[lucasvickers]

Hello,

I am working on an art piece that will involve the routing of an object in real time, based on input from various cameras etc.

I will be choosing one of the following three methods. I will be carving out of foam so that I do not have to worry as much about travel path. When I say outside software I mean software written by myself that in one way or another interfaces to the CAD software.

My possible paths are:
1) Most likely. Pre-defined shape entered into CAD package, but I need outside software to signal the CAD software to make a pass. This means that the machine will move very slowly based on external triggers.
2) Outside software simply triggers the movement of the router head, while the CAD software tracks location of the router and does overarching control. Think of this as an extension of mouse driven CNC.
3) Outside software directly writes to the stepper driver and moves the motors.

Does anyone have any experience on what CAD packages, if any, will work for this? I am an avid programmer so I do not care *how* the software communicates with the CAD software (pipes, database, socket, direct include of script, I am fine with all of them).

Preferably open source.

Thanks!
Lucas

[txcncman]

My possible paths are:
1) Most likely. Pre-defined shape entered into CAD package, but I need outside software to signal the CAD software to make a pass. This means that the machine will move very slowly based on external triggers.
2) Outside software simply triggers the movement of the router head, while the CAD software tracks location of the router and does overarching control. Think of this as an extension of mouse driven CNC.
3) Outside software directly writes to the stepper driver and moves the motors.

What does all the above mean? Are you capturing 1 raster source from a camera and wanting the router to machine the raster pixel by pixel?

Are you trying to overlay multiple images?

[lucasvickers]

I just realized using the term 'path' was incorrect since I am talking to CNC guys When I said paths I meant development paths pertaining to software.

I would like to control a CNC router head in real time.

What I really would like to try to do:
Link Python to Rhino3d (I know how to do this), and use info derived from Python (camera input) to create shapes inside Rhino. Then take cross sections of this shape and mill it into material along the y axis. I would like to do this in real time.

If, for example, you look at the attached image I have split the shape into sections. I'd like to find a way to in real time control a CNC router to create section by section.

I suspect there is no 'simple' workfow, so I'm trying to understand what I need to do. I believe manually I would probably export a segment, turn it into a g-code file, then pass this to CNC control software. I am trying to see how I can automate this all within software.

thanks!

[txcncman]

That helps...a little. I do not understand why you need real time control. In my world, I would construct my geometry "one slice at a time", but then compile all of the slices and compute all of the needed tool paths at once. Then I would output those tool paths to the CNC control for machining...all at once.

What you seem to be wanting to do is compute a tool path for once slice and then have that slice executed on the CNC machine. Then have the CNC machine sit and wait for you to compute the next slice.

Since this does not seem like a logical thing to do, I doubt if you will get much input for solutions on how to do it.

[lucasvickers]

you seem to be wanting to do is compute a tool path for once slice and then have that slice executed on the CNC machine. Then have the CNC machine sit and wait for you to compute the next slice.

This is more or less what I want to do, yes, and I'm trying to understand what software packages may exist for me to do this.

This is for an art project where an object will be created in real time based on audience movements. This is definitely not logical in terms of standard fabrication, but it's what I want to do.

A standard workflow would be what?

Rhino to make a shape
A package to go from the shape to the g-code
A package to control the CNC router based on the g-code

Could you let me know if there are any standard terms for the shape -> gcode and gcode -> cnc control software? This will help me research.

thanks for the input

[ger21]

Rhino to make a shape

CAD

A package to go from the shape to the g-code

CAM

A package to control the CNC router based on the g-code

Control or controller

When you're trying to do something that no one else is doing, there is really no standard workflow.

One way to possibly do what you want would be to use a control like Mach3. You can make a custom interface (screenset) that can contain custom buttons to create your g-code and run the machine. This can be done from external signals, from buttons or other devices. This way you could bypass both the CAD and CAM. But the learning curve may be very steep, as you'll be very much on your own.

[txcncman]

Since you are doing this real time, I would think you would want the process automated as much as possible. Let me offer the following:

1) Acquire digital image
2) Import image into custom software
3) Select "slice" of image to be vectorized in custom software
4) Vectorize "slice" in custom software
5) Convert vectorized "slice" into tool path in custom software
6) Convert tool path into G-code in custom software
7) Output to CNC machine
8) Execute output on CNC machine and standby

This is based off of my imagining that you are using a digital image of say 768 pixels wide by 1024 pixels high. And I am imagining that each "slice" to be machined will equate to 768 pixels wide by 8 pixels high. For the first pass of machining, the "slice" of pixel rows 1 through 8 will be vectorized and then machined. Once digital image #2 is obtained, the second pass of machining will be the "slice" of pixel rows 9 through 17, and so on.

