Haven't posted some of this stuff in a while, figured I'd throw a few screenshots up since I'm pretty happy with how this one is turning out.  This is part of my current job hunt - in my field, art directors and studios don't care about resumes, they care about what the stuff you make looks like.

If any of you played the original Dungeon Siege by Gas Powered Games back in 2000-ish, you might recognize this.  It was an awesome but pretty simple hack 'n slash with great visual design and some pretty impressive art for the era.  I always liked the visuals and the atmosphere, even if it hasn't aged so well technically.  Here's the part from the original game I'm redoing to modern spec.  It's the point at the very beginning where you discover the bridge to Stonebridge is... well, kind of busted, after some of the dumb orc-things set it on fire and then tried to drive a wagon across it.  Things just snowball from there.




I painted up a very quick concept, and originally wasn't even planning on including the bridge (actually, the original-original concept wasn't even based on the game, it was just a tree on an outcropping).  This took about 30 minutes in photoshop, just to get an idea of color palette, composition, light direction, etc.




Early version:




Added the bridge:



The texture maps for the pine boughs were 100% handpainted - along with the rest of the textures.  Basically, instead of edited photographs (very commonly used method), this entire scene is 2D paintings wrapped around 3D geometry.  For example - the grass:



Here's the process I used for the tree boughs -

I handpainted this:



And arranged those branches onto 2D stencils (everything that isn't the branch gets made transparent).  Then I make lots of copies of them and arrange them into larger, denser branches, and get this:



Then I take those and makes stencils out of them, and they become the stencil textures on the 3D geometry sheets that I build the actual tree branches with.


I didn't like how the tree boughs or the bushes turned out initially.  Except for the tree trunks, both those got redone later.  I kept the tree bough textures but went back and rearranged the geometry multiple times until I got the profiles and volumes I wanted.  They're basically doug fir boughs on a redwood trunk, so they arc upwards much more than you'd think at first.  Took me a while staring at the trees on our property until I nailed down precisely what my 3D models weren't getting right.



So after some rework on the trees, bushes, and the ground, I was MUCH happier with how the trees looked.  The new bushes also were much closer to the undergrowth-type look of the shrubs used in the original game along the edges of the terrain.  Still some work left but it's getting close to what I'd consider final.







A huge chunk of my work professionally has been machines, vehicles, buildings, and inorganic stuff.  It's been a lot of fun doing organic assets.  I'll post updates as I go.

This looks great.

Also I think the idea of showing what you're capable of doing without showing any original ideas for characters or vehicles is really smart as it keeps your best stuff from being "borrowed."

Wow.  Serious talent!

Thanks for sharing. I love this stuff.

Great work. I remember an article from Game Informer recently that talked about what you are experiencing - its all about your portfolio. Kids in highschool are making games, apps, and popular PC mods before they even start taking college classes or looking for jobs. Truly an interesting field.



I like to mess with 3d modeling and especially creating textures in my free time.

Thanks!

I'm less concerned with concepts being grabbed... there are lots of concept artists and tradigital painters that can blow my socks off without breaking a sweat.  I'd be thrilled if somebody thought some of my original ideas were worth using as base concepts.  



It's a neat field because age isn't a barrier to learning.  I started doing 3D by modding the crap out of Total Annihilation back in 1997, when I was in middle school.  I was working on a major commercial release (Pacific Fighters) before I finished art school.  There are some REALLY talented folks who have gotten work right out of high school and skipped college entirely, or gone back later for a classical art degree instead of an art technology oriented degree.  Some of their stuff just blows you away.

If you find yourself in the Dallas area, I've got a job for you.

PM me?

Added some undergrowth, 3D grass, and flowers to match the original game better.  Other lighting tweaks and some texture changes to the wood planks.

Looks good. I'd hate for this thread to disappear.

As a gamer, I remember playing this game and loved it.  Nice to see you recreating pieces with a modern touch.

So.  I'm more or less signing off the Dungeon Siege homage piece as done-enough-for-game-work and starting something new.

Every month, Polycount - my other hivemind home away from this hivemind - does a Monthly Challenge.  Pretty simple: you get a single concept image and a month to finish as much of it as you can.

Here's this month's:



So I broke it down into parts like this -






And after a day's worth of work (with time out spent on hold with two state government agencies... fun...) here is the rough blockout.  This was to map out basic volumes, spacing, proportion, silhouette, and give me an idea for the composition of the scene.



