Model building & visualization questions
Posted: Mon May 31, 2021 10:31 am
Hello all,
I've got two OpenSim related questions (I'm on version 4.2). The modelling question is perhaps best explained visually:
To the left is a common situation. For instance, a femur 3D model (with no mass), and a PinJoint on both ends, with a volumetric object connected to it which is assigned a density, representing the total mass of the thigh. In this situation, the femur 3D model defines the rigid body. I want to split it up into a proximal & a distal portion, connected by a WeldJoint in the middle, so that I can acquire the midshaft joint reactions during a simulation. I can then use these to calculate the bending moments using Euler-Bernoulli beam theory. My question is the following: can I "get away" with situation B, where the bone is still massless, and entire thigh volumetric object is only connected to the proximal femur? The WeldJoint should make the entire femur still behave as a single rigid body, correct? So then it doesn't matter to which of the two ends the volumetric object is connected? The other option is of course situation C, where the volumetric object has also been split up into two and connected to each bone, but this is of course more trouble if you want to do it for multiple segments & multiple models.
My second question is related to the OpenSim standalone program (although I'm using it to visualize simulations optimized with Moco via MATLAB). Are there any ways to increase control over the visualizations? One the things I'd like to do is to have a model, as it is walking, stay in one place while the world moves backwards (so the camera translates at the same speed as the model). I can sort of hack this by just editing the motion so that the pelvis has no forward translation, but you don't get the visual reference of the world moving backwards. Are planned camera angles/motions possible, and are there any options to increase the resolution/quality of the animations? I appreciate that OpenSim's goal is not to produce very detailed renders etc of the motions that are acquired, so are there perhaps workflows available that allow for more powerful visualization tools, without having to for instance recreate a rig in blender and assign all the joint angle time series manually? I think the biggest hurdle here is to actually have to recreate a rig from scratch, perhaps this has been attempted before?
Best wishes,
Pasha
I've got two OpenSim related questions (I'm on version 4.2). The modelling question is perhaps best explained visually:
To the left is a common situation. For instance, a femur 3D model (with no mass), and a PinJoint on both ends, with a volumetric object connected to it which is assigned a density, representing the total mass of the thigh. In this situation, the femur 3D model defines the rigid body. I want to split it up into a proximal & a distal portion, connected by a WeldJoint in the middle, so that I can acquire the midshaft joint reactions during a simulation. I can then use these to calculate the bending moments using Euler-Bernoulli beam theory. My question is the following: can I "get away" with situation B, where the bone is still massless, and entire thigh volumetric object is only connected to the proximal femur? The WeldJoint should make the entire femur still behave as a single rigid body, correct? So then it doesn't matter to which of the two ends the volumetric object is connected? The other option is of course situation C, where the volumetric object has also been split up into two and connected to each bone, but this is of course more trouble if you want to do it for multiple segments & multiple models.
My second question is related to the OpenSim standalone program (although I'm using it to visualize simulations optimized with Moco via MATLAB). Are there any ways to increase control over the visualizations? One the things I'd like to do is to have a model, as it is walking, stay in one place while the world moves backwards (so the camera translates at the same speed as the model). I can sort of hack this by just editing the motion so that the pelvis has no forward translation, but you don't get the visual reference of the world moving backwards. Are planned camera angles/motions possible, and are there any options to increase the resolution/quality of the animations? I appreciate that OpenSim's goal is not to produce very detailed renders etc of the motions that are acquired, so are there perhaps workflows available that allow for more powerful visualization tools, without having to for instance recreate a rig in blender and assign all the joint angle time series manually? I think the biggest hurdle here is to actually have to recreate a rig from scratch, perhaps this has been attempted before?
Best wishes,
Pasha