Sculpting Details onto a Model
I wrote a series about how I used semi-parametric design in Blender to make the Living Bones model from Gloomhaven. Here's the final product that you can download from Thingiverse:
The bulk of the model was crafted semi-parametrically. The main shape for the thigh bone, for example, is actually just 4 vertices in a row forming a single line. I then used the Skin modifier to give it depth and changed the size of the bone at each of those 4 vertices. Almost the entire skeleton was formed this way, which allowed me a lot of flexibility while I got the pose into a shape that I liked and I figured out how long the limbs needed to be in relation to the everything else. That said, there are a lot of details in there that I could not make parametrically; once I had the main shape in place, I used Sculpt mode to add those small details.
For example, I had no idea how I could possibly make the skull via parameters. So, I did some quick googling and came upon this guide to sculpting a skull in Blender. I did my best to follow their process and came out with a pretty serviceable skull model. There's nothing parametric about that process, but I think it makes sense to use the tool that's best suited to each job. Since I liked the experience of sculpting the skull, I decided to see if it could add detail to the rest of the model. Here's what I had put together without the sculpting (except for the head and feet, which are 100% sculpted):
See the difference? I was able to add a lot of detail to the loin cloth, the straps and even some of the more prominent bones. In each case, I applied the Multiresolution modifier to my shape, then pressed the Subdivide button (on that modifier) until it looked like I had a decent number of faces to work with. After that, I changed into Sculpt mode and made liberal use of the Draw tool, either Adding or Subtracting via that tool to create the details that I wanted. I also found the Smooth tool to be invaluable, to make things look more natural.
I felt like the sculpting process was very intuitive to use. I started with the larger sculptural details and, as I added smaller and smaller details, I just pressed that Subdivide button more times to break up the model into ever smaller faces. When I was done, I could apply my modifier and end up with a nice, uniform mesh that had all of the details that I wanted. If I really messed up, I could always delete the modifier and start over. Why use muliresolution instead of subdivision surface? The difference between the two is basically how real the change is.
A multiresolution modifier stores a lot of data. It stores the subdivision and vertex manipulation as it relates to the source object, and it needs that source object to be fairly real. A subdivision surface modifier, on the other hand, is entirely calculated based on the source geometry. You can't change data for the subdivision surface itself, only change the underlying geometry to change how the computer calculates the subdivision surface (or tweak some of the parameters of that calculation).
In practice, this difference became meaningful to me because of my use of the skin modifier. I can throw a subdivision surface modifier on top of my skin modifier to make the skin nice and smooth. The skin's geometry doesn't actually exist (until you hit apply) though, and so you can't have a multiresolution modifier applied to it. There's nothing there for it to build on.
So, in case you were wondering how I got those extra details onto the model, that's my process!
The bulk of the model was crafted semi-parametrically. The main shape for the thigh bone, for example, is actually just 4 vertices in a row forming a single line. I then used the Skin modifier to give it depth and changed the size of the bone at each of those 4 vertices. Almost the entire skeleton was formed this way, which allowed me a lot of flexibility while I got the pose into a shape that I liked and I figured out how long the limbs needed to be in relation to the everything else. That said, there are a lot of details in there that I could not make parametrically; once I had the main shape in place, I used Sculpt mode to add those small details.
For example, I had no idea how I could possibly make the skull via parameters. So, I did some quick googling and came upon this guide to sculpting a skull in Blender. I did my best to follow their process and came out with a pretty serviceable skull model. There's nothing parametric about that process, but I think it makes sense to use the tool that's best suited to each job. Since I liked the experience of sculpting the skull, I decided to see if it could add detail to the rest of the model. Here's what I had put together without the sculpting (except for the head and feet, which are 100% sculpted):
See the difference? I was able to add a lot of detail to the loin cloth, the straps and even some of the more prominent bones. In each case, I applied the Multiresolution modifier to my shape, then pressed the Subdivide button (on that modifier) until it looked like I had a decent number of faces to work with. After that, I changed into Sculpt mode and made liberal use of the Draw tool, either Adding or Subtracting via that tool to create the details that I wanted. I also found the Smooth tool to be invaluable, to make things look more natural.
I felt like the sculpting process was very intuitive to use. I started with the larger sculptural details and, as I added smaller and smaller details, I just pressed that Subdivide button more times to break up the model into ever smaller faces. When I was done, I could apply my modifier and end up with a nice, uniform mesh that had all of the details that I wanted. If I really messed up, I could always delete the modifier and start over. Why use muliresolution instead of subdivision surface? The difference between the two is basically how real the change is.
A multiresolution modifier stores a lot of data. It stores the subdivision and vertex manipulation as it relates to the source object, and it needs that source object to be fairly real. A subdivision surface modifier, on the other hand, is entirely calculated based on the source geometry. You can't change data for the subdivision surface itself, only change the underlying geometry to change how the computer calculates the subdivision surface (or tweak some of the parameters of that calculation).
In practice, this difference became meaningful to me because of my use of the skin modifier. I can throw a subdivision surface modifier on top of my skin modifier to make the skin nice and smooth. The skin's geometry doesn't actually exist (until you hit apply) though, and so you can't have a multiresolution modifier applied to it. There's nothing there for it to build on.
So, in case you were wondering how I got those extra details onto the model, that's my process!
Comments
Post a Comment