Shading! (Overview)

Shading department is the department I worked in and managed most of my time in Snowball. I did most of the shading work on R6, and had great help from Chen Hillel towards the end of the project.

In this stage we get the finished, rigged model, and do everything we need to prepare it for rendering - unwrapping > texturing (and sculpting) > shading (setting up the materials) > defining render settings. Sets get lit in this stage, too.

I plan on making some in depth posts about some of the shading in R6, but for now here's a basic walk through.

As I mentioned in an earlier post, we did all the rendering and shading in 3Ds Max 2010 and Mental Ray.

We had two other major applications we used, both of which were in beta stages when we started out, and both of which it's been a real pleasure to see develop across the course of the project and still.

Unwrapping:

The first one of these applications is Unwrella, which is an automatic unwrapping tool. This saved me A LOT of time, and we use it extensively for other projects as well. I used it to unwrap almost everything except for a few specific models I had trouble with at the time (the current version uses UV seams, but the beta version didn't).

If I wanted different parts of a model to be given more/less texture space, I'd capture a morph target of the model in its normal state, deform it so that its parts' sizes in relation to each other matched the relation I wanted to have UV space, unwrap it with Unwrella, and collapse the stack and apply the morph target to bring the XYZ space object back to its original form.

Texturing and Sculpting:

The second program is Mudbox. I decided early on that I want to avoid using ZBrush completely. I've used it a lot in the past and wanted to see how well we could manage without it, for various reasons. We managed well. We used Mudbox to create probably over 80% or 90% of the textures (with help from Photoshop), not to mention sculpting. The first version gave me a lot of grief, right after Autodesk bought it. It crashed occasionally, and didn't play nicely with meshes or UVs that weren't made to it's exact specifications (and the UVs were difficult to control because Unwrella worked automatically). But by the time I learned what I needed to do for Mudbox to like my meshes and UVs, Mudbox improved it's flexibility considerably, and it's stability. Today we use it for everything, alongside Photoshop.

Shading, Lighting and Rendering:

I did a lot of research before I decided to stay with MR for R6. I tried out Brazil, Final Render, VRay, and MR. Each had it's advantages and disadvantages:

Mental Ray is our default engine. MR has great materials, the Arch&Design material can do nearly anything, and the SSS+ material does most of the things it can't. They also just released then the new final gather system for flicker free GI (projecting along camera path, and capturing points' values per frame). However, the Anti-aliasing calculations were slow and gave bad results, the area lights were a pain to use, and render time in general was high.

Brazil had terrific anti-aliasing and filtering, as well as soft shadows. It renders quickly and gives wonderfully smooth and sharp results. However, it's materials arsenal is poor, and more importantly, it's pixel displacement wasn't good enough for what we needed.

Final Render had the fastest global illumination system. It was quick, accurate, easy to use, and flicker free. It even used MR Arch&Design materials. However, it too had less than satisfactory pixel displacement, the anti-aliasing was moderate, and it didn't have a proper SSS material. It also gave a slightly artificial look in it's renders I couldn't quite put my finger on.

VRay had the best lights I've tested. It was relatively fast, had very good anti-aliasing, and great pixel displacement (In modifier form! Like it's meant to be). However, we had a lot of issues with their floating licenses, and there was a poor material and memory management tools arsenal.

Eventually, with the help of the MR exposure control, the new photometric lights, and some other features, we decided to stay with MR.

Like I said, most of the materials were A&D materials. The rest were mostly SSS+ with disp. and SSS++, or scanline Matte/Shadow material for certain render passes later on.

The mental ray SSS materials are intrinsically very good. They behave well and are fast to render, and the SSS+ material, by Master Zap, introduces the A&D material's reflection channel. However, the biggest issue they presented is that they absent in render elements altogether - they simply output black. When I posted this in a relevant forum, Master Zap wrote the SSS++ material which allows for the material to show up in several render elements, and can have other elements such as reflection and specular made for it as well. But it to this day still doesn't have a displacement slot, nor does it apply exposure control to it's specific elements. Ah well...

Download the SSS+ and SSS++ materials here.

The lighting was done using the MR Sunlight and a regular Skylight for the Hallway set and the Forest, and a host of photometric lights for the Flowers set. For extra lights I in the sun lit sets I used photometric lights as well, or spotlights with MR area shadows.

Rendering was done only later on during the production, but all of the render settings, rendering methods and render passes were defined here. We used MR final gathering for GI. The new system allows for flicker free GI with camera movement and object movement.

