Radiolarian
03-17-2005, 05:52 PM
These are a series of models that I'll be working on to improve my workflow and modelling skills. Some of these models will be just practice sessions, others I'll improve and texture. I would like to hear your opinions not only about the models but also about the methods that I'm experimenting with.
The first model was an experiment in speed modeling. The idea was to create a single mesh that would serve as a starting point for many models. The advantage of this would be that a single rig would animate many characters. Which is what Taron does on his people movie. And it works up to a point. I did this model in a couple of hours based on the initial mesh made in wings. All the face detail is modeled in Zbrush, that part in the original model is just a flat grid of polygons with no edgeloops. I thought that this way I could model a cyclops or a five eyed monster with 3 mouths if I wanted.
The problem is that you do get constrained if you start with a detailed model. A lot of the character in a model is in the proportion and posture of the body. With this technique all my characters would look like people in rubber suits. I still think is an useful technique to have in your arsenal but not as useful as I expected. Back to the drawing board for the next one.
On thing I did learn is how to improve your time using the Smart Resym tool. It used to take me more than two hours to smart resym a 4 million poly model on my computer. I found a way to do it in less than 10 minutes. What you need to do is to smart resym the mesh at a much lower subdivision level, then do it at the next level, and so on until you do it at the highest level. This will cut the time dramatically. Try it!
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Uno-Combined-Render.jpg
The second model was done completely within Zbrush. This method is slower but I have a lot more freedom in the development of the character. The clean topology was done by adjusting the resolution of the zspheres at key points: hands, feet,and head.
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Dos-Face.jpg
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Dos-Body.jpg
This is the third in the series. I started with zspheres and sent it to wings to improve the topology around the legs and arms. I also made sure that the model would reach the 3 million mark if subdivided. I'm trying to see what kind of details I can do with different amount of polygons. In the end it was a limiting factor in the face to do the kind of detailing I wanted to do. I also made things difficult for myself putting a lot of polygons in the eye area on the first subdivision layer.
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Tres-v2.jpg
The first model was an experiment in speed modeling. The idea was to create a single mesh that would serve as a starting point for many models. The advantage of this would be that a single rig would animate many characters. Which is what Taron does on his people movie. And it works up to a point. I did this model in a couple of hours based on the initial mesh made in wings. All the face detail is modeled in Zbrush, that part in the original model is just a flat grid of polygons with no edgeloops. I thought that this way I could model a cyclops or a five eyed monster with 3 mouths if I wanted.
The problem is that you do get constrained if you start with a detailed model. A lot of the character in a model is in the proportion and posture of the body. With this technique all my characters would look like people in rubber suits. I still think is an useful technique to have in your arsenal but not as useful as I expected. Back to the drawing board for the next one.
On thing I did learn is how to improve your time using the Smart Resym tool. It used to take me more than two hours to smart resym a 4 million poly model on my computer. I found a way to do it in less than 10 minutes. What you need to do is to smart resym the mesh at a much lower subdivision level, then do it at the next level, and so on until you do it at the highest level. This will cut the time dramatically. Try it!
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Uno-Combined-Render.jpg
The second model was done completely within Zbrush. This method is slower but I have a lot more freedom in the development of the character. The clean topology was done by adjusting the resolution of the zspheres at key points: hands, feet,and head.
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Dos-Face.jpg
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Dos-Body.jpg
This is the third in the series. I started with zspheres and sent it to wings to improve the topology around the legs and arms. I also made sure that the model would reach the 3 million mark if subdivided. I'm trying to see what kind of details I can do with different amount of polygons. In the end it was a limiting factor in the face to do the kind of detailing I wanted to do. I also made things difficult for myself putting a lot of polygons in the eye area on the first subdivision layer.
http://img.photobucket.com/albums/v205/Radiolarian/Zbrush/Tres-v2.jpg