Composition tests

Early tests compositing CHELL inside an environment, with DOF, MB and 32 bit glows, reading from a multichannel EXR.

Some thoughts:

• Mocap files don't hold up well in slow-mo
• She's too busty, I need to correct that.
• Light is a little bit too harsh.
• Game textures are not high-rez, so, at 720p they look poor.
• I need to use Indirect Lighting for it too look good. No big deal, but that means 300 frames require 14h of rendering. Oh well.

Skinning - Weight painting

To be honest, I never managed to make the "Paint weights" feature of the Skin modifier work properly. Until now.

I didn't grasp the concept behind it because the Painter brush thingy in 3DSmax always defaults at the wrong preset. In order for this feature to work, you need to set the pressure very low, like 0.01. If you fail to do so, and leave it at default 1.0, it gets out of wack extremely quickly, making any adjustment literally impossible. Radius, on the contrary, seems to have little effect on the general behaviour, it just covers more of the polygonal area. 

Thinking about it, it must be because I use 3DSmax with a Wacom, which, in my experience, is not common between 3D artists. Pen pressure combined with default settings in painter give us unpredictable results...just a word of advise! 

Working on some animations

I've been busy working on the environment (which would be a "level" in a game) but it's getting frustrating. I've discovered I have no talent whatsoever when it comes to designing maps. So, I'm dragging myself through this stage. To cheer me up, I'm refining the animatic, picking differents parts of the music and testing some new animations. Here we can see the underwhelming presence of 3dsmax biped.

However, without his great -underlying- help, no animation would be possible. I believe real animators tend to use CAT inside max or even Motionbuilder, but from a filmmaker perspective, nothing beats the simplicity and ease of use of the classic biped. I've been working with it for more than 10 years. You learn to love it and hate it at times, but most of the time, to deal with its capricious behaviour.

Storyboard/Animatic V01 Completed

Time to get away from 3D for a while and focus on the story. Now I have a working animatic, mostly hand-drawn using my astonishing artist skills in photoshop (*sarcasm*).

My impressively cinematic storyboard shots, in action.

Some issues, of course. Being an overkill shooter as I am, there are LOADS of shots to complete. Well, I imagined it, so I won't start complaining. It's time to start crunching shots. I've identified some problematic shots regarding simulations, but once I get my mocap set working the rest of animations should be pretty straightforward (providing that the general workflow is producing good results as expected).

Runtime: 2 minutes at 24 fps and 720p. Rendering times might be an issue in the long run but I'm not too concerned. I'm not planning on using GI or some fancy things like DOF or Motion blur, all that will be done in post. My general approach is using HDR dome lights and Vraylights with no indirect lighting. That tends to create nice soft realistic shadows without compromising render times.

Beware of the M-turret!

Working on some of the graphics and still deciding about the general look&feel of the piece.

I rigged my conceptual turret and simulated a slalom drive, just to see if I like the design in a practical use. I believe it behaves nicely, although center of gravity is a bit too high to be safe. But then again, the original design was intentionally clunky, so that u could grab them and toss'em easily. I don't really know if placing 10 or 15 of these puppies in a scene will bog my system down...I'll try. Anyway, I could always animate them separately and bake the animation information into each model, or even use some kind of animated vrayproxy or point cache. I believe it's a matter of complexity in terms of animation information more than mesh density.

PORTAL MOBILE TURRET (M-TURRET) CONCEPT

So I tried to give our beloved turrets some mobility. This is the concept I've created. Initially I wanted to give them tracks like a tank (minitank :D ) but I do believe it would be a nightmare to animate. Instead. 4 wheels like a tractor would allow the turrets to go through rough surfaces and still look a bit like a toy, which I believe was the intention of the original design. A deadly toy, though. I have tried with red tires and other combinations but this one in particular looks convincing to me.

CHELL - Expressions

Some expressions for Chell, modeled in Zbrush and transferred via OBJ to Max.

PORTAL RIGGING/SKINNING TEST

Skinning early tests. A generic mocap file of a female in a gym. Some problems here and there (like the hilarious happy shiny armpit) and obviously the animation is not tweaked to fit the exact proportions of this character. That's why both hands go through the floor and there are some hand-body contact issues. But that was not the goal of this test, but testing if my approach to rig her with the biped in 3DSmax works as expected. This is a chance to see the shaders working too.

There's a particular thing about Chell. Even if it's not apparent the first time you look at her, she's always on high heels. This is causing all mocap files to work in strange ways. The solution is tweaking these files globally, creating a keyframe to "elevate" the heels in order to get a nicer and more natural appearance.

I have some doubts regarding mocap sessions. Should the actress wear high heels or just regular shoes and correction keyframes should be applied? If the actress is wearing regular shoes, she will be able to pull better (in the athletic sense) moves, or , at least, I believe so, but on the other hand, heels would be a constraint a real Chell should have to live with, thus rendering -much more- faithful moves. I will probably try both approaches and see how they work. More on that on next posts.

Chell full body test

Some tests, experimenting with diverse levels of dirt and wounds. I think it's coming out nicely, much different from the classical (and cleaner) version of Chell we all know (and love). She's not fat at all, I don't know what Glados was talking about the whole time. The fatness is a lie.

PORTAL SURVEILLANCE CAM

Someone (or something) is watching us! Early tests converting the classical cam which was so popular in Portal 1. I've remodeled some parts, improved some textures, created vray materials and rendered it with a bit of HDR spice. Added some DOF in post. In the final piece, it's animated (as much animated as a camera can be, anyway). I believe this is pretty close to what we'll see in the final video.

CHELL (Portal main character) test renderings. Face and expressions!

This is an early test for CHELL,
the main character (and actually the only one). I custom modelled the eyes, and revamped some of the textures, uprezzed them and added some new features, like dirt and blood. The idea behind this new look is creating a beaten version of Chell, similarly to the new Tomb Raider game (2012). She has been passing tests over over again, and she's exhausted and a bit angry. That's how I imagined you would be if you were forced to go through endless test chambers, the same way the characters from the movie Cube were extremely pissed off.

To accomplish this look I've combined the Vray SSS2 material with custom painted specular maps. I tried to create SSS maps, but results were poor. So I believe I'm sticking to diffuse map, normal map (painted in zbrush, skin pores, lip wrinkles) and specular amount and glossiness maps. HDR maps help a lot too.

I've tried to create some expressions too, but due to the fact that 3Dsmax is pretty crappy when it comes to organic selections (even with soft selection on), I've resorted to trusty Zbrush for this particular task and...it works great. I discovered the "Topological Move" Brush works wonders when creating muscular displacement. I'm still investigating the matter, though.

Portal 3 Fan Art Project

I'm starting a personal non-profit new project inspired by the Portal universe. It's based loosely on the Machinima scene, since I'm using some parts of the original models from the games, but replacing textures, creating new shaders and animations and using a 3D modelling package and rendering engine to deliver the final images. Since it's not using the game engine and it's not real-time, it cannot be called machinima per se, but I'd like to think it recalls some the original spirit. The objective: loads of R&D in order to -hopefully- improve my skills.

Will I be able to create a short ad on my own, although I'm not an animator and not especially proficient at creating full CG pieces?


This project will include the following aspects:

• 3D Modelling, texturing, rendering of some of the Portal game assets, modifying the original meshes extracted directly from the game, to reach an acceptable level for cinematic purposes.
• Shot planning and storyboards
• 32 bit multichannel compositing
• Motion capture using PS3 cameras and some software to be determined
• Manual face animation
• Music to be determined too, probably not sound (music video-style)
• Editing
• Viral success! :)