9 September 2004

The time is now 10am, and I have about 23 hours before this journal is due. Looking over it all, i realise i need something *big* to top it all off. Something that will require I use all the skills I have learnt this semester. And so, I give you...



To do this, I will need to first model the character. Then, I will need to optimise it for animation, setup a bone system for it, rig and weight the bones, and finally do some kind of semi-believeable animation with it. A simple swim past the camera should be enough to show that it works.

To begin, I first searched around for something that would show me how fish are supposed to move through the water. In the end, i found this immensely useful movie (14MB): here.

It became apparent that I would need to construct my shark so that it could bend horizontally. That was the main concern. Here's a quick GIF animation of the kind of swim I need to animate:



I decided the best way to approach the character's design was to model in lightwave and convert across due to my lack of knowledge about the maya modelling tools. This meant that whatever modelling I did, it would probably have to be converted to polygonal geometry before I converted across. I thought it would be best to assume so as modelling this way sometimes feels easier for me anyhow.

To begin modelling, I constructed a low-poly model so that I could easily tweak the mesh's overall look. I also tripled the geometry to check for unwanted creases. The low-poly model was created with simple polygonal modelling commands such as extrusion and beveling so i won't go into it.



The first thing I added a nicer, smoother look to was the rear tailfin. This was done by subpatching the low-poly mesh and tweaking control vertices. I then froze the geometry to give this mesh:



This shot shows the newly created fins being mirrored to avoid having to do more work. I completely scrapped the low-poly geometry for this area as it was too undetailed.



I cut an inset into the higher-detailed face mesh for the eyes, then used them as a different surface so that it would be easier to apply materials to them in maya.





This shot shows the model semi-completed. Most of the geometry was created with subpatches (NURBs cages), while some was created by boolean additions - for example the ridge running along the side of the shark. The blocky fuel cells? or whatever they are at the back were created by modifying polygonal spheres and then subdividing the faces with a smoothing algorithm to give a semi-rounded look. Note the blue lines around the teeth at the front which show the NURBs cage around the derived geometry.



I began applying smoothing groups to the various surfaces of the model to make it look nicer. In particular, note the smoothing around the small fins achieved by moulding the geometry together instead of just letting it intersect.



At this point, I would like to point out the progress of the model from the low-poly guide mesh to the high-poly version. This animated GIF should show the differences.



I added some insets to the rear boxy-thing. No particular reason for it, execpt that I thought it might look interesting somehow and break up the smoothed geometry.



I then saved the file one last time and went into maya and imported the model with its part hierachy intact, an added bonus as it would mean easier setup. Read on to see how i went with setup.