Trying to get at least the head finished for one of the poster critiques, I started thinking about making the hair. For a ease of mind I did a quick one on Zbrush to avoid looking at a bold head all the time. It was quick and frankly... awful, but at least I had an idea of what I was going for.
I decided to go for a more final fantasy look by making planes with textures and transparency rather than going for geometrical hair. This last one was used in games like kingdom hearts for the main character (Sora ) on game play, but not for the cinematic
Technically speaking I wanted to get an end result like this:
At this point in time the texture still a work in progress as I want to give it requires a lot of attention and I'm currently battling between engines (Unity and UDK).
So far I've created a few simple brushes in photoshop that will help me get hair lock-like strokes
Something that I never really considered much is the fact that my character's got two colours of hair (of course) the top part is blond and the bottom one is dark brown. This reduced my texture space quite significantly but hopefully I'll be able to achieve good quality of with specualr maps.
Hai Phan is an artist I came across when I was looking for tutorials that involved the entire pipeline. It was in CGCircuit that I found his videos and the fact that he wasn't going for a realistic approach but rather a really nice stylised that made it so appealing.
His tutorials were helpful as he didn't approached a software specific but rather explaining the general concept and industry standards.
Hella is the character he mainly worked with and it was amazing to see most of the process on the making.
His portfolio is fascinating and quite extensive (have a look!); the amount of different characters and creatures that he creates varies a lot yet you can still sense a style within.
Not only his videos were really accessible but it was really easy to get in contact with him and he was kind enough to answer some questions that will be useful as primary research for my dissertation.
So at the end I managed to get a basic house together; I realised until I started doing some test renderings that probably I went in too much details with some of the textures while the ones that I left at the end are a bit sketchy. This is a final group of the house without any deformation. Not sure if being a bit less than 14000 Tris is good or bad for a game environment, but probably I could have done with less, specially on the windows (take in count that there are windows on all four sides of the house).
Also had some troubles with the roof: the main roof tilling is at different scale as the one for the sticking-out windows. I guess I should have done some testing before going through all the trouble of specular and normal maps.
There are some of the textures that were made:
Unfortunately there I ended up making more assets that I could even use like the square's fountain, road sign, chimneys, stairs and even grass. Some are textured but I couldn't put them in context for the render so I decided to let them out. Just for the sake of portfolio I'm planning to finish this scene during the summer.
Above are the unused props, the working texture for the tiles of the town square's floor and the fountain textures. This last one was the first texture I made as a test of the quality I wanted. Spent more time on it that I should have. Also I'm aware that I wasn't as efficient as I should have been; it was until the final crit that I was suggested to reduce the number of textures/shaders in the whole house, like having one or two for the whole thing. At that moment it was quite late as I was already half way through most of them.
So the "attractive" part of my responsive project was meant to be texturing; I had intended to have a simple and low poly environment that could potentially be used in a 2D camera as a sidescrolling with having an extra 3D/free camera angle when it came to exploring and item searching.
The hanging sign is the fist object that I completely finished:
This is just a screen capture, but I worked on the painted texture, a specular map to give it a shiny texture where needed and a normal map for the bumps.
The chains where quite tricky to get them look like actual geometry when they're just two plains, but I thing that by exagerating their normal map it worked out quite well.
For the normal maps I used Quixel's software called nDo2. It works as a plug in for photoshop and it's pretty easy to use. The idea is that once you have a texture you can convert it into normal maps by the click of a button, instead of going all the way of having a low and high poly mesh and bake it. Probably it's not as precise and it does have better results when you work with photographic textures rather than hand painted ones, but it sure makes life a lot easier, specially with features like live 3d preview, live sculpting, presets and a good panel control to adjust all levels of the normal map.
I'm aware that it can make specular , height, diffuse and other maps, but at least in specular maps I think I prefer to work on them on my own. I found this video on youtube that can show you an understanding and importance of a good specular map. It does come a lot harder when you are working with hand painted assets as you have to create everything from scratch.
I've also been working on other assets like this fountain below. My main problem at the moment is the time put a dedicated amount of time on each asset as it drains a lot of time by doing so. I thought for my submission I could just focus on having as many assets as possible but with good finishing quality rather than having many unfinished. I'll aim to have a house finish, but I can't certainly assure this at this point.
When I was doing some research about making a cartoon style I came across with this thread on a Polycount forum where a user took a good time to "open" the game The Legend of Zelda: the Wind Waker and made a really good break down and analysis of them.
What caught my attention initially was just to see how professionals do textures but then it expanded to more helpful stuff of how to make scenes more efficient by adding baked shadows to the textures for scenes with no changes in the illumination and also something really iconic in the game where the particles.
There are also some libraries of assets that I found. All of them are cartoon style but it's surprising how iconic and similar the explosions look to this saga.
You can check the rest of the post for more information. It breaks down a lot of secrets that where used to make this game a total jewel.
UV mapping Albus was more than a challenge, it was just time consuming. having all the little pieces separated made it easier, but still it took longer than what I expected.
Most parts went out smoothly, but having to model the environment and UV map it at the same time made me not to be able to texture Albus until really late stages of the project.
