Door Unwrapping

After finishing the final vine, I decided to deal with the last problem before moving on to UV unwrapping. This problem with the mesh was the bunching up of edge loops caused by the extra geometry need to add the leaf onto the tree. This wouldn't be much of a problem, but the mesh is of a smoother, more organic shape and I would prefer to have only quads on the mesh. Eventually, I decided to delete the entire edge loop up to the leaf itself and then split the, now 5 sided, polygon where the leaf meets the tree in half. This left a tri, so, realising I had just enough space with out causing the edge loops to bunch up again, I created a short edge loop that added an extra vert on the tri, making it a quad. This seems to have had the desired effect.

I next moved on to unwrapping the roots that make up the doorway. Being relatively new to UV Unwrapping I decided to start by faces that made up a visible area and then planar mapping them. Once I had done this with all the polygons I began attaching as many of the shells together as possible to make larger shells that would run smoothing from one to another. Although time consuming and sometimes tedious, I did not find this difficult.

 I am fairly happy with these UVs, although I am a little annoyed that one piece will not attach and stay in scale at the same time (the large piece on the top right). The only reason this annoys me is that it leaves a visible inscision. I am going to have to be careful when texturing not to leave a visible line.

The next thing I unwrapped was the vines. These were especially difficult due to their awekard shape. I figured the best way to do this was to use Automatic Mapping and then straighten up the pieces with the align tools. I then sewed the edges together to make a rectangle, making sure that the edge that was not attached was on the side that is not visible. I then put the ends in the corner of the map, since they are not visible. The mushrooms were fairly easy, The "fruited" mushroom I used Automatic Mapping to create the pieces and then stitched them together so that the incisions could not be seen. The closed cap mushroom I used cylindrical mapping to get the majority of the mesh mapped and then planar mapped the end.

The final thing I needed to Uv was the door. After looking at the Uvs I found that all I needed to do mas manipulate them into a more accurate size.

The only remaining task is to texture, light and render. I have condered making two versions. One that is lower poly and can be taken into a game engine and another that is higher poly and looks smoother.

Armed & Dangerous: Base Mesh Completed

After a couple of weeks, I have finally finished the base mesh for the "Six Armed Nutter". The production of the arms, torso and legs were all pretty straight forward. The biggest problem I had was with the head. Because of the cartooned style I have chosen to use, his mouth is out of realistic proportion. This made it very difficult to work out the shape and topology using the box modelling technique. I ended up using the edge extrusion technique. I found that, when using this technique, it was much easier to work out the correct topology than if I were to use box modelling methods.

(Left) Edge Extrusion Head. (Right) Box Modelled Head.

As shown above, the head on the left has a mouth that better matches the original proportions shown in the concept art, although the topology is a bit messier than I would ideally like.
The body, arms and legs were all made with boxes and cylinders with their werts moved into the correct position.
After the model was finished, looked at it in subdivision preview mode and found that random faces appeared and other faces detatched from not being merged properly.  Most of this I managed to clean up with Maya's clean-up tool. The rest had to be manually deleted and merged into it's intended place.

Knowing I would be painting the model in Mudbox and not taking the UVs into Photoshop to edit the textures I felt it was save to create the UVs with Automatic mapping. Automatic Mapping flattens out the UVs in the quickest way possible whilst not having anything overlap. If I were taking them into Photoshop, I may then take the time to straighten up the UVs to be clearer.

Once this was done I found it safe to take the model into Mudbox. However, after taking the model into Mudbox, I found that some of the geometry creased in certain areas. Mainly around the forehead between the eyes and at the crotch. This was due to having too many edge loops running throught the middle of the mesh. I had added these when I had miscalculated the number of edge loops need to attach the head and trousers to the torso. However, now that the meshes had been combined, I found it fairly easy to delete a few of these. Only with one did I have to rearrange the geometry around the chin to effectivly get rid of it. This also, by chance, removed some of the polgons making up and meeting at the center of the eye. There is still a 5-pole in the centre of each, but I am considering sinking these into eye sockets and then using 2 spheres inside each other for the actual eyeball.

Front and Side view of the final base mesh

Depending on how the sculpt goes, I may model the hair in maya and then bring it into mudbox. I may also make some weapons for him, since the them of the competition is "Armed and Dangerous".

Topology

Having finished the sculpt I then needed to finish my final piece of work for the sculpting project, topology research. I decided my first task was to find a definition of topology. My first stop on my researching quest was the forums of cgsociety.org. Someone had already asked what the definition was in this (http://forums.cgsociety.org/archive/index.php/t-416197.html) thread. There were two replies which interested me. The first was a more literal definition using the greek words that combine to make the word "topology". "Topos" means "place" and "logos" means study (newellteapot, 2006). Topology literally means the study of places. The second definition was, "basically for organic mesh you will want all quad using edge loops to define muscle and bone of the mesh. Edge loops function as defining the lines/curves/creases of an organic mesh. Also, it is important that all polygons have a similar density and size in the mesh for an efficient and accurate subdivision and deformation" (MasonDoran, 2006).
The user goes on to say that the topology need not be so strict for non-organic meshes such as building, vehicles or mecha. Although attention should still be paid to the size, shape and placement of the polygons if the model is to be subdivided or distorted.
To sum up, topology is the study of places, in 3D modelling it is the study of the "placement" of polygons to ensure proper subdivision and distortion.

After finding the definition I then needed to look further into what facial topology should look like, and why. After a bit of research I found this website (http://www.phungdinhdung.org/Studies_paper/Realistic_face_modeling.shtm#3.2-%20Topology:) which explained that the topology not only needs to define the face, but also needs to be able to deform in the same way the face will when animated. This means that the topology can change from face to face depending on the personal preference of the artist, the shape of the face itself, the art style and the animations the face needs to do. After seeing some illustrations on this page where the artist had drawn over some photographs to find deformations in the face, I decided to do the same. Luckily, when photographing my model, we took a few mock photos where the model pulled a few faces, this was perfect for figuring out how his face deformed.

Using these pictures, and illustrations from Stop Staring; Facial Modelling and Animation Done Right by Jason Osipa plan out a very rough topology for the 3D sculpt.

Although this rough plan wasn't perfect, I saw this as enough to start work on the 3D sculpt itself.
I am fairly happy with how the topology of the sculpt turned out. However, knowing that triangles do not subdivide or animate well, I had to create a couple of 5poles (a vertex where 5 polygons meet) to avoid them. I ended up with three on each side. This is 2 more than would have been ideal, since having at least 1 5pole is virtually impossible when avoiding tri's. Another small problem I think I had, is that I sometimes had to split polygons to ensure that the quads (a four sided polygon) of one edge loop (an edge that loops back on itself; these are useful for connecting distant areas connected by muscle, such as the eyes and the area above the mouth) remained a similar size. I would much rather keep every edge connected by edge loops, but this would have left some polygons unbalanced in size.

Although I used a book for reference I did not copy. Even though this means I may suffer for it. I would rather attempt topology without copying, fail and learn from my mistakes, than copy the topology and learn nothing about working out topology myself.