We are working, trust me. The blog has been subject to an undeserved time-out; its parents have largely forgotten and abandoned it. But daddy is here. Pep-pep will make it all better.

Here are the fish, back from the dead. This shot has turned into a genuine collaborative effort- Jesse added some SWEET math to make the movement much more natural, Dave was a trooper and modeled all the proxy geometry of the boards, and I’m bringing it home with shading and compositing.

It’s not quite finished. Dave has tweaked and added boards which, might I say, now perfectly match the shot (you’ll notice there’s some gaps of nothingness, especially on the left side). As per Max’s suggestion, I’m going to do an occlusion pass for the boards so we get some interactive shadows on them. And of course, the flicker. I hate flicker. Other than that it’s just a bunch of nitty gritty stuff until this shines.

I haven’t posted in a while, but there’s two movies here to make up for it. I’ve been doing a lot of baby steps in various areas of the project with not much to show for it until now. First is the circuit breakers shot, which I actually did a composite of last week to throw on my demo reel as an in-progress clip. This is the updated one which is starting to look better. Using Dave’s awesome matchmove, I modeled the proxy geometry, set up render layers for mia_material_x, and tweaked the textures, shaders and hdr.

I’m using a technique for the indirect pass which involves projecting the footage as a texture onto the proxy geometry so Final Gather picks it up in the secondary diffuse bounce. This gives me a pass with some colorful bounce light which I toned down a lot in shake, but still contributes to how well they’re blending. They need some more work, like more glossy samples to reduce that crackling, but it’s getting there.

Next is the first render of the barnacles in the living room. This shot gave me the chance to use my geometry placement MEL script I wrote in Ken Huff’s class. The script will look at the grayscale value of a texture and duplicate geometry randomly, but with higher chances where the texture is black and progressively lower chances as it approaches white. I changed it so it only instanced geometry instead of duplicating it (3000 barnacles and it’s only 2mb), and added a scaling feature that makes the geometry smaller the more white the texture is. The result is accurate placement with natural looking randomness. This has a lot of flicker from final gather but it’s only on the indirect layer, I was messing with overdriving some of the secondary diffuse bounce values in both the shader and the render settings which is what’s causing it.

I’ve also been working on the fish with Jesse and plan on having a test comp of it by Wednesday.

I’ve been doing a lot of little things as well. Setup the Houdini scenes that needed the camera matchmoves. Got a decent start on the eel model/textures. Sent it all off to Jesse for animation testing.

Finalized the crab model (ack!), to send off to our rigger Liz Hersee. Wasn’t able to use the autorigged version from Patrick. Setup machine overkilled it, and everything was so buried it was hard for my unexperienced eyes to edit it efficiently. I’ve uploaded a screeny from zbrush to show where I’m freezing the model, and a fresh render showing the legs (almost done texturing), and some of accent geometry (the antennae, and leg flaps). The displacements should still be tweaked a notch or two, cause I still can’t get those spikes to sharpen up! The textures are almost done, I’m still not happy with the legs, and I need to do some new textures for the hairy flaps on the legs, and the antennae. I suppose this all means it’s time to start breaking it up into appropriate layers and figuring out the composite.

I looked into doing some sky replacement for one or two of the shots, and it’s not looking very good, but if I can pretty it up at all I ‘ll try and do it.

Two urchin posts in a row, but significant progress has been made on the li’l guy. The motion is 4x slower so it looks much more realistic now, I also fattened up the spines and furs, added a mouth, and introduced more variety into the spine and fur length. I added a fan under the mouth so the spikes don’t fold in on themselves so much and painted scales on the mouth spines so they drop off more drastically.

I rendered one pass with just a spotlight and deep shadows through the micropolygon renderer and another pass through PBR with an environment light and HDR map. As you can see from the split image below the PBR render took significantly longer but the HDR introduced a lot more color variation and brought occlusion into play.

The movie below is a hybrid of the two passes, in Shake I did an Add/Mix with 25% deep shadow and 75% PBR, this enabled me to keep most of the color and occlusion information of the PBR render while introducing the directionality and light modeling that I got in the deep shadow pass, it also reduces the feel of tranlucency slightly. It’s a subtle difference but I think it is effective. There is still the issue of geometry interpenetration but since these won’t be as large as this in the frame and there is so much motion going on I don’t think it’s something we will need to deal with at this point. Next up: Spec pass and reflectivity.

The urchin is coming. I started with a poly sphere here and gave it the proper number of divisions so that the points would line up with where the main spines should be. I copies small lines to the spine points and brought them into a popnet as a soft body and ran some turbulence through them to get some slight motion. I also scaled up the urchin shell a bit and brought it in as a collision object to limit the rotation angle so they didn’t swivel around though the shell.
Then for the main spikes I did a sort on the points and set the point normals equal to their worldspace position minus the position of the next ordered point. This makes the normal point whatever point is next, then I deleted the outer points and copied the spike geometry to it. for the minor spikes I follwed pretty much the same workflow except I copied 4-point l-system lines with a width attribute to the points with manipulated normals and turned those into guide hairs which I ran into a Fur SOP.
The resuts still need to be slowed down a bit but I think it looks promising. Rendered in about 15 seconds/frame with deep shadows and 1000 fur density, I’m using a basic surface shader on the fur so it should look better once I put a CVEX shader on it to pick up anisotropy and up the fur width a bit.