3rd Challenge entry - Futuristic rts game

[User137]

I haven't worked it out fully yet. I need to make lots of UI elements to make the plans for the AI, and i'm propably going to hard-code all the UI. There will be no bounding box selection, and if i'll allow unit selection it will be for 1 purpose, to show details about it. Bit like other strategy games, you can select a building from toolbox, and set where the construction yard will be at. But unlike other games, i also determine the building size upon placing it. Then nearest available workers start coming to build it, with no interaction from me (this is how Settlers works). I can determine how many workers (they fly in this game) i want made for the empire, and in games we know (TA) it's sometimes taking away from military production queue.

I'll also divide the map to 10 sectors horizontally and vertically to minimize amount of processing done. Each have their own list of units, so say if bomb is dropped on sector (2,4), then i only need to check unit collisions from there only. If the blast radius extends near the border, or even within corner, in worst case i'd check 4 sectors for that blast. But if AI has 500 units at sector (8,6) or something, they will be completely left out of calculation. It is also a boost for rendering. I can also use the mass collision i used in the other demo, with the sector optimization, and 1 other trick.

Yes it's propably a bit ambitious idea, which is why i won't start from 3D animations. There'll only be Skirmish game mode, no story, and no identifiable commander, unless time permits and i get new ideas. Other obvious problem will also be to make the enemy commander AI. I need to get enough playtime myself to know how it should be played, so that i can tell AI how to play. I am also cautiously worried about outposts, and UI related to them. World might just be a low varying heightmap from simplex noise, with single tile type for now. No looping from borders this time. I have done it before, but it does add bit more confusion and i can't afford the time.

[User137]

Will see what kind of enjoyment level i can achieve with this, or will it again leave a technical and dry experience...

Progress so far, i've implemented some game objects and models (10 different units, 2+1 temporary models). Their ingame properties are configured in ini file. So much configuration options are in ini, that one could almost redesign the whole unit concept without touching source code. So with these working, i have first units on screen, with physics enabled. It is slightly different from the walker demo, basically i am simulating units with different mass. Sector system is also stable, but it's not used everywhere yet. Maybe i'll hold first screenshots till later, there is really not much to see yet

[WILL]

Will see what kind of enjoyment level i can achieve with this, or will it again leave a technical and dry experience...

This is a questions I ask myself a lot with pretty much all of my own projects. And it's a good question to keep asking from time to time. I would suggest to be sure to try your ideas before you outright discount them. You never know how all the elements of your game will fit together as a "whole experience" until it's there and you have a decent session of testing to see where your pros and cons lay. Like soup sometimes all it needs is a little "salt."

[User137]

Sorry, it's getting along very slowly, i have been spending my time elsewhere afterall... But i decided to pick up some pace today, by finishing the things i kept having motivation problems on. When i start coding i can accomplish alot in big bursts... as long as i have slight clue what i'm doing.

Procedurally generated simplex-map terrain also with calculated light normals. Terrain and game objects all properly follow sector system now. Added simple smooth shadows before first screenshot. Mouse controls are done for scrolling/zooming the view. Doing some planning for UI on text document. I think main elements to come soon are unit construction toolboxes and mini/battle-map. It is easy to lose track on where the base is without it.

[User137]

I found out another cause for low framerate with high object amounts. I separated smoothing groups for flat faces, and this caused at worst like 800 calls to vertex array rendering per object, if each face was 1 group. I added new procedure to nxPascal model class to combine groups with same material, to 1 group. This gave a significant speed boost, and the process itself is unnoticable. While taking this screenshot with 500+500 objects, game was at full 60 fps, and 0% cpu use (some new things to be spotted there too ):


The AI behavior and its controls for player are already planned out till the end, but i hope i can implement them...

[Sascha Willems]

Looks mighty interesting. Do you think you get it done in time? Even if not try to at least make a public prototyp of it, so we can take a look at it

[WILL]

Wow, that's quite the amount of little guys on the screen.