I defined the render passes with the render pass manager we've been using since the advent of the studio, Grant's RPmanager. 

All the beauty passes were rendered with MR exposure control, and to 32bit float linear space images. That means we had to work with gamma correction for the textures and the final output.

This is a turntable of the Preacher after he was shaded. All the sculpting and texturing was done with Mudbox, Unwrella, and Photoshop. His skin is an SSS+ with disp. material. Rendered in MR with a Skylight and a couple of area lights.

Our major assets

I think I'd better straighten this out. Here are the major assets we had in R6 and their names, so that I can use the names from now without a problem.

Our main character, Snap:
Snap is number 17 in CLR clan (team). He has clan uniform/armor, and carries a standard issue clan weapon called Ashmedator (seen in the first image).

       

the Preacher:
The preacher's a large and strong animal. He guards the goal to which Snap's team is supposed to get the pigskin. His metal head has a shielding mask he can deploy to protects his eyes and ram people he doesn't like with. In this image the mask isn't deployed.

the Pigskin:
This is the game's ball.

the Hallway:
This is the spawn zone. When a player is fragged, respawns in the next round here. This is an early render.

the Flowers set:

The sets have lousy names. I'm aware of that. This set is part of an underground alternate route in the area which leads from the spawn zone to the forest, where the goal is.

the Forest:

This is the area surrounding the goal.

Modeling! (Overview)

This post will cover a few basics about our modeling process.
Most of our modeling was done by Igor Kreinin, with a little help from me. None of the art had orthographies, so much of the art interpretation was left to Igor. In certain places I did detail sketches by demand to fill in things that weren't clear enough from the art.
Modeling was done in Maya and Max.

We decided to have the models in need of sculpting, for which I used Mudbox, divided into two: the models that required sculpting for animation (such as the ground in the forest set and the Preacher's [the monster] anatomy), and models that didn't (such as cloth wrinkles in Snap's [the main character] shirt and dents in his armor).
We wanted to start with the layout and animation as soon as we could.
I started working on sculpting the models Igor finished while he worked on the rest of the modeling. The sculpting that wasn't relevant for animation was left for the shading stage.
Everything intended for heavy sculpting was kept strictly low-poly, and as quadrified as possible so that I'd have a cleaner, more evenly detailed mesh to work on.
The Preacher, for example, had his head properly modeled, and was given his basic bulk in low-res only. I took that into Mudbox, worked on it, took it back to Max for retopologizing - since Mudbox doesn't have retopologizing tools yet, and then back to Mudbox for projection and Displacement map creation.



Here are some of the clean models:

From Modeling assets went on to Rigging, and then to Animation and Shading.

Art! (Overview)

2D Art and design was done at two phases of the production.

The first part was visualizing general ideas as we figured out our story. Based on the principles of the game (Fragball) R6 is based on and the world and feel I wanted to have in R6, I started threshing things out, comparing ideas with the synopsis, and trying to zero in on the design language and style I wanted. I decided eventually to go for a modern sports meets sci-fi soldier designs for the characters, and have something more surreal and aesthetic as far as the sets went (Steering clear of the usual industrial gray and brown colored gritty look). These are some of the early sketches:
  

After playing around within these guidelines (doing a lot of sketches, looking up references on the net, watching game trailers and cinematics, etc.), I ended up with the football derived design for the team players, the bulky and gritty design for the Preacher, and a host of sketches and ideas for the sets. Here are some of these:

We have a lot of left overs from this stage we still use to further develop the Fragball world.


The second part started after we had the synopsis and most of the storyboard done, and I needed to start fine tuning the designs to providing more accurate art for the 3D production to refer to.
Here I a) Figured out lighting and color schemes for the sets:

b) Collected photo references for materials, sports gear, anatomy etc.
c) Made detailed art for various parts of assets:

d) Generally made sure the next stages in the 3D production had all the resources they needed to be in sync with each other and with the artistic direction, which meant mainly collecting a lot of references from the net
(Some of them were also used for compositing and shot lighting reference later in the production).
We ended up starting the 3D production with a few dozen paintings and sketches and a few hundred reference images from the net.

In future posts I will show the artistic progress which lead to the final design for the team players.
There was also the art and graphic design for the whole frag-list (scoreboard) and timer shots, as well as the titles, but I'll keep that for another post, too.

Workflow and production

This post will cover some of the basic information about the creation process of R6. In later posts I will get into more detail about each stage.

I've segregated the production of R6 into 3 stages: Pre production, 3D production, and Post production.