Am I getting close?

There are other ways to do this, some using currently available software. I am just using the above as example for a base line for further discussion.

[lovebugjunkie]

You are waaaaay above my head here but if I understand you correctly (and even with your explanation, I'm not sure I do) I think my first choice would be to use a laser. The most common (hobby field at least) input to the DSP (controller card for laser) is a graphic's (Coral draw) program. You bypass the conversion of "picture to vectors"
No CAD, or Cam needed.

[txcncman]

I made up the attached graphic as an example of my imagining what it is you wish to do. Each "slice" of the example would actually represent a "slice" of a different photo, taken in time lapse sequence, for your purposes.

[lucasvickers]

You are waaaaay above my head here but if I understand you correctly (and even with your explanation, I'm not sure I do) I think my first choice would be to use a laser. The most common (hobby field at least) input to the DSP (controller card for laser) is a graphic's (Coral draw) program. You bypass the conversion of "picture to vectors"
No CAD, or Cam needed.

Laser has it's own issues, most notably it does not allow for 3d (or 2.5d if you will) surfaces.

[lucasvickers]

Since you are doing this real time, I would think you would want the process automated as much as possible. Let me offer the following:

1) Acquire digital image
2) Import image into custom software
3) Select "slice" of image to be vectorized in custom software
4) Vectorize "slice" in custom software
5) Convert vectorized "slice" into tool path in custom software
6) Convert tool path into G-code in custom software
7) Output to CNC machine
8) Execute output on CNC machine and standby

This is based off of my imagining that you are using a digital image of say 768 pixels wide by 1024 pixels high. And I am imagining that each "slice" to be machined will equate to 768 pixels wide by 8 pixels high. For the first pass of machining, the "slice" of pixel rows 1 through 8 will be vectorized and then machined. Once digital image #2 is obtained, the second pass of machining will be the "slice" of pixel rows 9 through 17, and so on.

Am I getting close?

There are other ways to do this, some using currently available software. I am just using the above as example for a base line for further discussion.

You are correct, minus a small abstraction. Allow me to make a little presentation to better explain myself.

See image. The input from computer vision affects the surface that I am working with. I will isolate sections of the surface and then route this. So yes, txcncman you are correct in your analogy. I will of course make an effort to try and make the final shape smooth, but that is not the concern of this thread.

I understand how to get to the point that I have shown in my visualization, but as I need to figure out how to get from 5 to 8.

I did a little homework today and found RhinoCAM , which appears to generate the G-Code within Rhino. I am going to speak with the company and try to understand if I can automate the use of rhinocam using Rhino3D scripting languages (python mainly), and potentially control the CNC router. This will be ideal.

I would like to discuss the other options you mentioned.

thanks everyone

[txcncman]

You are correct, minus a small abstraction. Allow me to make a little presentation to better explain myself.

See image. The input from computer vision affects the surface that I am working with. I will isolate sections of the surface and then route this. So yes, txcncman you are correct in your analogy. I will of course make an effort to try and make the final shape smooth, but that is not the concern of this thread.

I understand how to get to the point that I have shown in my visualization, but as I need to figure out how to get from 5 to 8.

I did a little homework today and found RhinoCAM , which appears to generate the G-Code within Rhino. I am going to speak with the company and try to understand if I can automate the use of rhinocam using Rhino3D scripting languages (python mainly), and potentially control the CNC router. This will be ideal.

I would like to discuss the other options you mentioned.

thanks everyone

Rhino does not directly have CAM capability, but it does have plug ins that will allow CAM. Also, Rhino does not directly have raster to vector conversion, but it does have at least one plug in that will do raster to vector conversion. If you are wanting Rhino to act as the CNC machine's controller also, I suppose that is possible, but I am not aware of a plug in that will currently allow that. But, with enough programming, I am sure a plug in could be created.

My options are:

1) Take digital image with digital camera
2) Load image from camera to PC
3) Open image with image editing software (such as PhotoStudio)
4) Crop image to section desired to be machined and save cropped image
5) Open cropped image in MasterCam
6) Use MasterCam ArtCam feature to convert image to surfaces
7) Assign tool path to surfaces with MasterCam
8) Create G-code from MasterCam tool paths in MasterCam
9) Output G-code from MasterCam to CNC machine with RS-232 cable
10) Set desired X, Y work zero on CNC machine
11) Execute G-code program on machine to machine material