Final Update (2012-03-25)

And there we are, the project is over! I just thought I’d make a final post to say that I had a blast doing this character, and certainly I learned alot!

A quick screencap from within Unity;

Not much left but to offer up my final playable version, post mortem, final presentation and a short video clip for those who doesn’t have time to play it.



Progress Update (2012-03-14)

Been a while since my last update, and I’ve gotten quite a bit of work done too! Took longer than I had originally anticipated though, but here’s my low(er)-poly character at a grand total of 14993 tri’s, just under the 15000 limit I had set for myself. What’s left on the character is baking maps and sorting out texturing.

I’ve also retopologized, UV-mapped and baked normal/cavity/occlusion for the sword & shield. I should mention that they are not included in the character polycount. They are 404 tri’s and 388 tri’s respectively.
Next up is baking normal, cavity and occlusion for the character itself. After that I’ll be doing some quick tweaking to the maps if neccesary, aswell as texturing the character, sword and the shield using the occlusion/cavity maps as a base. Not much time left, but it’s all coming together at long last!

Note that the character maps will be at 4096×4096 and the sword & shield at 2048×2048.



Progress Update (2012-03-07)

Alrighty – some (quite a few) images of the high-poly – the poly count is through the roof due to using extra edges to make the mesh smooth correctly – I’m surely going to avoid this type of pipeline in future projects, perhaps in favor of learning to do the hard-surface directly in zbrush.

Anyway, I’ve opted at using XNormal for baking as it can load each piece individually if I export every subtool one by one – it doesn’t seem to have any issues with extremely high-poly meshes either, I loaded a 13mil (one-piece) mesh up and it worked like a charm both for normals and ambient occlusion.

Aside from the character itself, I’ve also finished the shield and the sword high-poly meshes.

The character took a bit longer than I expected as I went further than originally intended with the detailing (and so I’ll sneakily withdraw my estimate of being done two days ago). Nevertheless the final result was well worth it.

If anyone wants me to I’ll upload images using another shader, but for now I’ve gone with a shader that emphasizes detail properly.

Progress Update (2012-03-02)

Another weekly update. This week I’ve been quite busy figuring out ZBrush to the extent that I’ll be using it, that is to say – mostly for detailing and alot of masking. I’ve also gone through the pipeline of high-poly to low-poly, and gotten somewhat used to working with shaders in unity.

Now, today I’ve run into some problems, I did have slight worries in regards to this — however it came from another source than I originally anticipated.


When attempting to export a mesh from ZBrush (I’ve been using subtools to split the mesh up somewhat) I ran into an unexpected problem! It’s not possible to export all subtools at once without merging them, at least not to my knowledge, having looked all over the place for information.

Now, this wouldn’t be a problem normally (I’d expect) but in my case I’m working with a ridiculously high-poly mesh since I’ve done the hard-surface work in Maya, and in order to get proper details in I have to divide the mesh 3-7 times depending on the mesh, more or less resulting in extreme point counts, upwards 33 million.

My computer at home has no problem handling this amount while sculpting though, as subtools lighten the heavy load. The issue also lies in that ZBrush being a 32-bit application can’t utilize more than 4096mb of RAM and chokes up while trying to merge the separate pieces together.

Anyway, I know a few different methods that could act as a workaround. Starting with the simplest one being to drag the division levels down before exporting the mesh for retopologizing. This part wont really be an issue and it’s not something I’m worried about.


The issue comes at the point where I need to bake the maps. I have layers of objects where I want only the outermost ones showing (example being breastplate occluding the chainmail and straps occluding the breastplate where applicable).

To combat this there’s a couple of approaches that I’ve thought of so far, such as baking the subtools separately and combining the maps afterwards, or separating the parts where you wont see any seams from the main body. There’s also the option of baking directly in ZBrush, but that’s not something I currently entertain – as I have no good idea of how well it does it (not to mention, if it can handle the workload).

I’ve also toyed with the idea of making each object a dynamesh and then reverting back to the original mode so that each face of the mesh is similar in size, which should allow for more details in the spots that require them rather than at the edges where the detail is currently being put due to the way the mesh is built (the maya hard-surface modeling followed by smoothing to allow detail work).


Anyway,for now I’ll update with a few pictures of the current progress, seeing as the work I’ll be doing in ZBrush wont be too extensive – mostly bigger details (as seen on the shoulderpads in the images below) and scratches etc – it shouldn’t take too long now that I’m more or less aware of what I need to do.

I expect to be done with the modeling process by sunday evening. And hopefully if I manage to make the mesh lower-poly than it is I can bake everything besides the moving parts at once as I originally intended to.



