Finally HD 720p. Click HD button or watch on youtube. My laptop doesn’t even run the video smoothly without an optimized encoding.
Generalist visual effects from a limited selection of CG work (r03), as a “student thesis” for NYU 2009. We don’t actually have a thesis, just optional classes and projects, only attendance is required (yep college headcount). Projects from Star TV, CFs, and other freelance gigs.
See my other videos for full length animation and tools presentation.
HD download & descriptions on my site: gophercg.

Version r03 is much more organized, newer clips, timing, by subject (kind of?). A pure effects reel is usually under 2 minutes, and animation showreel well above 3 minutes. I’m hoping this generalist demo is not too long.

3 minutes is pretty short, so all my clips feel like a difficult quantity to prune down. Next stop would be to focus and raise the quality. I mentioned the work permit issues here, and keep getting interviewed for some reason, so yes, I’m looking for a (legal) job!

Melscript written for a director awhile ago to test out trail generation with nurbs. It’s much easier in Houdini, but remains a good mel exercise. The VFX tool also updates the trail on demand if character animation changes.

Some recently finished modeling. A wheel and a baseball, sounds boring eh? Well not at all, it’s quite challenging.
Previously I’ve done mostly organic modeling (and not even high detail), where topology’s more important near detail areas, and each vertex (for subd or smooth) is more forgiving. Hard surface on the other hand, can have parts with careless topology, like flat areas. But once precise curves and flats must join, every detail area needs both perfect topology and vertex location.

The wheel somewhat references a Lamborghini Reventon Roadster, since the 2009 version debuted with a strange looking wheel, actually similar to some older Reventons. A problem was lack of reference photos for the wheel (there are only a handful of unique Roadster 2009 photos online). But I wanted to challenge the features, not the exact positions of shapes, so the details turned out nice.

The tire treads are a challenge, since I wanted connected treads, separate extrude treads, and even circular treads (inspired by cloud patterns on JP wear), besides very dense fillet. Everything turned out well, except the top cap of some treads. The tire process lends itself great to procedural/history based modeling, which Maya does to a limited extent (I had to deal with a messy node graph and reconnect nodes and hope it works). And it switched halfway from a Nurbs model to a dense poly, so unfortunately some top caps are separate geos and have gaps with the sidewall fillets. I planned to keep all nurbs, but SW and MR render with the highest nurbs settings couldn’t remove visilble gaps at trims. The best I got was random black dots along the seam. Ok, still learning Nurbs, maybe Rhino or some methods can give correct nurbs rendering.

The steel rim, spokes, center and bolts are also by a Nurbs -> poly process. Nurbs was great setting the precise curves and exact cuts. Though I’m still gauging when the poly conversion should best occur.

The baseball also took lots of experiments, and now it’s curvatures are precise to baseball spec, and all Nurbs. The trims get away because the render’s small enough. And the 108 stitches are fairly light weight since it’s paintfx.

Slideshow:
>

Or see all images:

First visit to JP, Kyoto, Osaka, Nara and the neighborhood. Learning a lot of differences between Taiwan and Japan, also a real visit enlightens my previous cultural preconceptions of JP (like from anime) and subcultures.

If you’re into fashion (I’m not quite), it’s nice to observe how traditionally the Kansai citizens wear in greytones, compared to the more kiddie colorful Taiwan–more influenced by American jeans, or the classy leather, sunglasses, thick-layered New York. Within the black,grey,white scheme, Kyoto wear has more expression in elaborate patterns, clouds, squares, Chinese classic shapes. Same goes with the temples, though there are bright orange facades (開?) and well groomed gardens with color to accompany.

Click the pic for my 14 tourist albums:

I’m still freelancing for now, and thought a 13 second ID is cakewalk! Well, not quite, but was doable. Can’t speak of specifics yet, but a 1 person do-all faces lots of issues I thought were easy.

Transparent mapped polygons and shadows have lots of issues.
Set shader shadow attenuation=OFF. Shadows (ray) cast are affected by angle of surface normals, so it must be off for any objects like architectural people boards.
Reflections are easy with useBackground, but shadows were a pain. I’m sure there’s a better method, but my shading experience is limited now.
useBackground for a groundplane doesn’t work, because this shader doesn’t have the common shadowAttenuation. So I get weird clipped out shadows and the ugly poly shadows. A makeshift method is use a normal shader instead, but colored. Key the color out later. It’s not great because the ground borders shouldn’t be visible in shot (curved up corners like a greenscreen set), and real uniform light for the plane is needed, since ambient/incandecence and other tricks of uniform color washes out shadows completely.

