One last thing / regarding my mention in the other thread about using Auto Position. Your response was discouraged about creating positions for each frame.
It occurred to me throughout this conversation that you might think frame is like animation frame. Where there’s 30 per second or something....
The Auto Position calls a frame a resting goal position of servos. An action consists of a few frames. The algorithm interpolates the servo positions between two frames. Frames are goal positions - not per second or per movement. The system calculates that for you.
Are you sure it’s not the Auto Position you’re looking for?
The reason I suggested FBX as it is an interchange file format created by Autodesk, so that any animation software (3D max, Maya, motionbuilder and Lightwave etc) can use animation data across different software. Basically you are logging the scene, lights camera, angles of joints etc. in the FBX file.
Obviously the only thing useful to us are joint positions on bones on certain frames so we can use those to map to our robots.
Now I have no idea what’s going on. What is maya? Or motion builder? What program do you actually want this for? I’m not sure I know what it is you need anymore. If someone can explain the exact problem you’re wanting to solve and with what software and what format?
This conversation is a dream of million wishes but no clarity. I’m not sure i can help you at this stage. Identify between the few of you what it actually is you need and let me know once you agree
that sounds great actually, let me fiddle with it, run it by some friends who have the magic. Make sure it does it reliably by running it through a view different configs then send it over tomorrow once I know it works and you can do your magic.
I think that'd be really great as a consumer to get this software working with yours. So I started out as a scientist and got sick of publish or perish academia malarkey and decided to take my ideas, make them real, try to do some good and help people. So as an inventor early in my "mad scientist" career I've had to start learning a bit of everything. Where before I studied neuroscience and barely could wire up an arduino, now I have to do and learn everything from code to MIM manufacturing techniques etcs. For me and other makers, real time control with animation would be a great tool to create truly new things without having to learn freaking EVERYTHING.
I appreciate your double checking, yeah I read through your manual, watched a bunch of videos (good stuff BTW) and played with the software. Its not that its failing me algorithmically, its that as I said the kinematics in this machine are VERY weird. Setting a position goal would get it to move to that position but there are 17 joints in total and the NEXT movement, to be completed correctly will require some parts to move away from each other and others to move in not entirely intuitive ways. In short, I would love to just buy your kits and stay 100% within the platform but needing nema's, the alien kinematics etc extends past your original vision I think, though its so cool that the platform is evolving complexity like this. As a biologist, im all about that self organizing complexity ha
ah didnt see the last few bits. No nothing has really changed, I agree with @fxrtst that .fbs would be ideal, its simply a matter of quantifying what that means. Like you said, it has alot of complicated data in each file. Motion builder, Maya, 3ds Max are just different animation/game design type suites that autodesk has and fbs is (with certain limitations) interchangeable among them. I think taking fbs files should be the goal, like @fxrtst said someday but starting with the route we discussed I think is a good starting point. Frankly it has alot of potential getting fbs working through ARC. Itll expand the range of interested consumer to entire categories that are being ignored. Would you agree @fxrtst? I think we should start with this plugin then work up to general .fbs guided robotic programming as that will be a crazy involved process with a bunch of unknowns thatll probably fizzle out again like it did last year with the last guy wbs that tried. Instead with a bit of max script and magic plugin stuff from DJ we can have something that works to build off of as a community and not lose steam since it will take a long time to get fbs ARC programming completely realized.
All these are exactly the same concept, altho if not coming from an animation background its hard to wrap your head around.
Autodesk owns 3D Studio Max, Maya and Motionbuilder so the animation systems are basically the same. The .FBX files from all these programs are the same. Everything we have been discussing be it 3d Studio Max or Maya all have the same animation processes.
The endgame is the same, exporting animations (or real time) from a virtual robots inside an animation program and applying the rotational values to matching servos.
How we get there doesn't matter to me. Which program doesn't matter to me. I added the links above because someone has already created a python plug that does this with ardurino and sometimes reverse engineering gives some or all the answers.
I built the above system from Dan Thomson in 2012 and got a servo to move in realtime by moving a virtual jaw in Maya (video somewhere?). But, I just followed his directions. I have no coding skills or I would have tackled a plug in for ARC a long time ago
It has been a dream as I said for 30 years. But, in this thread are others thinking the same way as me, so I assume Im not the only one who sees value in a system like this.
That said this is an extremely time consuming concept. You have a business to run. Perhaps some of the fellas here on the forum would like a stab at moving forward (PTP ?)
