Just a bit about the mocap I have been doing. So like I have said I have used the 30 day free trail of IPI Soft marker-less mocap and I am super impressed and defiantly considering purchasing it in the future for me realistic animations. After capturing the data I was finding it difficult to process the recorded data. Not because of the software put the hardware. I had borrowed a friend’s laptop to record the data on, as I needed a portable windows machine to download the trail on and use to record. Unfortunately the laptop wasn’t always available to me so I lost out on a few days of processing time. On the days I did have it, the 2 mins of mocap data I needed to process was estimated at taking 3 days! At that rate I would never have got it done. So my tutor managed to get it installed on one of our windows desktop computers in college and what was going to take 3 days took 35-40 mins to process. So the approximately 3 mins of mocap data I had was done in a few hours and jitter removal.

Something I noticed with IPI was that it can have trouble identifying arm rotation. Sometime the shoulder joint or wrist joint would snap hard on the rotational axis.

To get the mocap onto my character:

  • Export the character mesh and bound rig as a FBX
  • The export tab IPI studio – and import target character. – Connect corresponding bones of the imported fbx rig to the IPI rig, using the dropdowns.
  • Export animation- save as FBX

Any controls added to the imported rig before it’s sent into IPI, they will not be connected once re imported to Maya.

My rig seems to work pretty nicely. Connections needed to be broken on the hip joint for translation, to stop the bobbing up and down. My character is in a fixed position and wouldn’t be able to translate only rotate.  Importing her back in Maya and on her podium, having the hip still able to rotate meant she was popping through the mesh where she was connected to the podium.

Going back to IPI studio I left the hip as unused. This meant there would only be movement above the hip and the hip was static, this would mean the mesh wouldn’t poke though the podium. Exporting and importing it back into Maya this disconnection from the hip seemed to make the whole upper body twist about 30 degrees to the left.

To try and eliminate this I tried the HumanIK rig in Maya, to see if it had a better relationship with the ipi rig.

I had had a play with it before, adjusting the number of bones and different components. Personally I don’t see the benefit of it, other than it the automatic naming conventions and the fiddly bits like the finger bones and feet are done for you. It seemed quite temperamental too. As my character had no lower legs or forearms I tried deleting them from the outliner. From then if I tried to change the rig in away way the deleted bones would pop back. It also didn’t like it when the bones were extended, they would later spike out when doing other amendment. When the rig did decide to behave and we managed to bind the mesh, testing the rotation of the bones with the mesh would terribly mangle and deform the mesh. When it did work and it imported into IPI studio there seemed to be no difference in the behavior from my rig.

Annabeth mentioned that the podium should be moving slightly with the animation also. I had already considered that but had gone on my ‘if I have time list’ and had planed to use a bend deformer. But we tried to see if we could use that hip data to get movement into the podium.

This meant combining the mesh of my character to the podium as one full piece. I’m porting that into IPI using the human IK rig gave the podium a dress like effect when the data was applied. This however didn’t work great. There were a lot of amendments and connection breaks I would need to figure out. Also having combined the podium to the body, the texture and UV map on the podium were over ridden by the UV’s and character texture.

So having come across all of this I decided to just go with my custom rig and just have the hip data set to un-used. No its not perfect and yes I should have done more testing so I was ready for these issues before needing to apply the data due to time running out. But from the results I have gotten it looks so much better than I had expected!

To get a closer mocap performance to simulate the constraints of the podium I could have bound the legs of my dancer, but I didn’t think that was wise for health and safety reasons. The dance I needed was very erratic and she really needed to keep her balance, binding her legs could have resulted in injury.

Advertisements