Day..... eight?

The shots have been matchmoved! At last! Albeit with some jittering. Progressed has also been made on the golem and fracturing.

For the golem, Mr Ron showed how my rockgen asset could be used on all the pieces of the golem at once. the displacement could then be baked onto a texture (using an asset from the Exchange) that would be applied to the model at render time. All this to allow me to work without my viewport hanging for a minute everytime I press spacebar. However, the baking did not turn out as expected. The UVs were all stacked on top of each other! To solve this, I separated the pieces into different groups corresponding to the arm, leg, etc, and merged and auto-mapped them in Maya before importing them to Houdini. They were then displaced and baked out individually. I guess there's no escaping these tedious tasks.. :(

More progress has been made on the ground and stair fracturing simulations. The RBD sim was going way too slowly to be usable, so Mr Ron recommended using particles to drive the effect instead and set up a rough network using the foreach node.

So basically, what the network does is doing a for loop on each of the pieces created by the voronoi fracture, creating a point at its center. It also generates a stamp index ID, associating the piece with that particular point. These particles are then grouped and piped into a particle network. the output is then fed into another for each node which pairs each piece back to its respective point. And voila! We have our fracture, 100 times faster! However it's not as accurate, that's the price to pay. I tried to get back the same cracking effect by using a sphere to collide with the particles, so the pieces wouldn't just fall together.

This method also allows multiple particle systems to be combined together. I used the second system to raise the pieces out of the ground and have them in a sort of swarm, encircling each other. An orbit node was used, with a custom axis and radius generated for each particle based on its ID. This introduces a lot of randomness into the behaviour, without which they would all orbit in only one axis.

Thanks to Mr Ron for all the help!