kagura update

Here's what she's looking like lately with the same data set seen many times. Using FXAA now since it's trendy. Will I have the lighting and AO in a working state for more than a week? Hopefully that'll all come together in a couple of months while working on the new editor. Might have to give up on using blender with the telemetry system after all due to workflow issues.

I should do a write-up on the telemetry system used for editing ( lights, materials, etc ) inside the webbrowser at a later date, however now I'm working on supporting more things with the telemetry system besides TTY / HTTP transport and may hold off until that's worth talking about.

Cool!

I saw this on altdevblogaday a while back. Embedding a webserver in your game engine, and then using javascript on the webbrowser to visualize it.

Just tried it tonight... btw the tutorial is wrong in several parts of the examples from C to javascript to JSON, but it's the idea that counts. :D

Bon Odori test

I wanted to make a full O-Bon demo, but instead I just spawned some dynamic torches ( animated lights ) following a simple path like a particle system real quick. Need to fix lighting and visibility again new design new breakage. Hurts framerate to spew all the debug and error test each render command too.

Haha! Fixed a dozen bugs texcoords ain't one.

Sometimes you need to step back an enjoy the test materials on your programmer art. About to get realtime dynamic light editing hooked back up via replication to edit levels 'live' again. Good times playing around in the sandbox.

Yeah... I'm going to need you to rewrite all the light and material shaders.

With E3 coming up I barely have time for my hobby engine. Still playing around with comic look. About to get serious about fixing the deferred lighting with SIMD based tile rendering for box, point, and spot lights. But for now the same broken shit. People asked how horrible it currently looks, so a quick peek of a debug session.

Postprocess fun

I've been playing with using basic laplace edge detection to add more definition to the scene as a 'sharpen' style of convolution using inverse color. It could use some work, but I really like the effect. Not bad for pure image space I guess.

I wonder about modifying it so can be filtered per material. That would be pretty easy using a spare component out of one of the G-buffers.

HHmmm...

Found a new bug causing my colortargets to get cleared when I don't really want them to be. Likely due to be being 3am and I haven't wrote new buffer code yet. Parallel lighting on PC is going to fucking suck.

Here's an init prelight pass for a test character for fun. Not much to look at, but better than my 'oh shit' debug pink.