Wednesday, 28 November 2012

Avengers Assemble - Vehicle

So yeah this vehicle project is just about to wrap up, the deadline is Friday but just like any other project this year we have got permission to tweak them at a later time for our portfolios. I chose to model the Quinjet from the recent Avengers film, just cause it looks awesome. I should be told off for running with terrible reference, and trying to guess some of the geometry. The limited concepts and the final product were kind of different, so I tried to find a delicate balance between the two. The concepts were better for reference than me pausing the Movie for screenshots, but the thing is I wanted it to look more like the movie one rather than the concept. Then I ended up putting my own twist on it, but keeping it fairly recognizable.
      Anyway, the thing with this project is there seems to be something slightly off putting about the way were supposed to present the work. I have no issue with the subject matter, I like vehicles and environments but its this whole screen grab from max, viewport shaders and touch up with Photoshop that I just don't quite grasp. If I'm going to be honest, we should really be demonstrating how things look in a proper game engine; seems we have access to both CryEngine and Unreal I just don't see why we aren't using them. I complain because the vehicle looks kinda bad in the MAX view port, and too make things worse it just doesn't look that great using the Xoliul shader either. When I render it looks fine, but that too isn't a good representation of in-game graphics.
Max screengrab - Running off the Nitrous driver (Realistic with Materials)
Render :(
I've configured the driver to set view port textures to the best resolution it can provide, and thats basically the result I got ^. I have an inkling that the problem is to do with the diffuse being pretty much grayscale, and some kind of graphic compression is causing this.
I may have left it a little to late to start fixing these problems, especially the Xoliul shader because that's what the brief requires, but as an alternate I just built up the vehicle again inside Unreal and took some screengrabs. This is the result.

So of course there is a difference, but still unsure if this is permitted for the marking. I'm finding it quite frustrating that everyone else seems to get there's looking rather sweet in MAX using Xoliul, and as the brief states to use it for real time screen grabs I feel like I'm not going to do as well as I should. On the other hand I feel more comfortable presenting my work by using an in-engine screen grab because that's essentially what all game models run in, not 3ds Max's viewport.
Anyway heres's more presentational stuff:

Max renders!


I'm gonna come back to this model, just so I can fix some of the problems aswell as continue with it. I still need reflective maps, to give the metal look it doesn't quite have yet. The model is around 6,300 tris I believe, which gives me roughly 3700 tris to mess around with. An interior is priority, landing gear would be cool but the wheels will eat up more triangles than I can bargain for.


Okay so I sorta managed to set up a shader in Max, not the xoliul but the Car Paint shader Codemasters provided for us. I broke down the original example file, which I really should have done in the first place and saw that the body of the phaser tank was using this shader instead of xoliul, so I tried it out on the Quinjet. It works a lot better. The diffuse is still a little flat, it needs more texture to it so I guess that's my next task.

looking better
return to the sky - realtime this time
Photoshopped version. Still messing around with editing screengrabs.
The reason this one is sharper is because I managed to up increase the anti-aliasing
in Max's driver configuration to max (8x I think) its essentially the same as the one above it but with smoother lines.

Monday, 12 November 2012

FMP Pitch

This year is a mixture of 3D projects, we've done a rooftop (environment), currently working on a vehicle so I guess the only thing left to model is characters! I may not be the strongest at this particular segment of 3D but I feel that 4 months of blitzing out some character models in dioramas would be a pretty good way of training myself, and hopefully I can push my skills further especially considering all this work is going to go into a portfolio. For those who don't know I'm still undecided of what kind of artist I want to be. Ideally it would be great to be some sort of Lead, or Director - and so its best to show a strong talent across the board, being able to concept and model anything. That way I'm not too fussed of what role I'm given if I get into the industry; and I would be completely versatile If they needed me on another job.

Here are my ideas so far, they're a little undeveloped at the moment but in time Ill have the idea fleshed out. These characters are from a hypothetical video game, that maybe someday I can progress into an Intellectual Property, but for now its kind of fun to design these anthropomorphic characters, and I hope that I can replicate the art style in 3D. In conjunction with this project, I may choose one of these characters and create a medium sized level. These ideas are; an icy mountain village and a haunted house. Two contrasting environments each with their own pros. In the next few weeks or so as I begin to flesh out the ideas more, there may be some change, for example I may come up with other environments and such. I'm happy with the ideas so far, I know I just need to push them further.

Our Protaganist Penguin
Our main Antagonist - Walrus - still under development

General Ingitov

Scrapped Concept for Medieval theme

Friday, 2 November 2012

Rooftop - Final

After four weeks of solid work, this is what I managed to produce for the rooftop project. As a quick introduction the idea - the design of my rooftop was a 'Prince of Persia' style palace, trying to emulate how Ubisoft successfully combine fantasy and reality Arabian architecture. Some concepts:

Sultans Office
Walls and Pillars

                                                                            Rooftop Concept                                                                                       
 These initial concepts were a basic illustration for the size and architectural type of the level I was going for. The level was going to be crafted using tile sets, reason being was that last year during our group project I found that building architecture in UDK was so much simpler with static meshes in comparison to using clunky BSP. The tilesets were kept fairly simple, for example floors and walls were generally 128x128 (3dsmax units) planes. The only thing I needed to ensure was to line up more complex architecture like the arches to the grid, so that by instancing them they would snap together seamlessly (modular).
    Once I'd matched the Rooftop Concept in engine, I felt that the level needed 'more'. Using tile sets allowed me to continually push the level expanding in all directions. I always feel there is room for improvement in all my projects so I am never settled, I mean I will be more than happy to accept at this stage the level was decent but I find that I have fun trying to find new and clever ways to improve upon what I have achieved. I guess that's because I was so enthusiastic about this project. I was actually disappointed this project ended but fortunately we have the remainder of our final year to continue working on it. Anyway here is some real-time screenshots of the rooftop.

View from the balcony.

The sultans office was initially supposed to be crafted from more elegant materials like Marble with reflective floor tiles to make it look more impressive. I decided to craft the office using wood rather than marble, which gave it more of a 'office study' feel.

Sun Lounge - with Camels!

View of the palace/mountain backdrop
I guess the only thing left to do now is to render out a fly through, there were a few animated dynamic elements in the level that would be cool to show.