VFX STUDIO I - Winter 2009 - Jordan Rempel

--- Joe Pasquale supervising ---


OBJECTIVE - To produce a full CG short containing approximately 12 shots in 10 weeks.
SYNOPSIS - A small unmanned submersible is traversing the depths of the ocean when it encounters a ominous mysterious object. Suddenly the sub is rendered powerless and is pulled inside the "mouth" of the monster. In the end, it becomes clear that the monster has absorbed the sub and made it a part of itself, along with other man-made machinery that has been lost in the ocean over time.
Check back here regularly for updates on this project.


--------------- PRODUCTION BLOG ---------------

Enjoy ;)




Well, after just 10 weeks, this short has been completed. There are still a few very minor things I would like to address at the very beginning of next quarter (such as artifacts and sound effects), but overall I feel this project was a success. It has been very rewarding working on this project, even though the scale of it was a bit daunting at times.

In an effort to get people to go to the VFX Quarterlies tonight (7pm at Arnold Hall), I'm going to wait to post the entire short on this blog. However, I've posted some stills from some of my favorite shots.



Class 15 presentations have been given. I have shots 1,2,4, and 6 finaled with shots 3,5, and 7 ready to be cranked out. There are a few minor things I'm going to tweak with the finaled shots (such as the tether being visible in the beginning of shot 1), but those are minor fixes. Chris also has shot 8 and I hope to be giving him shot 9 as well soon. I told Joe I would have shots 3,5, and 7 done for next week so I'll be busy trying to get that done. In the meantime, I've crudely compressed my current animatic so you can see my progress.

Now that all the assets have been completed, it shouldn't be too big of a deal to crank the last of these shots out. I started shot 3 just the other night and have it nearly completed. One caveat to be aware of is that I haven't paid much attention to continuity of color until recently, so that will be fixed when I time the shots over the weekend.



Work is continuing on shot 6. I've decided to go for a kind of bioluminescent look to the monster when it lights up. I have a work-in-progress still below.
I've also started animation on shot 8 so Chris can get started on the effects. I've given him the scene file and he's already converted the scene to fbx so he can use Houdini. I've rendered a playblast of the animation below.



Work has continued on shots 4 and 6. Here is a render of shot 4. Inspiration for the monster has been taken from the Alien movies. Keep in mind that these shots have yet to be comped for continuity, so the colors may be a little off from the shots around it.
I'm now ready to start on shots that have more of the monster in it. Also, I need to get the effect shots laid out so Chris can begin work on the "sucking" effect (the monster is going to pull the sub into itself).


02/08/2009 - MIDTERM

Shots 1 and 2 have been finaled with a couple fixes. I've decided to wait to final those two shots until I have a few more shots in the pipeline. This way I can further develop the look as I work and then apply the changes to those shots as necessary. I'll post these shots when they're completely done.

I've also started work on shots 4 and 6. This forces me to solidify my ideas on the monster. Here is an early image from shot 6 (still very rough).


Click here for a full resolution still.



Because Travis is still working on the textures, I have decided to forego any development of the sub shaders. I would hate to do a bunch of work and then have to redo all kinds of stuff because Travis painted a sweet texture or whatever. So instead, I've focused primarily on getting the sub's lightfog rig working properly. I've also done some early tests of look development using shot 2 (above). I've done some work with the terrain shader, but it still needs a lot of work. Also, you'll have to forgive the odd line going from the sub to upper screen right. That is the tether from the sub to the surface that hasn't been dynamically simulated yet.

I'm thinking of possibly adding a fissure in the terrain on screen left. I think that might add a bit of depth to this image that is a little lacking right now. The next priority however is seeing some progress on the sub shading/texturing with Travis and also getting an early monster asset into the pipeline.