Yeah, this is fair. I think starting out with a plug-in would get the train rolling to get users with the skills to take the software there. Basic group psychology, show people something works and mass psych forces take over. "Hey look fire cooks food," that cought on and now I'm heating up pizza bagels in my microwave.
I haven't had the dream for 30 years but I would love to see this work done as well. I'm working with my buddies on cleaning up that script that produces the csv file. It works-ish at the moment but there's a few irregularities that should be worked out tomorrow. As long as we all agree and DJ is still willing then we can in a sense, program a robot with an animation. That I think should grab the attention of the more skilled users.
Anyway, let me know if this is cool with everyone. Unfortunately my skills are pretty limited so I can't help much beyond this light data extraction stuff but I'll do what I can!
Hey guys, great to see this dead threat is hot again!
I had a lot of stuff going on in my life lately and @WBS00001 had an accident which is the reason he could no longer work on this...
But its good timing to dig this out again since I have been doing a lot of robotics lately, including the continous work of exporting animation to my robot out of 3ds max!
Writing the servo positions to a .txt file is not a problem at all...live control is also possible! IK is possible...I made a fully rigged JD for testing purposes! I wrote all this in Max Script and can share it if needed!
At the moment my robot is FK rigged and I am exporting the servo positions as a plain .txt file, I am still struggling with an IK setup...someone specialized in mechanical rigging within 3ds max would be highly appreciated!
The main problem though still persists, which is sending a lot of data correctly timed to the EZ-B or any other micro controller board...the EZ-B will not send those with the correct timing, since Windows is a task based operating system. And other microcontroller boards that I tried could not handle the traffic due to their limited onboard memory, and dropped out due to memory buffer overflow!
At the moment I am sending the positions frame by frame via LAN, rather than all the position in one coherent array..I had a bit of trouble with Wifi and switched to LAN for the sake speed and stability.
Also I realized that even when I am sending 30fps to the servos, the motion seems to be a bit choppy and needs to be smoothed out.
Just enough not to cause the movement being not carried out correctly, but only smoothing it a tiny bit in between the frames!
One last thing / regarding my mention in the other thread about using Auto Position. Your response was discouraged about creating positions for each frame.
It occurred to me throughout this conversation that you might think frame is like animation frame. Where there’s 30 per second or something....
The Auto Position calls a frame a resting goal position of servos. An action consists of a few frames. The algorithm interpolates the servo positions between two frames. Frames are goal positions - not per second or per movement. The system calculates that for you.
Are you sure it’s not the Auto Position you’re looking for?
The reason I suggested FBX as it is an interchange file format created by Autodesk, so that any animation software (3D max, Maya, motionbuilder and Lightwave etc) can use animation data across different software. Basically you are logging the scene, lights camera, angles of joints etc. in the FBX file.
Obviously the only thing useful to us are joint positions on bones on certain frames so we can use those to map to our robots.
Here is where Id like us to be someday.
Here is a plug in written in 2009 by Dan Thomson https://www.highend3d.com/maya/plugin/servo-tools-for-maya-for-maya
..and it demonstrated here:
Now I have no idea what’s going on. What is maya? Or motion builder? What program do you actually want this for? I’m not sure I know what it is you need anymore. If someone can explain the exact problem you’re wanting to solve and with what software and what format?
This conversation is a dream of million wishes but no clarity. I’m not sure i can help you at this stage. Identify between the few of you what it actually is you need and let me know once you agree
that sounds great actually, let me fiddle with it, run it by some friends who have the magic. Make sure it does it reliably by running it through a view different configs then send it over tomorrow once I know it works and you can do your magic.
I think that'd be really great as a consumer to get this software working with yours. So I started out as a scientist and got sick of publish or perish academia malarkey and decided to take my ideas, make them real, try to do some good and help people. So as an inventor early in my "mad scientist" career I've had to start learning a bit of everything. Where before I studied neuroscience and barely could wire up an arduino, now I have to do and learn everything from code to MIM manufacturing techniques etcs. For me and other makers, real time control with animation would be a great tool to create truly new things without having to learn freaking EVERYTHING.
