Finishing the asset

I`ve continued using the same technique on all of the different elements. Clay Tubes + Blob + Flatten and a bit of hPolish here and there. The aim was to create detail in a logical manner. I`ve tried to imagine the story of this piece of architecture.. how it might have started aging, what might have been the causes etc. For example: if a brick has a huge chipped off part on its lower side then you should continue or add some destruction in the upper area, as a continuation to the brick below it. Anything that caused the upper bricks destruction surely did not stop where the brick ends and continues on the one below it.

I had to be aware of the maximum polygon limit which still provides decent frame-rates. That threshold for my computer is around 20 million. To really get all those cracks in there, I`ve calculated that one asset should not go over 1 million as I have 20 elements. Exception can be those surfaces which are bigger, thus more important for the final image.

After the high poly model was done, I created a low-poly version with the Decimation Master plugin, found within ZBrush. It's a very powerful tool as you can convert your high poly mesh to a low poly one on the fly. It also allows you to set a specific triangle count, so you don't have to tweak it a lot after the fact.

Next step was to unwrap the low poly model.
I`ve loaded the low poly version into Headus UV Layout. Whenever I have a bit more complex model to unwrap, I use Headus for it, because the workflow of it is very fast. The relaxing algorithms it uses are just way ahead of that of 3DSMax or Maya. Unwrapping actually took no more than 10 minutes, whereas it easily would have taken up to an hour in Max.

I`ve took the low-poly mesh into 3DSMax to fix the topology up a bit.
I removed several edges that had no effect on the silhouette of the object whatsoever. The surface details will be dealt with by the normal and the color maps, so there is really no need for geometry there, except for some bigger cracks.
I`ve ajusted the smoothign groups again, because the unwrapping process for some reason smooths out the entire mesh as a whole, which gives very bad results. I`m sure it is the settings in Headus which is doing this, I will have to look into it.

Texturing was done with the help of XNormal, Photoshop and nDO2. I`ve baked the information from the high poly onto the low one in XNormal and then took those maps inside Photoshop. In Photoshop I`ve loaded up nDO 2 and generated several maps (e.g. cavity map) to further bring the small color details to life.

For the final render I`ve used a somewhat low angle camera shot to try and give the impression that this is a big object. Looking up on something gives that feeling of it's quite big... or at least not small :) .

Here is the final image:

ZBrush work

After the blocking stage was complete and the necessary topology was set up, I could take the assets into ZBrush to begin detailing them and creating the high poly version.

I`ve decided to challenge myself a bit more and instead of the clean surface these bricks have, I aimed for an old, ruined like look instead. This allows me to look at the various effects that different brushes have.

I`ve started detailing the column:

I have used the Standard Brush and the hPolish for the column you see above. I`m really not happy with the result as the mesh looks to smooth and "fluffly", instead of harsh, stone chipped off look.
In one of the Environmental tutorials I have watched I remember a technique where the Clay Tubes brush was used to establish the main details and then the Blob Brush came into action with a small brush size to give the surface a bit of noise kind of, roughed up effect. As the last brush, the Flatten brush was used with a square Alpha to trim away the noise and establish that harsh, stone detail.

Here is the result:

I have found that this combination of brushes work much better than what I had used previously. Also, even though silhouette is absolutely the basic element to keep in mind when modeling, surface detail is also crucial. If your surface detail (modeled or illustrated via a texture)is not properly done, than your model might end up something completely different than intended. Some colors make the detail pop out a bit more. Because I have a greyish specular color at certain area, using a grey color when rendering the object is not the best idea. It will blend in with the specular and make all those small details almost invisible. I chose a different color for these second set of images and it really made the detail pop out a bit more.

Putting the new workflows to the test

To practice and test the new knowledge, I will be re-creating the object you can see on the image below. I`ve chosen this reference as it has various shapes which can be found on bigger architectural pieces. It has cubical elements, along with some arched parts and also some more organic shapes on the top (flower decorations).

Plan:  - create the main block out in 3DSMax

         - build base mesh with sculpting ready topology in 3DSMax

         - subdivide in ZBrush and sculpt high poly mesh

         - use the Polypaint feature in ZBRush for texturing (with high attention to color theories)

         - create a low-poly version with the Decimation Master in ZBrush

         - Unwrap the low-poly model with Headus UV Layout

         - use XNormal to bake the mesh from the high poly model onto the low poly

The reason behind this workflow is that it allows me insight and practice in game-asset creation (where polygon count and resource management is key) as well as in cinematic/film work (due to the high polygon modeling).

Focus will be on merging the theoretical elements (color, form, silhouette) with the technical ones'.

That said, I went ahead and "deconstructed" the reference image so I become more familiar with the structure of this piece of architecture.

Blockout mesh in 3DSMax:

I`ve set up the model with proportional topology by setting up Smoothing Groups and using the NURMS subdivision method based on these groups. This way, I can get a pretty high polycount and maintaining the silhouette of the model. If I hadn't used Smoothing Groups, then the model would have curved in at the edges, becoming too smooth.

By using Smoothing Groups I get a high enough polycount so when I subdivide the mesh in ZBrush, the edges will not become too rounded off. I could use the Crease feature, but that takes more time, plus it also will produce crease at unwanted places, thus I would have to "clean up". This method takes too much time and result would be not even as good as with the Smoothing Group method.