

Hi Mat, I can confirm you that I managed to get at least 3 devices working on the same computer.
The main point are the cables. Be sure to be using powered USB cables. 


i will try to find some time in the near future to clean up the versions I have and post them somewhere useful.

ok. interesting. because in touch i can sitch easily both devices. 
 but lets see.  i will try that. 

so there's no point in built the external from rob's git?

thanks Federico for this supermegafast answer. 



@maybites I downloaded your version of jit.realsense. It's recognised by Max 8 just fine but clicking the toggle object to start the stream produces the following error in the Max console. 

Using device 0, an Intel RealSense D415
    Serial number: 821312061537
    Firmware version: 05.13.00.50
realsense: Couldn't resolve requests
wait_for_frames cannot be called before start()
jit.realsense: error type unspecified
wait_for_frames cannot be called before start()
jit.realsense: error type unspecified


Is Anyone else experiencing the same behaviour? I have realsense sdk, max-sdk, and libusb installed. I'm running MacOS 11.7.1


Intel Realsense development

dailytlj

Hi, quick question (with lots of details and specifics following):
are there any tools or workflows for debugging usb/serial stability issues? 
(Windows 10, app built from Max 7)
either in Max, external software, or is there something I can learn from Windows logs? 

I know there are about a million forum questions about serial errors, but I haven't been able to find any that cover the problem I've run into: initial correct operation, with delayed failure. 

I have a computer running a max patch for an art installation. It's been installed and operating correctly for 2 months, but in the last week, serial errors have begun occurring. Not at startup, but at seemingly random intervals through the day: sometimes 30 min, sometimes 5 hours.  

The error message isn't a lot of help:
serial - error writing to serial port

the serial object sends values every ~5 seconds. Once this error occurs, it continues to occur on every subsequent attempt to send values until I close and reopen the serial port in max with the following steps:
-send "close" message to serial object.
-get new error: "serial: error resetting port attributes"
-send "close" message again
-send "open" message

I've hacked together a temporary fix that automates the above process using [error] and some [route] and [delay] objects, but I need to diagnose the underlying problem and find a true fix. Added complication: the installation is about 5000 km away. I have remote access to computer via teamviewer, but can't fiddle with the physical usb port/cable/etc. 

Any ideas appreciated!

Hi, quick question (with lots of details and specifics following):
are there any tools or workflows for debugging usb/serial stability issue

debugging-serial-instability

Hi, I'm having a very similar problem. I'm trying to use an arduino based sequencer with MAX. The serial communication works just fine from arduino to MAX but when I'm trying to send serial data from MAX (not even to arduino yet, just testing the serial communication with MAX console printing), MAX gets somehow jammed and starts banging "error writing to serial port"

What I'm trying to do is to send metronome clock to arduino to light leds to indicate which sequenced note is being played. I'm using serial communication at 9600 baud. Tried with the default 4800 baud and it's a bit more stable. with @dtr 1 the serial is also a little bit more stable, but gets jammed in the way as DAILYTLJ explained.

I have the same problem and need it to work for an art installation in two weeks!
Did anyone find a solution for this problem?

Hi Django,
this is not really an answer to your questions. I am merely sharing some thoughts and experiences.
When it comes to installations that need to run (unsupervised) over a long period of time I usually avoid using serial connections. In general my recommendation would be, to see if there is an other protocol that you could use.
Besides occasional stability issues that the serial protocol has, I find it also problematic because if accidentally unplugged the port usually needs to be closed, then reconnected and opened again. In this process the serial port may show up with a different ID on your computer. In that case ports need to be rescanned... 
These are in my experience too many steps for museum/gallery personnel and require some specialised technical knowledge.
If possible you may want to consider if using e.g. MIDI or OSC is a viable option for your work. Both have the advantage that they reconnect automatically when unplugged and they are usually far more stable over a longer period of time. 

Your absolutely right Jan. I used Teensy 4.1s with ethernet connections for installations and it is a joy to work with. 

I’ve also found that - worst case scenario - you can usually reset (Close open etc) serial port in the background after a certain amount of time (using some logic to make sure nothing important is going through the serial port at the time). You can even make a stand-alone just with the serial object which you open and close with the shell object, and use udp to talk to it from your main patch.

This doesn’t work in all cases but can be a solution if you’re in trouble. I’ve had installations running for years with serial objects doing this.

