Random "work in progress" shots

category: general [glöplog]
tomkh, well spotted :)
added on the 2017-05-27 22:21:12 by Virgill Virgill
Playing around while working on PBR and postprocessing: (this isn't for any demo in particular, just laying the groundworks for common things)
BB Image
BB Image
BB Image
BB Image
BB Image
I'm still not quite happy with the DoF, but I love the effects you can get with rough reflections.
added on the 2017-05-28 12:43:21 by cpdt cpdt
cpdt: Looking very promising! Waiting to see what you can come up with.
added on the 2017-05-28 13:07:32 by noby noby
cpdt: the 3rd one looks amazing!
cpdt: :D
added on the 2017-05-28 14:54:31 by ferris ferris
@cpdt: yeah, 3rd picture is ACE!
added on the 2017-05-29 11:24:14 by rez rez
Thanks guys :) I've got some more pretties that I'd love to show off, but I'd better keep some stuff secret for a prod :P
But in the meantime, I've got a synth to write...
added on the 2017-05-29 11:50:18 by cpdt cpdt
cpdt: Cool indeed! BTW. can you reveal us the tech you use for the blurry/rough reflections?
added on the 2017-05-29 12:27:36 by deepr deepr
deepr: reflections (from raymarching) are written to a separate texture along with roughness, which is then blurred before being composited back on the original texture. I think this is similar to how ferris did it in Engage, and I believe he goes into more detail in his Engage Breakdown video.
added on the 2017-05-29 12:43:51 by cpdt cpdt
yes, that's how the reflections in engage work, plus the detail that the blur ignores samples with significantly different depth, normal, and roughness (which I'm sure you gathered implementing it, but just for completeness here)
added on the 2017-05-29 18:41:37 by ferris ferris
I also did a bunch of tests with adjusting normal and using different kinds of blurs; in the end perfect reflection blurred with an edge-aware box blur looked the best. I did _not_ however try to weight samples based on a brdf-based approximation or try any temporal convergence (both mentioned here) which I'm sure would give better results, but this isn't the worst-looking compromise (however it's not the most performant thing, but that can be solved other ways if it's a bottleneck)
added on the 2017-05-29 18:44:21 by ferris ferris
I think the thing the technique misses most is the "contact hardening" mentioned in those slides, so I'd like to play with some more things to improve that at some point
added on the 2017-05-29 18:45:27 by ferris ferris
and "specular elongation" (they're both "artifacts" of the same phenomenon as I understand it)
added on the 2017-05-29 18:46:06 by ferris ferris
cpdt: Very pretty!
added on the 2017-05-29 19:42:26 by cce cce
ferris: yep, i'm currently only doing edge-aware in regards to depth and haven't had any issues in my limited testing, but I guess there would be some cases where other properties are required. I did try to get some sort of 'contact hardening' by changing the roughness parameter based on reflect distance, but couldn't get it to work well. That presentation looks very interesting though!

cce: thanks, glad you like it!
added on the 2017-05-30 08:29:06 by cpdt cpdt
I started something new :D

BB Image
added on the 2017-06-06 03:33:52 by rez rez
^ <3
added on the 2017-06-06 03:59:46 by Rob Rob
Rez now thats porn
added on the 2017-06-06 08:17:51 by leGend leGend
Rez: Pico-8? Love it! :D
added on the 2017-06-06 09:02:40 by Raven^NCE Raven^NCE
pico8 road rash remake!
added on the 2017-06-06 10:35:33 by bloodnok bloodnok
hang on!
SUPER hang on!
added on the 2017-06-06 12:16:22 by leGend leGend
I see a boat !
added on the 2017-06-06 22:21:36 by skarab skarab
finally a decent version of DONKEY.BAS
shots how hall of the mountain king evolved:

BB Image
BB Image
BB Image
BB Image
BB Image
BB Image
BB Image