Progress Update (2012-02-27)

Almost done with getting the mesh ready for sculpting — images of it will be uploaded soon enough (only the sabatons and gauntlets left to make), got somewhat sidetracked with testing out some zBrush functions for making detailed normal maps along with baking and shading.

Here’s an image of the (earliest) chainmail mesh in zBrush. I’ve used the noisemaker function to tile a texture across the entire model, then displacing the geometry with it.


(Image below)

This is the normal map applied with a simple bumped specular shader in Unity (to the left) and a custom shader (to the right) that uses a baked ambient occlusion multiplied with a cavity map to separate the darkened (deeper) areas aswell as to provide the shader with information as to where highlights on the chainmail links are supposed to be.

The shader also uses a cube map to provide reflections, and it’s masked by the occlusion/cavity map mentioned earlier to make sure that the reflection is limited to the actual links.

This is by no means a final result, but I felt I had to go through the process of going from high-poly to shaded lowpoly.

I haven’t played around with the occlusion baking options enough, as shown by the lighter areas at the lower part of the right model. The final version should be more evenly distributed to give the entire mesh the feel the torso has right now.

This ‘low-poly’ is 2200~ tri’s, but keep in mind that only the parts visible through the plated armor will be part of the finished mesh.

I should also mention that the rings may be made bigger than what they’d be on a realistic chainmail due to aliasing issues, but I’ll hold off my judgement on that until I see the chainmail below the rest of the armor.

Progress Update (2012-02-24)

Early morning update with some new images of the modeling progress – it’s coming along nicely. I’ll hopefully be ready for sculpting by tomorrow at the current pace.

Just ignore the random objects in the background, I tend to keep copies of things thrown around the scene if the need for something specific arises after I’ve altered the original.

In interest of saving time, instead of modeling a base mesh for the face I grabbed one of the heads available at the Polycount wiki, called ‘talon bust’ so that I have a basis to sculpt a face from.


Progress Update (2012-02-22)

Updating with some modeling progress (The basis for the high poly). – I’ve included descriptions of what the different pieces are going to be.

Note: The shoulder plate shown will be situated below the bigger shoulder guard in the concept – this smaller shoulder guard is the ‘top’-piece of the layered plates along the upper arm. Also, once I add the belt along with the steel plates going down the legs the character shouldn’t feel as thin as it does now.

I’ve also added an updated progress plan (that includes the change of concept art) aswell as the concept art to the MS2 folder, located under the Delivery category.


Progress Update (2012-02-21)

Following MS2 I thought I’d mention that there’s now a new testbuild to be found in the MS2 folder under Delivery that includes the rigged base-base mesh with the basic movement animations in place (playable).

During the presentation I mentioned that I’ll be reworking my concept. This is mainly because I honestly don’t think the old concept would pose much of a challenge in terms of difficulty (and I also wanted to avoid making a mantle) — and since I’m now about to enter the modeling ‘phase’ of the project I thought I’d make sure to do so quickly.

So, I’ve spent around 8-9 hours straight in photoshop, concepting on a new, more interesting design along with my friend Milad Ghanbari — using a posed version of my base mesh and doing a paintover. I’m really happy with the final result (shown below). It’ll be a ton of fun to model for sure!


Note: I do realize that I’ll most likely need to alter my animations due to the new design, but that was something I would’ve done in the end regardless – so no loss there.

I’ll update MS2 with the new concept aswell as noting down the change of concept in my project plan (which will also be located in MS2) tomorrow.

Progress Update (2012-02-17)

Going to be more of a daily update today, since my wednesday update more or less covered what I’ve been up to this week.

Anyway, I’ve reworked my run-cycle a bit following the feedback I’ve recieved – giving it some sideways motion to make it more interesting from behind, aswell as making the leg movements less extreme – and I have to say, it looks far better now.

I’ve also worked a bit on the turn-animation, to make sure the feet weren’t sliding around, and I found a pretty simple way of doing it – using (temporary) joints as snapping points, making sure that the heel pivot – which is where my characters feet rotate around – doesn’t leave the spot. I had to do it this way since my character rotates around it’s origin when turning in-game.

There will be no video demonstration today, but I’ll be sure to post something following MS2 on Monday.

Progress Update (2012-02-15)

Just updating with a small demo of the current animations in action.

The video demonstrates the Idle, RunningForwards, TurningLeft, TurningRight and WalkingBackwards as it is. I’m not quite satisfied with the WalkingBackwards animation yet – it’s a bit stiff, so I’ll probably revisit that soon.

For now I’ll be focusing my attention on the strafe animations – possibly followed by the jump-related ones.