Same problem. i have no problem with comunication from Arduino to Max , but if i try the opposite max print the same error message.
- Arduino in port " f : COM6 "
- I use the object " serial f 9600 "

Can one track that serial problem to specific OS or Max version ?

I never had any problems on Mac, but was allways using either custom made USB cables in case longer ones were needed, and taking baud rate into account in that respect.
Interesting is that all reports state that it works for some period of time, and then suddenly starts droping out.
Would mean something has changed in the meantime,
or OS keeps some COM port logs which get corrupt ? who knows.

In case it is a problem of max serial object itself, it might be
worth trying jasch comport object instead, which aso was allways reliable for me.

debugging serial instability

chapelier fou

I'd like to get the currently playing clip name. I have no problem getting the name of a session clip, but I want to keep track of the currently playing clip, either in session or in arrangement.
The only way I see for now to get the clip currently playing in arrangement would be to get all arrangement clip ids, get the current transport time, and check for all of them if the current time fits into the starting and end points. That can't be that twisted, can it ?  
Any help ? 

I'd like to get the currently playing clip name. I have no problem getting the name of a session clip, but I want to keep track of the curre

get-playing-clip-name-also-in-arrangement

Did you ever end up figuring this out? I have the same question.

Hi, i have a very useful patch made by Julien Bayle, 
hope that help you

That patch from Julien Bayle only gives us the name of the playing clip in session view. Has anyone figured out yet how to get the name of the playing clip in the arrangement?
I'm currently working on a patch that wouldn't work at all in the arrangement view, thus also not when you're exporting your arrangement, as long as I don't solve this problem.

Okay, I got a little bit further. I have a solution in theory but it doesn't work in practice.
Clips have a property named is_playing. So I thought I could just iterate through all the arrangement clips and check if is_playing is 1 for one of them. But even if I turn on the arrangement loop and fill the whole loop with clips, all clips have is_playing 0.


However, when I try to read is_playing from a clip in the session view, everything works as expected.


Does anyone know if this is a known bug? Should I report this? Or am I just doing something wrong?

Get playing clip name. Also in arrangement. 

Klockner

I'm trying to automate the multistep process involved in using HISS tools' IR Measurement Device room profiling within a max project, but I'm running into an issue where the parameter I'm using to save my impulse response (message "Save 1", multi-second delay followed by message "Save 0") seems to operate differently than when I manually press the device's "Save" button that needs to get pressed.

I'm noting my issue in the difference of the way that the Convolution Reverb Pro interprets the IR file (whose name stays the same but is re-written upon each save with the IR Measurement Device): The bang-initiated "Save 1"/"Save 0" message saves the IR from the sine sweep and interprets it poorly, as only a .03 sec response, but when I physically click the "Save" button, it interprets accurately, as a .33 sec response.  Weirdly, it seems that physically clicking the bang button also finds the device interpreting it accurately -- (would a delay object somehow produce a different type of bang?)

Furthermore, while I'm unable to predict when this happens, sometimes I also receive an error message: iraverage~: not enough room in buffer 001IRA -- which I think I understand as the trim+normalize feature of the IR measurement just being filled larger than is allotted. -- it doesn't appear to 100% correlate with the more obscure issue above.

I should mention that I'm quite a beginner, so my working knowledge of objects/etc may be missing a very basic oversight by trying to crash course my way into something like this project.

I'm trying to automate the multistep process involved in using HISS tools' IR Measurement Device room profiling within a max project, but I'

issue-interfacing-with-max4live-device-parameters-ir-measurement-automation

[deferlow] is better to wait for disk operations to complete. [delay] output is high priority which is not suitable for save/load operations. If you connect a defer / deferlow after the delay the bang will be triggered in the "normal" main thread of Max and not in the timer thread. That also depends on overdrive on/off setting of Max. But there are some explanation about this in the Max docs.
I did not try your patcher but it sounds like it is related to this.  

I forgot to thank you for this response! your answer was exactly the solution; I was unfamiliar with those different thread aspects you succinctly described and I spent more time with the documentation there afterwards. 

