October 30, 2009
After edge-looping and subdividing about 30 percent of the walls, I decided to go another route. I found that this approach was taking a lot of time because the subdivide multi fractal was creating non-manifold meshed which I had to fix by hand. In addition, this was adding an insane number of polygons to the model.
I should be able to get a good effect by using a normal map for most of the walls. I reverted to an earlier version of the file and made a few corrections that I had run across during the subdivision work.
So, at this point, I am declaring the remodeling work completed and am moving on to texturing. This is where I believe that I can make the biggest improvements over the original model.
October 25, 2009
The model shows that some of the stone walls are ‘ruined.’ That is, there are portions in which there are missing stones. In these places, the wall should appear fractured. In the original model, these places were modeled with very simple topology. For the new model, I want to show more detail.
The technique that I have decided to use is to add more vertices on the faces of the fractures by using Blender’s Subdivide Multi Fractal function. In order to avoid affecting the geometry of other portions of the wall, I am first adding an edge loop around the section. This isolates the rest of the wall from the addition of more vertices on the interior faces.
October 22, 2009
I finished sculpting the ground topology which in general falls off to about 7 m towards the San Antonio River in the West and South. The North falls about 1 m from the elevation of the church and the East is about the same elevation. (All within the area of the ground plane grid.)
October 18, 2009
I completed stitching together the dirt piles and ditches to the ground mesh. Along the way, I learned how to minimize the artifacts due to poles in the topology.
In general, poles are no problem if the mesh is not smoothed or there is no deformation associated with a vertex that is connected to the pole. The solution is to surround any deforming mesh like a ditch or pile with an edge loop that is at ground level. Then the vertices comprising the loop can be connected via poles without causing any artifacts in shading.
My next task is to deform the grid to conform to the overall topology of the land. I may also work on improving the smoothness of the dirt piles around the Lunette.
October 3, 2009
The ground object consists of the dirt and fields around the compound. It also includes the pond/marshy area and the acequias. In addition to the poor textures in the original model, the ground was modeled completely flat. The piles of dirt in and around the compound were also modeled as separate objects and have the appearance of being ‘placed’ and not a part of the same material. The original model also used a different material on the faces that represented the roads from the fields. This produced a sharp and unnatural delineation.
I first simplified the geometry of the original model by separating the depressions (acequias and ponds) from the surrounding terrain. The geometries of these depressions were also simplified and made more regular. I then applied a 32 x 32 vertex grid overlay and cut out the portions where the depressions are. This grid will allow me to sculpt some elevation detail, including the slope down to the San Antonio River to the east as well as a gradual rise in elevation to the west.
After joining the depressions to this grid, I will make the connecting faces. Another advantage of using the grid, is that the long runs of triangle edges should be eliminated or at least minimized.
The final step will be to add to this mesh the dirt piles.