Pre production:
This stage includes everything up until the actual 3D work starts.
The major landmarks here are the synopsis and scripts, the art, and the storyboarding.
Basically, this was the process: we came up with the initiative to make an internal sci-fi short, akin to a video game trailer. The goals were doing something fun, learning from it, and showing off what we can do (to ourselves, as well). After a few ideas came up and went down, we finally settled on the basics for R6 which answered all of our goals (a fictional video game trailer, approached from a short-film point of view more than that of an actual hype-trailer).
All this time Ilan Baffet and myself were sketching and conceptualizing designs for the project. After we had a synopsis and a rough storyboard, I did the art the 3D artists would later use for reference.
The correct course of action would be to have a tight storyboard before the 3D production starts, but as things would have it, we ended up doing several, rather substantial changes to it after some of the assets have been created in 3D - because we had a few months hiatus from the project and decided to improve what we could in the story and the direction while we could.

3D production:
This is where things start getting more streamlined. In this stage I include everything done in 3D - from the modeling stage up to the rendering stage.
The 3D production can be readily divided into two parts: asset creation and shot treatment.
Asset creation is just that. Creating all the assets - the characters, the sets, the props etc. The three departments who are most active here are modeling, rigging and shading. This stage had to be done after we had a storyboard, so we could put more time into the things that will have more screen-time, more closeups, etc., and not waste time on things that were eventually not relevant to the project.
The shot treatment part includes layout, animation, and rendering. All the 3D work that is done per shot as opposed to per asset. Once a basic version of the assets was modeled, we started setting up cameras and compositions per shot and made the layout video board. Then the shots were distributed to the animators. The video board was automatically updated with each progress the animators made. Once the animation was done, each shot was prepared for rendering and then rendered. Render preparations included cloth simulations and effects.

Post production:
After all the render passes of a shot have been rendered and passed QA, they were all integrated and composited. I kept using the in and out of each shot (it's neighboring shots) as color and light reference. When a shot was finished, it was rendered to a 32bit image sequence and to a low quality AVI to update the video board with (the video board was constantly updated with the most recent stage of each shot).
Once all the shots were composited, I put the final sequences in a single composition and did a final color correction pass on them, added film grain, the static noises, and closed other lose ends.

Here's some more general info:


Apps:
Most of our 3D work was done with 3Ds Max 2008 and 2010. We used CAT for character rigging. Models were unwrapped and textured almost exclusively in Unwrella 2.0 and Mudbox 2008/9 respectively. Everything except for some specific render passes was lit, shaded and rendered with Mental Ray. Compositing was done with Eyeon Fusion 5.3.
Personnel:
We had about 5 major players in the production of R6, and another 10 or so guys who worked on it as well. Many of these guys worked out of their own personal hours only. For a full credit list check out the official homepage.
Schedule:
R6 had many personal after-work hours and weekends put into it, as well as internally funded work shifts. The production took about 12 months, out of which half the time was not funded and we worked in our spare time only. With client projects getting priority, these 12 months were spread across slightly more than 18 months altogether.
The Game:
R6 is a trailer for a video game concept we developed, initially for R6, but which eventually got a life of its own. The game, Fragball, is a WIP and is currently still confidential.

Alright, enough for now. Next time I'll get into some details about the art process.

Welcome!

Three days ago we finished Round 6. We've been working on it, on again-off again, substantially out of our spare time, for the past 18 months or so. I will wait for the website to be finished and this blog to have a few substantial entries before it's all published.

Round 6 is an internal project done at Snowball studios which emerged from the initiative of some of the workers. We had a few goals in mind (not by priority): Learning (both personally - and as a streamlined studio), working on our own project and doing the things we love doing with no artistic restraints, and showing some different work than what we usually produce at the studio in order to get our foot in the game trailers'/cinematics' door. (You can find more information, and watch Round 6, at the official website.)
The learning part is very important to us and to me personally. At initial stages I decided I'd like to share as much of what I've learned as I can with whoever might find it useful. So instead of having a "making of" clip, showing a few render passes, grey animations and wireframes - which really do little in telling how things were actually done - I've opened this blog and hope to fill it with actually useful information.

So here's the plan, and it is completely subject to changes:
I will start by writing, with the help of my coworkers, a basic workflow post, followed by general posts explaining the different stages in the production in further detail. Then I will start posting more specific post, almost tutorial like.
I urge whoever is interested to comment and ask for specifics, so I will know what areas to cover.