Issue interfacing with Max4Live device parameters (IR Measurement Automation)

Sorry in advance for my english. I hope, it will be understandable...
So...
I've been working a lot with genexpr these days...
Calculations and variable evaluations seem to be done for each sample, as soon as Params are involved, whether they are changing or not...
As soon as you put your variables in "if" blocks with a change in Param value  as a trigger for the the block to be evaluated, it works and your CPU is happy...
The drawback is : for this to work, you must declare your variables as Histories for data persistence over time...

I think I read somewhere in the forum that the compiler was already optimized for this purpose (that is, only evaluating variables on data change).
Could it be possible for the compiler to do the stuff for us  (I know memory is cheap but, hey... I have to ask)??
Anyway, I'd appreciate clarifications on this matter.

Thx!
version : Max 7.0.6 (32 bit)
system win 7


calculations-in-gen-and-optimizations

Bump from a while ago. I'm also curious how gen~ handles per sample vs per vector processing.  I found this in the gen~ references, but I'm curious if there are other values/variables that can be declared as lower priority. 

The compilation process for gen~ Gen patchers and GenExprs includes an optimization that takes into account the update rate of each operator, so that any calculations that do not need to occur at sample-rate (such as arithmetic on the outputs of param operators) instead process at a slower ‘control-rate’ for efficiency.

also curious how gen~ handles per sample vs per vector processing

only place where it’s per-vector is at the ‘param’ level - params come from event-messages in the outside max-world, similar to how you lock the event-rate to the start of each audio-vector using ‘in audio interrupt’ and ‘overdrive’ settings within the AudioOptions of Max-Proper, in order to sync/lock the event-rate to the audio-rate in a certain way, ’params’ within gen~ are automatically setup to work in this way. the rest of gen~ is sample-by-sample processing.

curious if there are other values/variables that can be declared as lower priority

someone can correct me if wrong, but since gen~ is all audio, there’s nothing within gen~ that’s actually lower-priority, it’s all the same priority within gen~(the params poll from the outside at a different ‘rate’ not ‘priority’; you could say that because events come from the outside world at a different priority, within this overall chain utilizing gen~ within it, gen~ receives the message with a lower priority, but within gen~ the calculation is sample-by-sample so the overall priority even in parts of the chain which are not required to be sample-by-sample will still run at the same priority - priority is more often a term used to refer to threading, within gen~ i’m pretty sure everything runs in the same audio-thread).
if you have a param connected to other ops that don’t require a sample-by-sample update(for example, connect a param, to ops that are all in a chain but only receiving input from the param like so: 
[param]->[* 2]->[/ samplerate]->…[the rest of your gen~ patch which actually needs to run sample-by-sample]…)… everything up to the point that does require a sample-by-sample update will only get polled/calculated at vector-rate(just a reminder: ‘control-rate’ can be thought of as the same as ‘vector-rate’; in the example above: [param]->[* 2]->[/ samplerate] all get calculated/polled at the ‘control-rate’ mentioned in that passage from the docs you quoted. the rest of it after that, will be calculated sample-by-sample… the reason is the [* 2] doesn’t require any sample-rate level of changing data, neither does the [/ samplerate] op, they both only need to update when the param feeding them input updates(or when the entire system changes by editing or setting a different sample-rate)… i’m sure they’re all still in the same thread/priority though because otherwise, the rest of the gen~ patch will suffer from timing/sync probs if data must be shifted to a different thread at this late point… i could be wrong: now that i think about it, ‘priority‘/‘threading’ is not really spoken about in the docs i’ve read so far - i think that’s because gen~ is all audio-thread only(same priority throughout))

the only other area of confusion i imagine someone might have is with History ops but where you might send the history op a message from outside(hey Cycling74, i’m pretty good at this! y’all should hire me for help on anticipating what will confuse users in your help docs, haha! XD)
most people probably won’t want to send a History op a message from outside… this is more for resetting a specific History(you can also send the ‘reset’ message from outside, so you don’t really need this history-specific messaging in a user-specific situation, it’s more for debugging a specific part of your chain with that ’history’ op in it) but if you have a history op named ‘gimme’, and you sent the message from Max-proper “gimme $1” then this message is an event message coming in at ‘control-rate’ from Max, therefore the History op within would be updated at ‘control-rate’ technically speaking, but since history ops have to be ready to process other input within gen~ at sample-rate, they will still calculate/poll for input at sample-rate.

