Monday, 23 May 2011

Worth a look at....

A good website that explains everything there is to know about animatics and storyboards for production. (completely off topic)

http://www.aboutanimatics.com/production.html

Enjoy!

Poly Count research

So we all know poly count is very important within games and production animation. I've tried to collate some research material on industry guidelines and tips on poly count within productions and games. This will also set me a limit on my own modelling as I can further progress by learning how to model to the specific poly count.

Article from http://forums.runicgames.com/viewtopic.php?f=6&t=2311&start=160

"This is adam ... I am the tech artist over here at runic. I end up putting in a lot of the game asset on the art side so I can probably give you some info on our format and restrictions.

Almost all of our textures our 512x 512 or 256 x 256 and even some 128 x 128. Most of the environment textures for the tileset are 256x 256 so that would probably be the way to go. We end up outputting the textures to dds format. There is a free plug for photoshop at:

http://developer.nvidia.com/object/phot ... ugins.html

We can support full alpha (DXT5) and 1 bit alpha (DXT1). In many cases 1 bit is a better option as full alpha can have sorting problems depending on if a model with alpha has to render under them or over them. The best way if to just try it out in the editor when it gets released. You can always change it to 1 bit if full alpha doesn't work out.

I suggest that if you are going to make textures with alpha always store it in PSD format or some other lossless format like tiff or targa until you are ready to integrate them as dds in the game as dds has lossy compression. Also it is best to create an alpha dds from a texture that has a dedicated alpha channel in photoshop, not like png where the alpha is just stored with the color values (although we use superPNG format that allows a png to have dedicated alpha channel). This is needed so that you can paint the area that gets alpha out a color matching the rest of the texture and helps prevent a white pixel halo. Formats like regular png end up sampling white as the blending bg color between the alpha area and non alpha area creating a halo effect when converted to dds.

Here is a link to the superPNG plugin
http://www.fnordware.com/superpng/

As for your second question it is a wee bit harder to answer. Since our levels are semi random the length of a level and the amount of props changes a bit. What we do is instead of aiming for a level polycount is we aim for each asset to hit a consistent level of detail in both model and texture across the game. Our main characters end up being higher than our props and our props usually end up a little higher than our tileset pieces for there size on screen. A main character might be around 1400 tris, a regular similar sized monster might be around 1000 tris, a props of about the size as a character might be around 300 - 600, and a whole wall tile might be only 100- 200. I know these seem a bit low but we are aiming at having a lot of monsters on screen plus we are looking to have computer specs that allow our game to be played by people that don't have the Rolls-Royce of computers :)"


In comparism to this article, another article from teamxbox.com has figures of poly counts from game assets.


I stumbled upon an interesting blog entry from Rick Stirling, a character artist, about the polycount of game character models. This post is interesting since it quotes some approximate numbers for the models.

Here are some examples that caught my eye:

Gears of War, Xbox 360, 2006
Wretch - 10,000 polygons with diffuse, specular and normal maps
Boomer - 11,000 polygons with diffuse, specular and normal maps
Marcus - 15,000 polygons with diffuse, specular and normal maps

GTA San Andreas, PS2, 2004
Characters - 2,000 polygons with 1 256×256 8bit texture
NPCs - 1,200 polygons with 1 256×128 8bit texture

Halflife 2, PC, 2004
Alyx Vance - 8323 polygons
Barney - 5922 polygons
Combine Soldier - 4682 polygons
Classic Headcrab - 1690 polygons
SMG - 2854 polygons (with arms)
Pistol - 2268 polygons (with arms)

Halo, Xbox, 2001
Masterchief - 2,000 polygons

Metal Gear Solid 3: Snake Eater, Gamecube, 2005 (Small mistake here, it's not on GC but PS2)
Snake - 4,000 polygons

Resident Evil 4, Gamecube, 2005
Leon - 10,000 polygons

Jak & Daxter, PS2, 2001
Jak - 4000 polygons

Jak II, PS2, 2003
Jak - 10,000 polygons*

Lost planet, X360/PC, 2007
Wayne - 12392 polygons (but finally 17765 polygons for compatibility with motion blur effect)
VS robot - 30-40K polygons
Background - ~500K polygons
Peak number of polygons per frame - ~ 3 million**

Dead Rising, X360, 2006
Peak number of polygons per frame - ~ 4 million**

The Legend of Zelda: The Wind Waker, GC, 2002
Link - 2800 polygons

The Legend of Zelda: Twilight Princess, GC/Wii, 2006
Link - 6900 polygons

Super Mario Sunshine, GC, 2002
Mario - 1500 polygons
Levels - ~ 60k polygons

Dead or Alive series, Xbox, 2001-2004
Character - ~10k

Vitua Fighter 5, Arcade/PS3/X360, 2006
Character - ~40k with diffuse, specular and normal maps
Background - 100K - 300K polygons

Medal of Honour: Allied Assault, PC, 2002
Character - 4096 polygons

Project Gotham Racing 3, X360, 2005
Cars - 80K-100K polygons (interior + exterior), damages add between 10K and 20K more polygons per car

Gran Turismo 5: Prologue, PS3, 2007
Cars - 200K polygons (probably interior + exterior)

Midnight Club, Xbox360/PS3, 2007
Cars - 100K polygons

Gran Turismo 3, PS2, 2001
Cars - ~2K-4K polygons

Gran Turismo 4, PS2, 2004
Cars - ~2k-5K polygons

Lair, PS3, 2007
Main dragon plus its rider - 150K polygons
16x16KM scene - 134M polygons (streamed into memory, not loaded at run time)

*Might be a cut-scene model
**No precisions whether it's actual rendered polygons or just the number of polygons sent to transform, pre Z-pass and culling.

