Continuing on from last time where we talked predominately about developing a character for one of our current projects, the character concept was further refined through several feedback sessions involving both facilitators and peers. The feedback received was, in a nut shell, simplify and brighten the colour pallet, move to a more simple style with the stripes and details in general and push the proportions a little further. This along with some other passing comments and talks with facilitators lead to Fimu3.3, as seen below.
Comparing this final concept to previous iterations, the painted style is gone, the wing is gone, the detail in the stripes/feet/eye are gone, leaving us with a more refined and style appropriate design. The next stage is to create a low poly model of the character suitable for the brief (500-1000 triangles, up to 2×256 texture maps, topology suitable for animation). So my process for converting the adorably simple Fimu into a deceptively low resource cost 3D model is as follows. A development note here, I did not create a model sheet for Fimu as his shapes and proportions can be broken down into simple primitives allowing me to save time in that stage of the development process to be relocated to other areas of production. The first step I took was to break down his shape into its basic primitives in 3DS Max and adjust their proportions until they were suitable for the design. I ended up with 2 spheres, 2 cylinders and a box, corresponding to head, body, legs and feet.
The second step I took was to align everything in 3D space appropriate to the modelling process, allowing me to then sheer off one half of the spheres and apply a symmetry modifier to the model to reduce the workload of modelling by half. Third each primitive was “sculpted” (not sculpted as in high-poly-super-detailed-3D-sculpting-I-don’t-know-how-to-do-that-yet sculpting, just moving vertices into the shape they need to be sculpting) into a more refined shape and then joined with each other. This left just one job for the modelling process: optimising polygon efficiency. When creating a basic sphere in 3DS Max it creates a lot of unnecessary topology which was cleaned up yielding a total of 568 triangles for the model. Success in my books.
Above is an image of the current model and its topology. I then took this model and unwrapped it (only the one side needed to be unwrapped, the symmetry modifier did the rest of the work) producing the below UV maps.
Having unwrapped the model and collapsed the modifiers I then exported Fimu as a .obj file and took it into Mudbox to 3D paint giving me the following result.
Happy with this outcome I then took the produced texture maps and applied them to my 3DS Max model fully expecting some wonderful results, however instead of the gleaming, polished and finished texture I expected, I was confronted with a host of problems and weird artefacts that I haven’t seen before. Here they are, all circled in blue.
At this present point in time I am not entirely sure as to my plan of attack to tackle these current issues. Although I am glad that when I take a model into Mudbox from 3DS Max it points out all the places where something has gone wrong so that I can easily fix them, when taking textures back from Mudbox to 3DS Max I have yet to encounter this issue.
I would assume that my next port of call is to research the issue and try to find a solution to the problem and failing that, consult with one of my facilitators.
All right, take a break, grab a drink, maybe a snack.
Back?… Wonderful. Onwards then.
Along side this character development project I have been working on creating a ‘before and after’ model. The idea was to create a skeletal hand, not particularly worried about polycount, and texture it to the best of my current abilities and after that to research and practice before re-texturing the model to my new and improved skill level to give a quantifiable measure to how far I have come over the course of the semester. So without dragging this out too much here is the results of what I have done for the before hand.
The first of the above images is in Mudbox where I painted the model and then hand-detailed (no pun intended) a bump map to give the bone some definition and nicks and what not (about 2 and a half hours of pains-taking hand painting). I then took the produced maps and applied them to the 3DS Max model and produced the second image. Much to my dismay, it appears that the bump map is having the reverse effect, raising the details up instead of indenting them into the bone. Yet another mystery to be solved.
Anyway that is a mostly complete update on the 3D side of my Studio 1 work.
Thank you and ’till next time…