and if you’re wondering how i come up with some of this as it’s not that specific in the docs, it’s because this is similar to how discussion of the AudioOptions settings(‘overdrive’ and ‘in audio interrupt’) has worked(described on these forums for years, eventually also further elaborated in the docs), and then i’ve also programmed a couple externals in C for MSP, so i can see the gen~ pattern is very similar(the ‘control-rate’ messaging to an MSP object is handled differently than the sample-rate perform-method and you can see when you program an external on your own how these are divided among threads(the ‘perform-method’ written in C code within an MSP external is like gen~, the rest of the MSP external is a bit more like Max in general), etc.), and finally! because this book helps alot:
https://cycling74.com/books/go 
(it doesn’t go into much more detail than what i did, but it describes it with much better, more technically-proficient semantics, haha, and then it goes into more on what will make gen~ most efficient for you which is often to be found in some basic design at the core of your algorithms which is different from this data-formatting/routing stuff, but this stuff is still important to know, no doubt! :) )


(damn, i rambled on and on and on and on and on and on here… :D… i’m not even sure i cleared anything up… ah well, makes more sense to me now at least, hahahaha XD)

Calculations in gen and optimizations

Robert Koster

Hey peeps, 

I've been digging deeper into lots of things and just searching for different options to play with when it comes to upsampling/oversampling, antialiasing and filters. I've got some examples of 4x interpolation; doing the process 4 times and then interpolating which filters out before nyquist... (from this forum thx to stkr and ernest)

but can I do a simple zero stuffing type upsampling in gen~? 
I dont need the accuracy of oversampling, just the temporary higher nyquist frequency. I tried using interp, input 1 my signal, input 2 being 0 with a factor of 0.5 interpolation but I didn't appear to work. Has anyone else tried this?

Hey peeps, 

I've been digging deeper into lots of things and just searching for different options to play with when it comes to upsampling/

zero-stuffing-in-gen~-upsampling

I have done a 2x up and down sampling patch. It's partially working but not really in a shape to share currently, but I can maybe talk through it a bit. 

The upsampling case has a signal input and two signal outputs for the odd & even signals. The input is written into a delay line. The outputs are the same delay line convolved with a sinc waveform (for the lowpass filtering), but moving through it in steps of 2. The first output is convolved with indices 0, 2, 4, .. etc, and the second output is convolved with indices 1, 3, 5, etc., which is the zero stuffing part of it. 

The downsampling case is near identical, but using two inputs (both writing into a delay line each), and summing the two convolutions back to a single output. 

// ASCII diagrams of upsampling:
// input signal as x[n], x[n-1], ... i.e. n[t] where t is delay index
// zero stuffed for 2x upsampling:
// n9 _  n8 _  n7 _  n6 _  n5 _  n4 _  n3 _  n2 _  n1 _  n0 _
// convolve with ideal IR kernels
// ir len == 9, buflen = 3
// out0                             _  b0 b1 b2 (1)b2 b1 b0 _
// out1                          _  b0 b1 b2 (1)b2 b1 b0 _
// IR len == 11, buflen = 4
// out0                       _  b0 b1 b2 b3  (1)b3 b2 b1 b0  _
// out1                    _  b0 b1 b2 b3  (1)b3 b2 b1 b0  _
// where b[t] indices the index into the IR buffer

// ASCII diagrams of downsampling
// in2 in1 in2 in1 in2 in1 in2 in1 in2 in1 in2 in1 in2 in1
// n6  n6  n5  n5  n4  n4  n3  n3  n2  n2  n1  n1  n0  n0
//                         _   b0  b1  b2  b3* b2  b1  b0  _
//                     _   b0  b1  b2  b3* b2  b1  b0  _
//                 _   b0  b1  b2  b3  b4* b3  b2  b1  b0  _
//             _   b0  b1  b2  b3  b4* b3  b2  b1  b0  _

