Monday, November 17, 2008

Update 3

A lot of long and fun hours going into trying to finish this level. I have added a lot of sounds and replaced a lot of the ground BSP with terrain, since they are not so flat and can have layered textures. I have also been re-recording voice sounds (because the old ones sounded too much like a robot reading from a script) and have finally figured out how to get sounds to do what I want them to in Audacity. I have also been adding consistency details all around and making the space more visually appealing (I had put this off because I wanted to have scripting finished first, but scripting was taking more time than expected so I had to juggle both). I mostly have all the scripted events done for now, and all of the AI and matinees are working nicely. There is still a problem trying to make a custom ladder work with matinee instead of a ladderVolume (which doesn't work unless pointing north). Here are some newer screens of what I've accomplished. Note that lighting has still yet to be built -- also, I haven't added new screens of areas that weren't much affected since last update.









(I'm still deciding whether I like the outdoor environment to be at night or during the day. I think I'm going to stick with my original plan, though, and make it early morning/light outside).

Sunday, November 9, 2008

WIP Update 2 (MAD)

Second update. Yay! There has been a lot added to the level, but most of it may not be visible in the screenshots -- as most of the work was done in scripting the events for the majority of the level. A couple of days ago I was able to jump into texturing, and the process is so different from scripting that it feels weird to go back and forth between the two. I like it though, because it provides me with more variety and makes each day a new challenge with unexpected results. As far as the AI, I have a more solid system now for evasion time; I originally went with a delay and the bot would go back to its default patrol -- but the delay fired every time the sequence was activated, so they wouldn't be actively pursuing the player for long. To deal with this, I have a variable that stores an incrementing counter value. When the player is spotted, the counter is set back to zero: to prevent the bot from going back on patrol prematurely. Once the player is not found anymore, the counter is able to finish its routine and then the bot is able to return to it's default action. This was a major breakthrough for me, as it was frustrating not having the bot stay on alert when the player is standing right in front of them.
Another major feature I added today is sort of a time-based puzzle. The player has a set of switches, and each subsequent switch can only be pressed if the previous one in the sequence is opened. This way, the player has to play sort of a game of "Simon" where they have to follow the lights and touch them in a certain amount of time. To add more dynamism and support the narrative of a nuclear reconnaissance operation, I thought it would be interesting if the player were to jump on a moving nuclear missile while it is being transported from one room to the next. I initially had nukes moving in the room just for aesthetic and dynamic reasons, but now they play a crucial part in the gameplay and it's actually really fun!
There is so much more to talk about but all will be said in the finished product. For now, here are the screenshots of my current progress:








Thursday, November 6, 2008

Transfer Maps and Parralax Mapping

I've been doing research on making good normal maps for a texture, instead of the default of taking the color map into a normal application like CrazyBump and playing with the settings. What I have been doing is taking my low-poly game mesh and importing it into Zbrush, where I assign the diffuse to the object and paint detail over noticeable deformations. Then, I'll import the high poly and low poly model into Maya and go to Lighting/Rendering>Transfer Maps. This achieves a much more fine-tuned normal map, instead of having random deformations on the surface if they're not supposed to be there (ex: a stain wouldn't cause the normal map to project from the surface, like what would happen in CrazyBump).

Also, another feature in Unreal 3 is the ability to create a parallax map: a combination of a normal map and a faked height map. I have played around with this, and it seems that the more subtle the offset the better, as details start overlapping and crawling over each other if the amount is too drastic.

I truly enjoy modeling and texturing, and now I'm even more exciting to achieve more convincing results in my textures!