Also it was really hart to think of a colour that would suit him, and me being not so good with colour theory I ended up opting for a "typical" bronze scheme.
Here's some of the progress images of Albuss texture:
1.- just the UV map capture.
2.- The textures finally painted...took me ages to do it and I also put a small clover as a signature :)
3.- This is just an example image of the final render of Albus. Looking great and so happy :)
This is just an idea that I thought could work (if we had more time). I wanted to make an occlusion render pass to the textures that would add a nice feeling to the textures, and probably help to give us a cartoon style with a bit of more complexity, but unfortunately this was just an attempt the process was becoming a bit more complicated that what I expected, and it had to be done to every single object in the scene, so probably the rendering time would have double and the time for the deadline was just upon us.
Woking on my plants for the scene, I ended up spending too much time on details for some of them.
At first I tried to do all hand painted, because I wanted to get this dull scene to look all full of colour, but then time was running low and had to start working on other assets.
After putting everything on Unity, I noticed that the normal maps I did for every plant where useless, as the shaders that accept back face culling dont accept any kind of bump mapping.
Also, never thought of combining the meshes in maya to avoid having too many objects in the scene.
Once I put all the plants in place the scene began to get a little bit laggy. The only thing I could have done at this point was going back to maya and combine them into one mesh, but because of the time limit (and illness) I'll leave it like this for now, as the important part at this point is more composition rather than efficiency.
After loads of attempts of trying to get Mudbox running on my laptop and nothing good, I decided to give it a try to Zbrush. My plan was to make the stalactites and stalagmites in high poly and then capture the details on a normal map. The product was quite good looking, but I think I spent so much time trying to get this level just for one piece and thinking of the amount of pieces that I needed to make it look decent.
At the end, and after spending so much time looking at tutorial after tutorial, I decided to just edit one big mesh and make it look as good as possible with a tileable texture. much more easier as Zbrush does the complicated UV mapping for you and it's really easy to export back to Maya once you set it up. After running a decimation to avoid killing Maya on the attempt to export an object with over 1 billion polygons, I got quite surprise that it changed from quads to triangles... never worked with them and my previous 3D professors where completely against working with them.
Didn't care...
So manage to get the normals working and to get it all fit and scale with the rest of the scene. So far is the object that's spending most of my poly-budget, but hopefully will be worth it.
I still want to play more with Zbrush, since it's a really different software, that quite took me a while to get my head around, but seems like one of those in which I can see myself getting lost for hours just trying and trying all the tools.
After duplicating the half of the cenote (cave) on which I worked on, and merging them together I saved and closed Maya, but is until now that I noticed that by doing that I completely destroyed the UV map and it was turned into a square, so now I either work with an automatic UV proyection or start working on a uv mapping of an object that has over 67,000 polys and loads and loads of organic bumps. God save my soul if I chose the second option. (which I probably will)
Jose Alves da Silva is a Portuguese 3d artist that has had loads of appearences in magazines like 3dtotal and 3dcreative.
Appart form loving his over stylised way of working, I found very interesting to find a some images of the same final product, but in the different phases of it's making. Like the one below, you can see the turn around of the character on actual clay, a final 3d model and the final product with textures and post production. Also de fact that he started his current 3d passion by doing a degree of architecture kind of game me a little bit of hope of being someone in the industry.
Check out his website and portafolio:
After finishing the main parts of the scene for the animation I wanted to start getting some textures just to start testing the looks of it.
This is just a scene of the scene before importing the dolls into it, but after a bit I added more stuff and change the whole arrangement of it:
I know the scene is simple, specially because my original plan was to recreate my own desk, but then doing too many details might cause the scene to be too distracting of the actual subject and also really time consuming (which is something we all lack ). This picture below is more or less the final arrangement of the scene. Even scanned the posters and post-its from my walls.
This picture below is trying some other textures, like some for the dolls that are done and also trying the bump on the book and desk. I hope that putting a scan of an actual manga I have in my room doesn't get me into trouble lol.
Even now I'm struggling with defining the style for my animation. I think is one of my biggest mistakes from the begging because I didn't established a style to follow and now I hope it doesn't turn out too mixed since some of the textures are too cartoonish and the rest aiming more to realistic.
I'll try my best to blend this with the illumination.
At the beginning I thought that my scene and the whole animation was going to be really simple because middling the Russian dolls wasn't much of a hassle. My initial idea was to animate their faces to enhance the expression of the movements and give them a nice personality but in order to do that I needed a proper UV mapping for the models. Now I've done UV mapping before (on 3d Max) but never came across with something as problematic as this.
It might be my lack of experience with this software or that I just cant retain any information but it took me three sessions to finally get the hang of it (that plus my file getting damage and causing me to do all over again).
Because of their round parts at the top both the inside and the outside, it require to cut the edges on the UV editor, then cut the edges on the back spine of the main model just to manage to properly unfold the figure into something that I could easily work on photoshop and after effects to get image sequence for the faces.
After finishing this I completely feel like the master of UVs... I just hope that I don't screw up with animating them since they all have different emotions during the whole animation.
I really hope the rest of the process works with easement since I'm tuning out of time for the final critic.