The actual IR I was using was 21 samples long, computed as a sinc function under a Blackman window. Here's the genexpr for that part (assuming a [buffer~ ir_buf @samps 21]):

blackman(x) {
	return 0.42 - 0.5*cos(2*pi*x) + 0.08*cos(4*pi*x);
}

Buffer ir_buf;
// cutoff frequency in Hz
hz = in1;
// gain factor applied to all kernel to preserve throughput
gain = 2*hz/samplerate;
// if they change, recompute the IRs:
if (once || change(hz)) {
	once = 0;
	// length of the IR
	buflen = dim(ir_buf);
	sinc_midpoint = buflen-1;
	// the *conceptual* length of the IR
	// +3 because we trimmed the first & last (0), 
	// and middle occurs once
	len = buflen*2 + 1;
	// convert Hz to radians per sample
	w = hz * twopi/samplerate;
	// for each frame of the buffer:
	for (i=0; i<sinc_midpoint; i+=1) {
		// conceptual phase of the IR:
		// (i+1 because we trimmed the first 0-valued sample)
		p = (i+1)/(len-1);
		// k==0 at the dirac pulse in the centre of the sinc
		// (this is 1 sample after the buffer ends)
		// k==1 at the sample just before it, etc. etc.
		k = sinc_midpoint-i;
		// per-sample multiple/divisor for sinc:
		kw = k * w;
		// compute sinc kernel for lowpass
		lp = sin(kw)/kw;
		// sinc function is infinite, 
		// window it here to force sinc to zero at IR ends
		lp *= blackman(p);
		// write to the buffer:
		poke(ir_buf, lp * gain, i);
	}
 	// lastvalue:
	poke(ir_buf, gain, sinc_midpoint);
}

How are you going with this Graham? Have you made much progress?

:O You can do it Graham! Oversampling would be a great addition to Oopsy Daisy.

I've been playing around with for loops and convolution trying to get something going myself but I keep going around in circles at this point :(
I've tried a few different things and I think the issue is the iteration & index of buffers not lining up all the time and perhaps that the second filter should be processing the zeros? in any case, heres an example; 

// declare buffers:
Data input(64);
Data effect(64);
Data output(64);
// the impulse response: (14 zero crossings, kaiser window)
Buffer IR("sinc");
// the length of the IR:
len = dim(IR);
// declare variables for sums and output:
filter = 0;
FX_init = 0;
FX_filter = 0;
Out = 0;
// for loop
for (i=1; i<len; i+=2) {

	//store input samples
	poke(input,in1,i,0,0);

	// Read IR:
	// (by default peek will ignore out-of-bound indices)
	FIR = peek(IR, i, 0);		

	// len-i is to reverse buffer for convolution of input
	In = peek(input, (len-i), 0);

	// accumulate their product:
	filter += FIR * In;

	//inserts zeros by only placing samples at iteration points
	poke(effect,filter,i,0,0);
 
	//read (&scale from 1st sinc filter)
	FX_init = peek(effect, len-i, 0)*0.812;

	//apply effect and filter (skipping zeros coz we can?)
	FX_apply = tanh(FX_init);
	FX_filter += FIR2 * FX_apply;

	//store & read to decimate
	poke(output,FX_filter,i,0,0);
	Out = peek(output, i, 0);

FYI - Ive kind of been using this as some sort of guide; https://www.willpirkle.com/app-notes/multiratepolyphase/

I'm also really interested in this. Robert, did you get any further?

Not really @DAMU but perhaps @Graham Wakefield can tease us with his updated Oversampling example which is coming in the next gen~ book :P ???

There's a new book coming? That's exciting. Do you know when that is due? I've still been working on this as much as possible but my DSP knowledge isn't enough to do anything better than rudimentary upsampling.

There is a 2x spline upsampler and downsampler in the 5D filter. I optimized the coefficients so it doesnt use much CPU. 

Hi Ernest, thank you so much for replying. I have looked at the relevant parts of the 5D filter and i'm quite confused about which pieces to use to make a more general up/downsampling algorithm in gen~ 

I'd ideally like to make a 16/32/64x version  which works like poly~ and allows me to do weird synthesis/waveshaping techniques in gen~ with minimal aliasing. 

Following Will Pirkles methods outlined above, I understand the general principles of how this is done but I definitely need some more guidance on how to implement it. I have purchased Graham and Gregory's new book today which will perhaps enlighten me when it arrives on Friday. 

I totally understand if you'd rather not get dragged into this as I understand it isn't a trivial matter, but any more light you could shed on the answer would be very much appreciated.

That little oversampling algorithm is pretty cool ernest. I noticed a similar one for the SVF. 
I haven't managed to get either of those to work in a more general context though, I got either the incorrect output or no output, not sure why.  I'll revisit all this oversampling stuff again soon. 

I just finished reading Generating Sound and it says that oversampling will be discussed in Book 2, which is exciting. I wonder if we can get a sneek preview if I ask Graham politely one more time?..