If you look very, very closely at the blockout, there's a little blue smear at the bottom.  That's a scale reference for a 6 foot human.

You'll also notice that there are some parts that aren't an exact match, like the framing for the bubble and some of the angles of the temple.  There's an explanation for this: fuck concept artists that can't solve and maintain simple perspective.  

"How does he get all that done so fast?" you don't actually ask but we'll pretend you asked anyway.



I don't.  It's extremely rough geometry, and it's done with modular pieces.  The entire ridge is made from copy-and-paste clones of three ridge pieces, three rock pieces, and one building copied and pasted around.  Only the main temple in the center is detailed past that, and the bubble frame and stairs are unique.

More tomorrow as I work on this.  Probably a few updates per week for the rest of the month.  Tomorrow is zbrush on the terrain module parts.

Ok, so here's a key part of how modern 3D art for video games gets made nowadays.  What you're going to see is a program called zBrush that is used for sculpting super-high detail geometry, which is then baked down into what's called a normal map.  This is tech that started to see use right around 2003-2004-ish, although the tools and techniques to really crank it to 11 didn't mature for a few years after that.

Long story short, we're going to make a really high detail 3D sculpt that will never actually go in a game, and instead we're going to use it to tell bald-faced lies to the mathematics that drive the lighting and rendering system.  

Here is the super low poly version of one of the cliff chunks from the blockout, imported into zbrush.  We start with less than 50 polygons.



That mesh gets subdivided to about 12.5k polys and I start working out general shapes, contours, crevices, and profiles for the rock formations.



A little while and 788k polys later, rocks happen.  This is not an award-winning sculpt.  I have a screaming migraine.  But it'll do the trick for something that's going to have dozens of other meshes and objects stacked on top of it, or be way in the back of the scene.  This mesh, all 788,000 or so polygons worth, gets saved out as a high-detail version.



Then we create a low-detail version that is about 1100 polygons.  Lots more than the starting mesh, but not even 0.1% of the high detail version.  This is the actual geometry that ends up in the game.



Here is a version with the wireframe overlaid so you can see how simple this version of the mesh really is...



Here's where some deep, black magic happens.  The high and low poly meshes get plugged into a tool called xNormal.  This is what game artists call a "baking" tool; what it does is use projected 3D rays to compare the high and low detail geometry, and then uses a set of imbedded coordinates in the low detail mesh to create raster image (bitmap) representations of different aspects of the high detail geometry.  We use it and similar programs for normal maps, ambient occlusion (keeps crevices, pitting, and deep holes 'unlit'), displacement mapping (where grayscale images are used to actually push the geometry in or out), and a whole bunch of more convoluted stuff.

This point is where everybody who went to art school curses quietly for not taking a lot of linear algebra in college.  Our entire trade is basically wielding weaponized calculus to make art.  When they told us there would be no math?  They lied.

This purple-blue image is a tangent-space normal map - a tangent for a 3D surface is a vector pointing outwards from it.  Bitmap (raster) image files on computers are stored in three channels: red, green and blue.  What this image is actually doing is using color information to encode 3D tangent vector data for the surface as a texture - a 2D image that gets wrapped around the 3D geometry by the graphics engine - instead of as geometry; the red and green channels represent x and y components of 0-180, and blue represents length (the z component).  "Straight out" from the surface is (90deg,90deg,1).  There's some math conversion involved because we're calculating 0-180 degree arcs as 0 to 1 normalized decimal values, and representing them with 8 bits, so the image is 0-255, and represents a 0.0-1.0 range of a 0-180 degree arc.  So the color representation of a tangent vector "straight out from the surface" is the RGB value (128,128,255) - that distinctive blue-violet color.

What those vectors are doing is lying to the lighting calculations and telling it "see that totally flat triangle this normal map texture is on?  Pretend the light bounces off it like all these surface details were there and the light was hitting them at a different surface tangent angle."  (...basically.)

Confused yet?  

Paying people who understand this (and can explain it better than I can) is why Call of Duty has a multi-million dollar budget.  



So once we have the game-ready mesh and the normal map that we generated from the high-detail sculpt, we import both into the game engine we're using for this - the UDK, or Unreal Development Kit.  Same underlying technology for Gears of War, Unreal Tournament, Bioshock, and a lot of others.  Believe it or not, it's free to download from the company that makes it, Epic Games - you only have to pay them royalties if you sell a game you make with it.

