Glitch pictures thread!

fmx · Post by **fmx** » Tue Dec 21, 2010 9:37 am

Look at the Teapot handle - shadow is just as pixelated.
The shadow on top of the Teapot handle, its smoother and less pixelated than near the bottom.

Seems more like a precision issue, you need to adjust your near/far values or use a larger shadowmap resolution, or dont use VSM but LiPSM or PSSM

Mel · Post by **Mel** » Tue Dec 21, 2010 12:25 pm

No, it is a matter of the VSM algorithm itself. Happens that the VSM is an statistical approximation of the shadowmap. Hence, if there are no samples for the stage, the probabilities for the shadowmap to be shading the environment become minimal, and the filter reduces to its minimum. I have "half" solved it inverting the process, the trouble is that now it produces "shadow bleeding", which is, again, a problem of the VSM algorithm itself.
The best of VSM though, is that it solves quite well the biasing problems of the standard shadowmapping.

I also thought it could be a problem of the texture filtering. But it isn't.

The issue with the texture resolution can only be adressed with the LiSPSM, or something the like. And i can't do it properly. Simply, the current LiSPSM implementations don't fit in the Irrlicht engine, and PSSM isn't an option because i have only 1 texture slot for the shadowmap. I have to try more stuff.

fmx · Post by **fmx** » Wed Dec 22, 2010 1:52 pm

Mel wrote:... and PSSM isn't an option because i have only 1 texture slot for the shadowmap

Its possible to combine PSSM textures into an atlas, or use an ARGB format and render multiple shadowmaps into separate channels of a single texturemap

Mel · Post by **Mel** » Thu Dec 23, 2010 1:54 am

That's what i like of you people, you think! unlike me! ^^U Thanks for the tip i didn't thought of that

lazerblade · Post by **lazerblade** » Thu Dec 23, 2010 9:14 pm

Not sure what caused this, but it happened when my screensaver activated
while the program was still running.

Cloudef · Post by **Cloudef** » Sun Dec 26, 2010 3:23 pm

Was doing some late night material system code for GL/GLES2.0 framework.

omaremad · Post by **omaremad** » Sat Jan 01, 2011 1:58 am

All the materials had the shiny tile map applied

Virion · Post by **Virion** » Sat Jan 01, 2011 4:17 am

omaremad wrote:All the materials had the shiny tile map applied

the entire interior was made from gold!!

stefbuet · Post by **stefbuet** » Sat Jan 08, 2011 10:45 pm

Ok, my turn

I've started again to work on my lighting manager, so I have some more to show:

I don't remember what I wanted to do when I get this but it's a nice one:

Fail when trying to compute pixel screen 2D velocity:

Clipping issue (not yed resolved :p) with my motion blur effect, my room might catch the Millenium falcon at that speed

:

The nicest : Using the wrong shadow map with dual paraboloid mapping (when comparing depth / no VSM here)

I got a few more but I'm too lazy to upload it

omaremad · Post by **omaremad** » Mon Jan 10, 2011 4:04 am

Nice pics stef, if you remove the window broders im sure u can get away calling them "modern art"

I love this glitch, but no matter how hard i try i cant make a useful effect out of it.

Radikalizm · Post by **Radikalizm** » Mon Jan 10, 2011 9:53 am

omaremad wrote:Nice pics stef, if you remove the window broders im sure u can get away calling them "modern art"

I love this glitch, but no matter how hard i try i cant make a useful effect out of it.

Looks like a motion blur gone bad ^^

fmx · Post by **fmx** » Tue Jan 11, 2011 3:40 pm

omaremad
looks like the kind of effect that occurs when the renderbuffer isn't cleared each frame, wild

stefbuet
those are some great glitches

if only all my glitches looked as pretty XD

Mel · Post by **Mel** » Wed Jan 12, 2011 12:27 am

Ah! the good'ol Hall Of Mirrors glitch

stefbuet · Post by **stefbuet** » Mon Mar 07, 2011 6:16 am

ChaiRuiPeng · Post by **ChaiRuiPeng** » Fri Mar 18, 2011 7:03 pm

not very pretty. but its much better than last time

real time deformable mesh gone bad
EDIT:
it was the mvTetraMesh (deformable mesh) factory "bug". the indices creater was off by one on a zero based array

so you can guess the results of that.