I've done a bit more digging into Will Pirkle's code, but one thing still stumps me...

lets assume we are doing 4x OS with an FIR of 512 taps;


- the FIR is split into 4 'sub band filters' - 128 taps each
- interpolation uses reverse indexing (eg; i-- instead of i++)
- due to zero stuffing we perform gain compensation (multiply input by the OS ratio)
- 4x convolution processes are performed on each input sample (parallel)
- this results in 4 new outputs for every new sample


... then we process each sample individually, let's say with tanh, that means theres 4 parallel process taking place and we store these 4 samples into a buffer for the decimator.

- then the decimator is the same as the interpolator but with forward indexing (i++)
- we accumulate these 4 samples into a single sample for output.


Now for the questions... At first I would assume we need a 128 tap buffer for the input to convolve/interpolate but then why does the decimator only use 4 samples for the input? If we don't need it then are we still performing 512 (128 * 4) multiplications (convolution) on each sample input sample? how do we step through 128 indexes (sub band filters) within 1 sample frame?

I think after knowing that we can crack this 🤓

Just thinking out loud now... if we sum all the coefficients for the sub band filters and multiply the input by this, wouldn't the result mathematically be the same as multiplying the input by each coefficient and adding them up? If this is the case and we could do convolution this way, this would be a breeze... 
say the input is 1.0 (and the rest are zeroes), filter coeffs = 0.9, 0.2, 0.5, 0.3;
1* (sum of coeffs) = 1.9;
1 * (coeff 1) + 1 * (coeff 2) + 1 * (coeff 3) + 1 * (coeff 4) = 1.9;
Does this sound right or am I forgetting something?
If this is the case then indexing wouldn't matter... 🤔

Zero Stuffing in gen~ (upsampling)

leemorgan

Hi everyone - I keep using what I suppose are glorified LFOs to provide moving variables in my patches, and have done for years - but they always seem, visually at least, to be kind of jerky.  Is there a most efficient way to do them (one example enclosed)?

 I try a different method every time - using counter for example, or ~peakamp  - but I never know which is most CPU friendly / smooth / reliable. Or is is just the visual feedback that's a little jumpy? 

Hi everyone - I keep using what I suppose are glorified LFOs to provide moving variables in my patches, and have done for years - but they a

most-efficient-wave-lco-patch

next time paste compressed patch, it is no fun to scroll arround to collect all your code.

Mixing signal and non signal objects is not optimal in any case.
If you need oscillator to drive a signal,
then do it all using signal objects, and then display it on GUI using
defered representation if you really need to have visual animation.
If you need non-signal oscillator, than don't use signal objects, or make better
signal to non signal conversion.
It is also better to check what range gets produced, in your patch 
you drive slider out of it's range


number–  has at best 20 ms report time, is that what your smooth ramp resolution is ?


Cycling '74 Forums

Max 8.5.1 / RNBO 1.0.1 Released

We have just released Max 8.5.1 and RNBO 1.0.1, which focuses on fixing crashes and other issues affecting OSes with certain language ...

assign midi map button within standalone application?

Hello Ruslan, yeah I did, there is an object called [mappings], I just looked into the help file and it was pretty easy at the end of the ...

Intel Realsense development

@maybites I downloaded your version of jit.realsense.

debugging serial instability

Can one track that serial problem to specific OS or Max version ? I never had any problems on Mac, but was allways using either custom made ...

Get playing clip name. Also in arrangement.

Okay, I got a little bit further. I have a solution in theory but it doesn't work in practice. Clips have a property named is_playing.

Issue interfacing with Max4Live device parameters (IR Measurement Automation)

I forgot to thank you for this response! your answer was exactly the solution; I was unfamiliar with those different thread aspects you ...

Calculations in gen and optimizations

also curious how gen~ handles per sample vs per vector processing only place where it’s per-vector is at the ‘param’ level - params come ...

Zero Stuffing in gen~ (upsampling)

Just thinking out loud now...

Most efficient Wave / LCO patch

Thanks - Roman / SA - much appreciated!

Custom function/envelope UI w/ Mira

I asked me the some question. Anyone here who can help?

[share] patches from my weeklybeats2022

Thank You! Enjoy(i should make a VST-instr out of that 'Lonely Dreamer'(sarod-like-karplus) subpatch...