Messages

I guess technically it's VBO mode, as I'm loading all of the chunks' vertices and data into memory at startup/chunk update. As for the amount of polygons, it renders two per block face (pretty standard, I don't have poly grouping or anything fancy), and only renders the block faces it calculates to be visible (colliding with an air block).

It's more like 500-800 now that I added placement, and I achieve this by grouping all the blocks in a chunk into a single object, so instead of ~300 draw calls per chunk like I was getting before, it's now only 1 per chunk.
Here's a video:

Yep, world size is 16x16 chunks, or 256x256x256.

Here's a good one:

It's working really great on my Minecraft clone. Looks good so far! Here's a video of its current status (terrain, physics (still buggy), block placement, walking, etc.)
https://drive.google.com/file/d/0B2Ez6zqWxZAZVWpyMjRKZEhtZWM/edit?usp=sharing

I've been looking into fractal terrain generation lately, as I'm currently working on a Minecraft clone and want to have a good randomized, realistic terrain. Well, I ended up porting an old C++ Perlin Noise routine into GLBasic, then adding a GUI wrapper to control its output.

At the moment, it only draws and exports 2D representations of terrains, but I will soon add in 3D viewing/exporting.




New 1.2 release on 7/29!
Download it here: http://sblectric.com/downloads/TerrainGen-v1.2.zip
C++ and GLBasic code included for the actual algorithm.

Looks awesome, seriously it brings that retro racer feel right back.

Oh jeez, a story... well, that will come soon enough, I guess.

Yep, I will eventually make a simple model from scratch to get rid of all the glitches and provide further speedups. The master plan is really open ended. Track editor is a possibility in the far future; while internet play is certainly more feasible in the shorter term.

For reducing the car model, I used AC3D's Reduce function, which simplifies your model. I went from ~22,000 polygons to ~2,500 polygons for the shadow model. As soon as I did that, it brought my frame rate right back up to 60. As for more simplification, I don't think using a plane with a .png would work that well, and would like pretty bad.

Something that is bugging me is that one of the problems I ran into with the stencil buffer was the fact that, with the full 20000+ poly car models, it ate up CPU usage to the point where the frame rate dropped from 60-70 down to 30-35 (this is with a high-end CPU and system, so I couldn't accept those frame rates if I were to ever release the game). So, I had to make a reduced poly car model for the shadow drawing, like this simplified code:

Code (glbasic) Select Expand


FOR fx=0 TO 1
    IF fx=0 // normal lighting
    X_SPOT_LT 0, 0xFFFFFF, 0,100,0, 0,0,0, 360
    X_DRAWOBJ car_full,0
    ELSE // stencil buffer shadows
    X_SPOT_LT -3, 0x000000, 0,100,0, 0,0,0, 360
    X_DRAWOBJ car_reduced,0
    ENDIF
NEXT


It works great, but there are a few glitches with the car shadows here and there.
I could always add in the full model to the shadow processing later via a menu option, eg. Shadows - Ultra or something.

Thanks everyone, I appreciate the positive feedback!

Finally, the second preview! I have fixed the vast majority of physics issues since the last update. 
Pretty much everything is new and VASTLY improved.

Currently, I have implemented:
- Set game resolution (fullscreen / FOV as well)
- Circuit race track
- AI driving routine
- 3D car engine sounds (old fmod library)
- Newton physics
- Tachometer (no gears/transmission yet)
- Race position in the leaderboards (a toughie)
- Stencil buffer shadows (a REAL toughie)

Soon to be implemented:
- Track obstacles (breakable fences, cones, etc.)
- Split time
- Lap counting
- Speedometer
- Gears and transmission
- Transparent windows

In the not-so-far-off future:
- More cars
- More tracks
- More sound effects (even a switch to OpenAL)
- Menus and settings

The project is coming along nicely, but sadly it has been slowed somewhat by my college priorities. 

Thanks, Moru. I looked at the API info, and found that you can get a short piece of info about a user. Now it is working perfectly. It seems that Reddit has a 30-second cache for this info, whereas I don't know what was happening before.
Perhaps Reddit caches larger pages for longer or something.
Now, it's perfect. The file user/username/about.json file is less than 512bytes, and stores all the user info! Woo!

Code (glbasic) Select Expand

dat$=NETWEBGET$("www.reddit.com","/user/"+user$+"/about.json",80,512,1000)
Now I can continue! Thanks everyone.

I see. Both NETWEBGET$ and NETWEBGET have the same issue for me.
Would there be any better way to perhaps access the karma and other info in a more direct way then NETWEBGET$ or NETWEBGET? Because I don't really like the fact that I'm either shoving 13KB into a string variable or downloading it onto the drive to then have to access later anyway.

As far as I know, the structure of the page stays the same. The karma is stored in between the 12kb mark and the 13kb mark.