Hardware build record

cubesky

daylanKifky from the internal blender console?

'D:\\Program\\Steam\\steamapps\\common\\Blender\\3.1\\python\\bin\\python.EXE'

>>> bpy.app.binary_path_python
Traceback (most recent call last):
  File "<blender_console>", line 1, in <module>
AttributeError: 'bpy.app' object has no attribute 'binary_path_python'

daylanKifky

Thanks!

we'll take a look at the other error asap

cubesky

daylanKifky
Hello~ Is there any update for this new addon?

daylanKifky

cubesky

Hi!

I found the problem behing the new error: The XML had a mispelled node name r-lowearm instead of r-lowerarm. I updated it now, you can find it in the same link

cubesky

daylanKifky
Ok. I will test it tomorrow. Thanks

cubesky

I just test the new xml, it almost correct, except the right leg is in wrong position, and right foot is rotate continuously.
This calibrate process is not stable, sometime it can not recognize the right direction of my arm and leg, you need to recalibrate it and hope it will correct this time.

Logfile for this process: https://gitlab.com/-/snippets/2309191
Video for result:

daylanKifky

Hello Cubesky,

Thanks for testing it and for your feedback.
We are currently working on a variation of the legs calibration procedure based on the results of our internal testing. Some of the issues you describe are consistent with our results.
Will let you know as soon as these modifications are available.

Regarding the rotating foot: It seems like a problem with the KCeptor itself. Please test it by it's own. You will probably need to recalibrate it, or there might even be a physical damage of the sensor.
You can also examine the raw sensor data with the Dump node. Sometimes a broken sensor will output wrong values in one of the axis. These wrong values are sometimes hard to spot, but other times you can clearly see them, like when you get a constant value even when you move the sensor.
In any case if have the chance to replace that sensor that's probably the best approach

cubesky

daylanKifky Sometimes a broken sensor will output wrong values in one of the axis. These wrong values are sometimes hard to spot, but other times you can clearly see them, like when you get a constant value even when you move the sensor.

Ok, I will test right leg and right foot sensor and recalibrate it, if it doesn't work I will try to change it.

Thanks for your amazing work for chordata, this is so great.

I'm now collaborate with a tailor to make a motion capture suit, if this suit design works, I will publish the design drawings with CC BY-NC-SA 4.0 license.

cubesky

daylanKifky Regarding the rotating foot

I re-test calibration, I think this rotating foot is a recognition problem in calibrate process.

I'm running multiple test, and the wrong rotation problem affect random sensor, this may related by calibration process on blender.

The algorithm misunderstood the direction of the sensor and calculate to the wrong direction.

Maybe this problem is same as that legs calibration problem.

https://gitlab.com/-/snippets/2326964

daylanKifky

cubesky I'm now collaborate with a tailor to make a motion capture suit, if this suit design works, I will publish the design drawings with CC BY-NC-SA 4.0 license.

Oh that's great!
If you want to open a separate thread to document the design process it would be a great knowledge base for the community, and we will be there to share our advice with you 😉

cubesky

daylanKifky
Ok, if painter start to work with this design, I will create a new thread for this.

daylanKifky

cubesky

It doesn't look like a pose calibration process, since it calculates the quaternion value just once and then applies it to the whole calibration session. You cannot get a rotating sensor as a result of a bad pose calibration.

It seems like the values coming from the sensor and fed into the kalman filter inside the notochord are not good.
Did you get the change to re-calibrate the sensor or replace it?

cubesky

daylanKifky
I randomly switch k-ceptor with another k-ceptor, but it seems it doesn't matter. Rotation error is randomly appear on random K-Ceptor node, everytime reset pose and restart calibrate process on blender will change the result.

cubesky

daylanKifky
It seems recalibrate the K-Ceptor itself doesn't work.

daylanKifky

cubesky
what about replacing it?

if you can post a dump of the raw values of the sensor while it rests on a table I can try to spot errors. IF you do so please use an external service like pastebin.com and include the link here

cubesky

daylanKifky
I don't have replacement sensor this time, COVID-19 is affect express delivery this time in my country. But I can dump raw values of sensor, I will use Gitlab snippets to post data.

cubesky

daylanKifky
I just collect all 15 K-Ceptor data, here is the debug log. I notice some K-Ceptor has wrong initial orientation. Maybe it needs recalibration.

cubesky

cubesky A K-Ceptor output wrong orientation, when I place it facing south, it output facing west. This error maybe affect the pervious foot rotation problem.

daylanKifky

Hi everyone!
Sorry for missing these posts. In fact the spam is terrible lately. We are looking for alternative systems to filter them since the current ones are not working. As cubesky suggested they are probably hand posted by humans.

cubesky I notice some K-Ceptor has wrong initial orientation. Maybe it needs recalibration.

Can you point which ones had this initial wrong orientation?

cubesky

@daylanKifky
And I think chordata team need to clean up this forum by removing spam threads and add effective captcha system to avoid spam thread.

NakedRabbit

cubesky I'm doing all I can to flag them, but the robots will beat me every time.

cubesky

NakedRabbit I see.... It seems reCaptcha can not detect them... Maybe they are manually operation by human?

« Previous Page Next Page »