One of the biggest components technically in this project is getting the light fog to work (and look good). I ran several tests to find out the best way to tackle this aspect of the short. The maya lightfog node seems to translate nicely into renderman, however, using more than one light causes weird problems. Renderman doesn't know how to combine two volumes of light (using maya's lightfog node) seamlessly together without canceling each other out.

maya lightfog rendered with renderman (with particles). Click for larger view.

same light, duplicated and translated. The math seems to get a little hazy here...(haha)


I also tried to use renderman light shaders. I copied the volumeFog.slo file from the shader library (/opt/pixar/Renderman_Studio-1.0.1-Maya2008/lib/shaders/volumeFog.slo) to my project directory and injected it into the RIB file before rendering. This was required in order to tell every shader in the scene to include this RiAtmosphere calculation. I then appended the volumeSpot.slo shader (found in the same directory) to a maya spot light to get the lightfog. However, I can't seem to get shadows working with this shader.

The light fog method I'm more familiar with is simply using maya's software renderer. The only downside to using maya for the light fog is that the images rendered with maya don't seem to line up perfectly with the ones from renderman. Also, I'm still having trouble getting the light fog to look believable and fall off with a nice smooth gradation (instead of seeing a ridiculous-looking cone of light). Also, maya doesn't support rendering particle points which I had planned on using for small debris floating in the water that would be illuminated by the sub lamps, but I suppose I could comp them in separately.

So now it seems I have a choice to make. I can either render the better-looking renderman fog but take a hit on having to break down every single light on its own layer or I can use the software renderer and take a hit on aesthetics. Because of the time limitations, I have decided to go with maya for this problem because I still have the hope of being able to finesse the fog into looking better than the way I have it now. Even if the light fog doesn't line up perfectly on top of the beauty layer, it shouldn't be too distracting because of the blurry nature of fog anyway. Besides, I still plan on having several dozen (if not more) lights with light fog attached for the shots with the illuminated monster, so rendering each light on its own layer would be disastrous.


Here is the maya lightfog set-up I will be using. Particles will be rendered on their own layer and comped in shake. The fog itself will also have to be rendered on its own layer.


On a brighter note, I have begun layout. The ending to the short is where I'm still a little hazy on, but I'm sure once I get the monster asset together the shots will start to come together more concretely.

I've also rendered a new turntable of the sub:
I've also begun work on the terrain asset. Because each shot takes place in an arbitrarily different setting, I've decided to design each terrain independently for each shot. Right now I just have a basic NURBS surface with a displacement map which is utilizing both GeoTiff DEM data as well as black/white photography. You can see some early (very early) displacement tests in the animatic above as well as the following from shot 2:




Modeling has been completed on the mini-sub. The next stage for this asset is texturing/shading which will be passed on to Travis Button. I still plan on having a hand in the shader development (especially since some of the objects in the sub model can be done completely procedurally). Click here for the most recent turntable.

On a side note, I've been having some difficulty with the SCAD renderfarm. It seems that the symbolic links set up on the dedicated farm servers that point to the core renderman files are broken. However, systems assures me that this problem will be fixed soon. When it's fixed, I'll have an occlusion turntable to make sure there are no modeling artifacts.

Next, I plan on working with DEM maps to extrapolate displacements for the ocean floor terrain. Also, I'm eager to test some of the technicalities with rendering with renderman, such as light fog and atmospheric effects. I also need to start collecting other students' models to build the 'monster'.



Modeling of the mini-sub has begun. Extra details on the exterior of the sub as well as equipment on the interior of the bubble have yet to be added. Here is an example sketch of the original concept work as well as the current turntable.

Click here for the turntable.



For Studio I, I have decided to produce a short film about a mini-submersible's discovery of a mysterious object in the deep ocean. This film will be completed with the help of Chris Wilson, Adam Flynn, and Travis Button. Some example sketches and references will be posted soon.

Joe thought it would be a good idea to get some early visual ideas for how the "sucking" effect might look. Chris Wilson is handling most of the effects in this project, so I've been working closely with him to define a look before researching the best way to tackle this effect digitally. Here are some examples that we have so far:


If you imagine these shots played backwards, it gives the illusion of dust and water being sucked into the ground. I think this is a good starting point for Chris to start playing with in Houdini.

A rough hand-drawn animatic has been completed in order to get a better idea of the number of shots and the general amount of problems that may be down the road that we can catch early. Once modeling is completed, a second animatic will be created in maya to get a more exact feel for timing and composition.

The VERY rough animatic can be seen here.