So far it’s all raytraced shadows, and soft shadows work but slooow. Although I should give mentalray a try, I need to hack my way in a short time and a test render shows mr shadows with the exact problem maya has.

A last issue I couldn’t fix with uping reflection limit or any other settings is when transparent mapped planes intersect, the shadow cast from that region is gone. So we get nice shadows with odd poly shapes cut out.

I’ve also defined an asset for “ambient occlusion” between these transparent planes. It’s really just final gather. But the boards were background elements so raytraced shadows sufficed. There are actually two ways for transparent AO (mentalray AO only supports plain geometry). One is normal FG, very easy and no special shaders. The other is the mr shader network you’d find if you google it. The FG shader seems more predictable, but I haven’t tested enough to compare their quality. I just find the simple FG has more varying shades (noise?). Theoretically AO is a white environment subset of FG, but getting FG to have clean results for sharp edge shading on white needs quite some tweaking.

Do I like freelancing? Well, it’s fun, sometimes. And exhausting, always. Don’t think of earning, because after you’ve done a few months, you’d be glad to earn a fraction of a month’s rent. I “landed” two official jobs recently, stable pay, insurance and all, but too bad this country’s foreign worker regulations make it impossible. What a shame. It’s a pretty island. If it doesn’t work out…well then it’s Singapore, Canada, EU, Hong Kong!

After much encoder wrangling, the screencap cam out.

BIRD RIG: wing & leg

Shows parts of the rig, wings & legs. The challenge was the character required more cartoonish flexibility, and while working non of my colleagues made a bird rig before so I was left exploring whatever problems I’d find.
Im quite surprised why I made 60+ shelf scripts (mostly MEL) just for the BIRD. And a few arent just short tool scripts. I’d need to organize and improve the scripts/rig into a better package as usual.
An “industry standard” is often to rig/animate in Maya, but keep the feathers and its complex motions to Houdini particles. Here I’m just attempting a simple & fast maya feather method. But because most renderers slow down with numerous transparent geometries, the sw isn’t too fast. I can try tweaking the renderer if feathers can look good enough for the bird.

Next up is the facial rig, which used the majority of script tools.

VineGrowth fixed

Also fixed the vineGrowth presentation, pardon if the previous vine video caused more seizures than Pokemon!

i wonder why screencaps are harder for me than coding and fixing 3d? anyways. I’m organizing the old bird rig into a screencap presentation. Publicizing works is quite important, as it keeps me more organized and on task. I’m still quite surprised why I made 60+ shelf scripts (mostly MEL) just for the BIRD! And a few aren’t just short tool scripts.
Hoping this video turns out well, I’ll do a first pass into revisiting just what this bird does….stay tuned

Screen capture of a few tools developed for past projects
It is similar to a previous maxscript I wrote for recursive plant growth. But since maya has paintfx, the recursive part and manual curve shape designs are not coded into the version for maya. However I added more flexibility with an easier UI.

Future work: more integration with Maya’s other modules besides paintFX.

Website! Yay cleaned up the homepage, mostly CSS details, and clean graphics (gophercg). I still prefer the minimalist feel (maybe major turn-off by making a text console style LOL), but now it’s all pleasant and tidy. As for the grid in galleries I know how to use divs/spans for the presentation, but so far my site is fine in firefox and IE, so before I decide on a pure flash site, I’m keeping em tables! Image galleries are converted to Flickr sets too.

At the same time I’m trying to focus on a career path, and learning a wider range of CG/web. I don’t get my oxymoron situation. It’s something in the realm of null, 0.? and 1+. I’ve too many 0.? or half-baked skills, and some have a negative apparent value. So I’m planning to level up my website to acceptable levels, since it’s a required part of either CS or CG. And at least know how to create cloth (on character) & hair, basics of effects. Then coding and houdini could be worth training for ++ levels.

I forgot about modeling…ok maybe some more of everything…