Simulations, concepts, 3D and more!

Very long time no post.
My bad, I'm terrible at keeping up with these things.
That said I haven't stopped working!
So here goes with a nice big update on what I've been up to.

First up is trying to simulate the spirt. In our film as mentioned in other posts, we have a golden spirit coming out of our main character. We discussed this being done using 3D particle simulations and as head of 3D it's my job to figure out how to do that.
I did a quick photoshop concept of the effect I wanted to achieve.

I showed this to the team and they liked the look, so I set out to figure out how to do it!

I grabbed up as many simulation tutorials as I could find and in the end I found a couple of tutorial sets by Digital Tutors that were super helpful. One was all about simulating smoke like effects using a voxel based system called Fluid Containers. The second was a particle simulator that runs off the ndynamics system in Maya.

The way I decided to create the effect was by using one or both of these effects emmiting from a 3D model of Oscar to make an etherial looking spirit simulation.

I started with the fluid containers tutorial set. I played around with this for a while trying to get the look I wanted. I managed to get the smoke effects emmiting from an object and being affected by gravity.

I found that to get nice smoke detail I needed to crank up the voxel count quite high which slowed down my computer quite a lot. The simulation time as well as the render time were enormous. I've since then found out that I can save out a simulation prior to rendering which speeds up the render by a tonne so that'll be helpful for future tests.
I quickly tried playing with some different colours of smoke as well.

I thought the effect looked nice but it could be taken further, though anything more complex will take a lot of time to simulate and render.

I realised that I would be able to cut down on simulation time if the emitter object wasn't full of voxels/particles. What I mean is that since my object was completely invisible I had to have a lot of voxels/particles covering its surface so that you could make out its shape. If I instead made the object visible I could have less simulation going on on the object's surface.
And so I created this.

After a lot of playing around in hypershade I managed to create a dynamic texture that looks like particles. I built the texture with multiple ramps connected to multipliers and a 3D noise texture so that as the object moves through it, cloudy particles seem to move along its surface.

I really liked that so I started playing with simulating particles flowing off of the sphere.
This time I used ndynamics. I watched another bunch of tutorials and eventually started to get the hang of it. I ended up simulating at a rate of 15,000 particles per second, which came out to about 80,000 particles simulated all together.

I was pretty happy with the result. Of course I didn't get this straight away, there was a lot of playing around and adjusting. In this simulation I added a turbulence field to give the particles more life. The blending between the texture and the particles came out really nice as well.

For reference, this is the look I was aiming for. It's not quite the same, but to render that many particles would have been impractical for me so what I have is good enough for now.

I showed the team what I had and they were pretty happy with how it's coming along. I have a bunch of stuff to try out with the next set of simulations. I want to add more size variation in the particles, I also want to play more with fields and try constraining particles to vector paths. But the main thing I want to try is compositing the smoke effect together with the particle effect to get more depth and variety.

So that's it for simulation for now, next up I did a little concepting!

I wasn't really on concepting on the schedule, but I felt inspired to do some environments.
I started by trying to do some matte paintings on top of existing environment photos.

 I thought they came out kind of muddy and it was hard to see the composition in them so I decided to break up the planes into colours.

I preferred the composition of the second one so I started to develop it further.

I removed the water so that all of the islands were floating in the sky. I really liked that composition so I decided to start to render it.

This one was just to get some atmospheric perspective. Then I added colour and started painting on the layers.

That's as far as I got with it but I liked the vibes I was getting. I was never really intending to use the concept since I wasn't rostered on concepting, but I showed it to the team anyway and it sparked up some good conversation about colour scheme and how the background should look etc. so I think it was worth it in the end.

Next on the list was the animatic. Oscar did the sound for that so I had to take that and match up the visuals as well as add a few extra little sounds to add effect.
I started by taking Tea's storyboard and cutting it into layers using Photoshop. I then imported those layers into After effects where I made a composition for each shot and dynamically linked them to Premiere Pro for the final editing. I also dynamically linked Oscar's Audition file so that I could edit the audio timing in audition. It too a decent bit of time to put it all together and add effects but I got it there in the end, so here it is.

I think that's generally what we're going for. The timing may change a bit and we may add in another shot or two. Exactly how the spirit world is going to work is still a little vague, but I worked with what I was given.

And lastly 3D. To be honest I haven't gotten super far with it. With all the simulation stuff and everything else modelling has taken a back seat for now. But I have some tutorials for making an organised human mesh and it's not the first human model I've made so I'm hoping it won't take too long. So far I have the symmetrical reference photos put together and my first basic shape has been laid down.

Hopefully Oscar doesn't mind me showing off his awesome body on here. You can see it's super early stages right now. The next step is probably a 5-10 hour modelling session to get it mostly done. Then I can move on to more simulation tests and other interesting stuff.

So All up I have a descent basis to work off for the rest of the semester as far as knowing what I need to do to get the effects that I want.

It's all bleet'n commin together!