In this section you can check out what I'm currently working in, practically in real time.
If some of this seems cryptic it's because the logs seen here come directly from my personal productivity tool prio, and are really just "notes to self".
Nevertheless, if you are interested in what I'm currently working on, this space will certainly be updated FAR more often than the news section.
19:07:51 sabk CLOCK: 0.1 days (02:05 hours)
19:07:47 sabk TASK: bigger map
19:02:54 sabk LOG
The world is now also twice as large (vertically), now 10 x 12 screens as opposed to the previous 10 x 6 screens.
19:02:04 sabk LOG
Significantly speeded up the application startup. The delay was the construction of the Tile instances, which were all using implicit Persistent construction, in this case very stupid (the check for unique instance keys).
Switched to the technique used in Saerfus, namely explicit loading of the Tile array.
19:01:08 sabk TASK: use explicit loading for Tile to speed up app startup
17:36:16 sabk TASK: remove deps of Logic on Tile
09:40:00 sabk CLOCK: 00:18 hours
09:25:17 sabk LOG
Finally back on this, working on expanding the world.
09:19:22 sabk TASK: remove nag screen stuff
00:53:21 sandblox CLOCK: 0.1 days (03:12 hours)
20:19:53 sandblox TASK: test shader based texture mapping for mt
20:19:46 sandblox TASK: change mt vertex format to include normals
16:07:21 sandblox CLOCK: 0.1 days (02:41 hours)
13:40:58 sandblox LOG
Moving membrane vertex format (3d position and normal) to dxut::vf::posnorm3, along with relevant utility methods. These are shared between ace, act, and sandblox.
10:29:36 sandblox CLOCK: 0.1 days (02:16 hours)
10:28:15 sandblox LOG
Vertical sunlight tests, very expensive, and requires accurate triangle collision.
Regardless, the mesh needs lightmaps / more uniform light sampling for this to work reasonably. This means that AO will also look crappy unless the mesh is mapped with lightmaps.
This probably isn't very trivial considering that this is a big 3d chunk of geometry = lots of lightmap memory. The mapping will probably also be quite wasteful as each triangle needs unique space, probably mapping onto the triangle plane itself to get equal coverage. I.e. the old hangar stuff, but with the added requirement of having to uv-map algorithmically.
Hard in other words.
However, if one were to stick with the vertical lighting idea, it should be possible to pre-generate a 2d floating point texture that represents the top heights per cell (a heightfield) and then use that in a pixel shader to calculate sunlight vs shadow. Something like this will take advantage of filtering too.
Other lights will have to be local / real time.
This will probably be adequate for lots of stuff. Daylight scenes will not need any other light, and night scenes / space scenes can use just local lights. The splitup of the mesh into chunks is very good for the real time lighting.
AO will probably be very hard and / or require lightmaps unless some kind of realtime solution is used.
10:21:26 sandblox TASK: mt lighting tests
15:15:20 sandblox CLOCK: 0.1 days (01:47 hours)
15:14:55 sandblox LOG
Got mt intersection to work, more stuff is moving into the model.
15:13:50 sandblox TASK: mt intersection
19:53:21 sandblox CLOCK: 0.1 days (01:38 hours)
19:53:15 sandblox LOG
Crazy good results using hi-res mesh and fractal noise! :)
18:14:27 sandblox TASK: more mt tests
18:12:33 sandblox TASK: fix menu escape behaviour
19:22:22 johno.se TASK: advert on imgui page
10:34:17 sandblox CLOCK: 0.2 days (04:20 hours)
08:36:06 sandblox LOG
Much progress on an MT terrain / world.
21:41:08 sabk LOG
BASSMOD worked like a charm!
There is a chance that the control offered (setting song position but still adhering to the "song", i.e. the order of patterns in the file) isn't enough for Mats, but he can then probably create a custom file where song slot == pattern and go from there.
21:40:02 sabk TASK: test mod player
21:39:55 sabk CLOCK: 00:17 hours
15:02:17 sabk CLOCK: 00:59 hours
14:01:31 sabk CLOCK: 00:11 hours
13:45:40 ace_bonus CLOCK: 00:07 hours
13:44:54 ace_bonus LOG
Roids splitting roids; didn't work for the simple reason that roids spawn at the center of their parent, resulting in a cascade of splitting.
13:44:18 ace_bonus TASK: test roid split roid
13:24:27 prio CLOCK: 00:03 hours
13:17:15 prio LOG
Some nifty css magic.
22:12:17 ace LOG
Did some tests with roids bouncing off of each other.
Interestingly enough the bigger roids are always earlier in the linked list than the smaller ones, and the creeps are always later than the medium sized roids, and this made the coding of the bouncing easier. In retrospect this should have been obvious due to how the gameplay rules are.
In the end I don't think that it really adds anything, other than perhaps making the game even harder. I also observed some weird edge bugs that I don't relish chasing down.
Leaving it out.
22:10:19 ace_bonus TASK: test roid bounce!?
12:30:05 eco TASK: open month of submitted receipt