Here is the 1k-ish polygon game mesh in the engine / toolkit.



And here is the low poly mesh with the normal map texture generated with a 788,000 polygon high detail sculpt.  90% of the detail and visual fidelity of geometry more than 750x as complicated - but the game engine and your graphics card / CPU only have to deal with very simple geometry (with a tiny bit of extra overhead for the faked lighting math).





And that is how we make games that have scenes of entire city streets or space stations or 200 airplanes or 10,000 Roman Legionnaires look realistic without flat-out crashing or melting your computer.





I have about five dozen more parts to do with this exact same process.  

Op this is incredible.

OST

No progress last few days.  Awful stomach flu with fever.

Going to go curl back up in front of the fireplace.

You've made me realize all I really want in a game is DS with new assets  I'll have to settle for Planetary Annihilation. All the best people from Cavedog went with Chris.

Hope you feel better man, this is good stuff.

fascinating.

Aaaaaand this is why I stopped playing with the Cryengine 3 SDK. My friends told me how fun it was, and got me all excited to play with it, and then I realized just how much work I was going to have to put into it to accomplish anything. There is so much more involved now than when I was modding games back in the early 2000's.

Looks damn good.

Yep, amazing difference.  I built some models for Microsoft Train Simulator, back around that time.  I didn't have to create the textures, thankfully.  Models are relatively easy, compared to good textures, and understanding how/when to use them.



 

(placeholder - working on something else right now and coming back to the monastery scene later)

OK.  So the most recent project is about to come to fruition.  Plus, it has zombies, and aliens, and modular environment construction.  When I get the OK to post pics...  They go here.

For now, to quote an old art director friend of mine, "imagine awesomeness until it gets here."

EDIT: Project got canned by the producer, bah... under NDA, no pics.  

Amazing OP.  That is REALLY cool!

And now for the latest project, trying to get the attention of a certain studio behind a certain Homeworld series that I damn near worship...

Homeworld: Shipbreakers - Enemy Fighter

Next up is going to be an end-to-end process pic sequence for one of my hilarious attempts at 2D concepting...

Here's that 2D work.

Base concept: semi-futuristic Eastern Bloc-derived VTOL base in a large cave in the Caucasus Mountains.

Early pass - grayscale



Followed by a little more roughing in and a first color wash pass.  I start base color very early, since it helps me wrap my brain around the character of objects, for tighter detail.

Paging Salmonaxe... Salmonaxe, please pick up the white courtesy phone.

neat

This is so cool Rogue-Sasquatch...thanks for sharing!  I love to see the creative process! I wish you could give me some lessons!!

Looking great

I'm nowhere near good enough to teach.  Hell, last Saturday I had a friend from art school over at my place giving me one-on-one tips.  

I'd love to eventually be able to do this someday, though:

Ladies and gentlemen, J. Aaron Kambeitz...

Awesome!

Now  someone shop lootie into it.

More 2D concept stuff.

Step by step so far...





Linework done



Colors start

Last update for the night...

This is cool. With those skills you could probably make a few bucks generating 3D porn.

I'm more an environment artist and hard-surface artist.  Don't do characters very well, and I'm extremely rusty with animation.

I could make you one badass next-gen 3D model of a casting couch, though.  Lasers 'n shit and all kinds of stuff.  

I noticed on your latest design that it really reminded me of the general Homeworld artistic style.  In my opinion, Homeworld has the best most interesting and visually appealing concept art of any game ever.

ETA: my first thought was.. Bentusi???

Good eyes!  I'm on the right track then.

Progenitor.

Homeworld is my bestest favoritest PC game ever.  Ever.  If I can someday paint like Rob Cunningham or Aaron Kambeitz, I'll die a happy man.

I remember the game.

It was pretty and it was long.

Lots of good loot too.   My donkey was packed down

http://www.homeworldremastered.com

Oh, believe me.  I'm aware of it and, uh, then some.  

Think I'll call this one done.  First major 2D piece I've taken end-to-end in a while, so it's rough, but it did what I wanted it to do.

Looks great man. Thanks for the updates.

Starting in on another one.  Experimenting with various lighting - direct vs ambient vs bounced.

New painting is done:

(Full res is 3000x5000 px)

WOW just WOW!! I wish I could do shit like that!