Could you give us a short introduction to DRPG?
I tried to make something between a roleplay game and a strategy game. The main part will be about discovering new creatures and destroying bridges. water, and plants growing in ways never thought to be possible.
This is your second SGA entry, how closely related is DRPG to your first game, DPSA?
They are both made to be easy to modify without restarting the game. The difference is that the physics is simpler to allow many units.
You are working on DRPG by yourself, how are you getting all the hard work done?
I use reusable modules for half of the code when it's near 15000 lines. Using a dynamic class module makes the code 1000 times slower but can be used for file access.
I use a graphics engine with a simple interface and a better file format than the *.x format. 3DState's *.wld format is easy to read in notepad when something exported can't be opened.
What tools are you using to create DRPG, and how have they been working out for you?
I use the full version 8 of the 3DState engine who I got from making an SDK for it.
Gimp is free and better than Photoshop for drawing but most people doesn't know how to use it since Photoshop is used in schools.
I'm using my own application for formal calculus to generate prooven code from my own language that will have the same speed on different computers. It compiles from infinite differential equations to finite differential solutions as a module to a specified language using multiple backends. I have reused about 90% of it's code from DPSA.
I will use my own sound creator. I use my own database system for version compatibility so that I can keep my levels after adding something new.
A lot of the 3D creating tools are a pile of crap and I'm working on a new 3D creator to replace them instead of cleaning up after them using my 3D editor.
You talked about using self evolving AI in the game, how is that going to work, and what precautions are you taking to make sure your AI don't go berserk and wipe out the human race?
Preparation: Before the real implementation, I made a small scale evolution to solve a 3D equation and find a gene blending method who is fast enough. If some properties are independent, the mixing of genes can make the time go from about (A+B)^2 to (A^2)+(B^2) depending on the problem to solve. Using a good blending and adaptive mutation, it solved a complex equation to machine precision in one second.
How it was done: I used it for balancing the personality by allowing the computer to play against a mutated copy of itself and letting the winner give gene segments to the looser. Therefor I must select vital parts of the code who needs fuzzy logic. The easiest way to make it hang free is to have a lot of arbitrary constants who will be choosing code segments and scaling formulas instead of the free arbitrary self editing function who would need automated verification.
What happened: After 5 hours of waiting in by bed an watching TV, the robot was better than me at balancing the economy and I had to learn how to play from the robot to know how many farmers I should start with and how many troops to gather for attacking. I didn't have a chance to defend my HQ because the AI had a much better economy. Later in the development, the game became too advanced for the evolution when I introduced defensive units. The evolution needs fast battle results to be useful since a longer playtime requires more performance. Therefor I had to remove that part of the game and just use the result from the evolution.
Sci-fi: If I would have a computer fast enough for free evolution who isn't bounded by rules, I would make these parts in my virtual machine since it can't communicate to the OS directly to cause any crash. If it would escape, it would only be made to attack an enemy referred to in the game and know nothing about the world outside. It would be unlikely that it would be a multi purpose AI. It would be like a bee flying in to a window because it wasn't made for it. Evolving a human like AI would require a randomized problem generator and more power than a supercomputer has just to run it.
What is your design philosophy for DRPG?
Quantity is more vital than heavy shaders and physics to be more like a strategy game. I try to make it possible to sneek with a small number of units for the RPG part but that's colliding with the strategic fairness when only the player sees everything.
Is your development strategy different this year from when you developed your last game?
Yes, I use a much slower database in a reusable class module because the optimizing tasks in the beginning will take a lot more loading time than the loading from the file. DPSA only used a static module to make it fast.
What has been the most difficult part about developing DRPG, and do you have any tips for other SGA developers?
The low level AI for finding the optimal path and optimized vision intersection was the hardest part. The vision intersection must allow long lines intersecting the collision blocks diagonally over the whole world like a voxel engine.
The most important thing when making the game is to think about AI before physics and physics before graphics. If you make a free polygon model without thinking about pathnodes or collision detection, you have to make the later parts 10 times more advanced and the CPU will not have power to many units.
Optimizing paths using nodes and links is only good for large static worlds. For a small world with destroyable walls, a simpler system like A* must be used since it takes seconds to make a node-link index.
It's worth to make a virtual machine for the story since the story has to be easy to remove without a trace in the core. It's even more important if many people are involved in the project since it's easier to specify the higher language than the game's core.
My virtual machine has a language close to lisp and fortran to make compiling fast and reliable to virtual assembly. For example: Call MyFunction 2.63, +(7,3), M(+(&Start,Offset)) : Set X to Out(0) : Set Y to Out(-(2,1))
What are your 3 favorite games and your most hated game? And how have these games influenced you and DRPG
Best games:
1 "Deus ex" because I can find new secrets in the story after finishing the game for the 14:th time.
2 "Stronghold crusader" because knowing the unit's weakness makes 90% of the victory.
3 "Sid meiers Alpha centauri" because lies is more dangerous than nukes.
Worst game:
"BackPacker 3" because it's not even good for learning when it gives the wrong answer to the wrong question. If you answer wrong about a year, it says odd things about flying elephants by mixing words by random.
That's it for this time folks. A big thanks to David Piuva for the interview. Remember to check back as there will be more news and Spotlights as we get closer to the deadline. Until next time, happy new year! May your year be full of gaming and game development.