It has always been interesting to me to see how many polygons were used on a certain game model. It's trivia, in most case, but I like these graphics related trivial tidbits.

Thursday, 12 May 2011

Modular assets

The problem I found when building the warehouse in district 1 was because it was basically 1 big extruded cube, it was very hard to add certain details without messing up the geometry. I found the solution to this problem in a book called Game Environments and Props by Michael Mckinley.

The solution is simple.... When creating large buildings the best thing to do in most cases would be to simply create different modules or parts of the building in different scene files. This technique has many advantages:


  • As the game engine is rendering in real time, that presents us with many limitations, one of these being that the more varied the buildings are, the longer the loading time is going to be.
  • Allows more detail to be put into buildings.
  • Variations of the same modules can give the illusion of different buildings simply by re-arranging them differently within the game engine and the more modules there are, the more varied buildings your game can include.
This is a screenshot of my modular building constructed of 5 different modules all constantly repeated to create the building. Game Environments and Props.




As you can see within district 2 I have created many variations of buildings with the same modules

Wednesday, 11 May 2011

Layout sketches to technical layouts

The transition from layout sketches to technical positioning in maya is one process that was made very easy thanks to the likes of photoshop and my scanner.


With the basic layout completed on maya I could then use the top view to print a snapshot of the area I had to work with and use it to improvise on my original layout sketches.


By working on top of the printout I could map out key assets such as the bridge network and the rail network in accordance to my drawings.


Using photoshop I layered the original drawing over the top view snapshot and then improvised the train lines positioning within the scene.


With the tracks laid down I then drew on the overhead bridge levels that would complete the layered look of the trainyard.

I found this way of working to be a very quick and easy way and also worked very accurately in terms of scaling and positioning. With this way of mapping out the level I could apply the same principles to district 2.

District 2 leading on from district 1

The slum maze of District 2

Referencing for game

This stage in the production of the game I believe was the most important for me as a level designer. It was the stage that would take my ideas from concept and by the use of real structures, use parts of existing structures I found interesting and that would provide me for a basis on my models.


I found that the surrounding area plays a very influential part when it comes to environment design and being in Birmingham, im pretty fortunate to have some great reference structures. Heres a few images I took during my research stage that have in many ways influenced my outcome.


This glass bridge inspired my own design of bridge which you can see within the environment.


A trainyard in india. 

Monday, 9 May 2011

Character Concept design

Although I never really focused much on the character side of things as it was handled mainly by my team members, I did a few concepts for the characters.


 My concept sketch of Barrel (The archetypal heavy weapons character)


Another concept for Barrel, later decided he looked more like an end of level boss.


Concept for Clip (Main protaganist) 
Concept for a roll-away suitcase that opens into a guided turret. (bottom right)


Concept for a turret made from a reinforced camera tripod.

As our main characters are from the wastelands, their equipment was designed to look like it was bits of scrap put together.

Sunday, 8 May 2011

Animatic & Storyboards

The final storyboards for the animatic






Animatic


video

The Final animatic unfortunately does not render the sound but this is the half finished version.
Finished version coming soon.....


Saturday, 7 May 2011

Game Creation

So its been a very long while since i've posted anything on here during the designing and developing stages of the game creation. As you could expect, alot has changed and much of the games original details have been developed.

The updated outline of the game:

- The game will feature an upgradeable armour and weapons system and has taken to other RPG elements such as improving attributes etc.

- The game will feature the main character who essentially is the average joe 'rookie' type character who can be developed however the player chooses. The other characters will be NPC's and generally a voice to guide you through the game.

- Although the game isn't a sandbox style game it will feature many elements and exploration will be very much possible over the vast 3 levels.

- The setting of the game is still post-apocolyptic. The Utopian city is in fact a pre-disaster Transport Hub which people began to settle in due to its stability and infrastructure.

- The central transport hub is a enormous pyramid structure.

The job that was left now was to begin the concept art and character design etc. Although I did a little character design I really wanted to focus on the environment side of things as I felt that a great environment can make a great game.

To begin on the environment I had to formulate ideas and collate reference imagery so with a hands on approach I took a digital camera everywhere I went for a few weeks and photographed reference images as well as reference textures. I feel that referencing environments is a very essential step because although I was working to create a post apocalyptic environment, I wanted its foundations to be rooted in reality so we as the audience can look at the environment and believe it to be something real and right in front of us.

A moodboard of reference photographs taken over 3 different countries.