I appreciate your double checking, yeah I read through your manual, watched a bunch of videos (good stuff BTW) and played with the software. Its not that its failing me algorithmically, its that as I said the kinematics in this machine are VERY weird. Setting a position goal would get it to move to that position but there are 17 joints in total and the NEXT movement, to be completed correctly will require some parts to move away from each other and others to move in not entirely intuitive ways. In short, I would love to just buy your kits and stay 100% within the platform but needing nema's, the alien kinematics etc extends past your original vision I think, though its so cool that the platform is evolving complexity like this. As a biologist, im all about that self organizing complexity ha
ah didnt see the last few bits. No nothing has really changed, I agree with @fxrtst that .fbs would be ideal, its simply a matter of quantifying what that means. Like you said, it has alot of complicated data in each file. Motion builder, Maya, 3ds Max are just different animation/game design type suites that autodesk has and fbs is (with certain limitations) interchangeable among them. I think taking fbs files should be the goal, like @fxrtst said someday but starting with the route we discussed I think is a good starting point. Frankly it has alot of potential getting fbs working through ARC. Itll expand the range of interested consumer to entire categories that are being ignored. Would you agree @fxrtst? I think we should start with this plugin then work up to general .fbs guided robotic programming as that will be a crazy involved process with a bunch of unknowns thatll probably fizzle out again like it did last year with the last guy wbs that tried. Instead with a bit of max script and magic plugin stuff from DJ we can have something that works to build off of as a community and not lose steam since it will take a long time to get fbs ARC programming completely realized.
@DJ, Its OK no worries.
All these are exactly the same concept, altho if not coming from an animation background its hard to wrap your head around.
Autodesk owns 3D Studio Max, Maya and Motionbuilder so the animation systems are basically the same. The .FBX files from all these programs are the same. Everything we have been discussing be it 3d Studio Max or Maya all have the same animation processes.
The endgame is the same, exporting animations (or real time) from a virtual robots inside an animation program and applying the rotational values to matching servos.
How we get there doesn't matter to me. Which program doesn't matter to me. I added the links above because someone has already created a python plug that does this with ardurino and sometimes reverse engineering gives some or all the answers.
I built the above system from Dan Thomson in 2012 and got a servo to move in realtime by moving a virtual jaw in Maya (video somewhere?). But, I just followed his directions. I have no coding skills or I would have tackled a plug in for ARC a long time ago
It has been a dream as I said for 30 years. But, in this thread are others thinking the same way as me, so I assume Im not the only one who sees value in a system like this.
That said this is an extremely time consuming concept. You have a business to run. Perhaps some of the fellas here on the forum would like a stab at moving forward (PTP ?)
Yeah, this is fair. I think starting out with a plug-in would get the train rolling to get users with the skills to take the software there. Basic group psychology, show people something works and mass psych forces take over. "Hey look fire cooks food," that cought on and now I'm heating up pizza bagels in my microwave.
I haven't had the dream for 30 years but I would love to see this work done as well. I'm working with my buddies on cleaning up that script that produces the csv file. It works-ish at the moment but there's a few irregularities that should be worked out tomorrow. As long as we all agree and DJ is still willing then we can in a sense, program a robot with an animation. That I think should grab the attention of the more skilled users.
Anyway, let me know if this is cool with everyone. Unfortunately my skills are pretty limited so I can't help much beyond this light data extraction stuff but I'll do what I can!
Hey guys, great to see this dead threat is hot again!
I had a lot of stuff going on in my life lately and @WBS00001 had an accident which is the reason he could no longer work on this...
But its good timing to dig this out again since I have been doing a lot of robotics lately, including the continous work of exporting animation to my robot out of 3ds max! Writing the servo positions to a .txt file is not a problem at all...live control is also possible! IK is possible...I made a fully rigged JD for testing purposes! I wrote all this in Max Script and can share it if needed! At the moment my robot is FK rigged and I am exporting the servo positions as a plain .txt file, I am still struggling with an IK setup...someone specialized in mechanical rigging within 3ds max would be highly appreciated!
The main problem though still persists, which is sending a lot of data correctly timed to the EZ-B or any other micro controller board...the EZ-B will not send those with the correct timing, since Windows is a task based operating system. And other microcontroller boards that I tried could not handle the traffic due to their limited onboard memory, and dropped out due to memory buffer overflow!
At the moment I am sending the positions frame by frame via LAN, rather than all the position in one coherent array..I had a bit of trouble with Wifi and switched to LAN for the sake speed and stability. Also I realized that even when I am sending 30fps to the servos, the motion seems to be a bit choppy and needs to be smoothed out. Just enough not to cause the movement being not carried out correctly, but only smoothing it a tiny bit in between the frames!