Re: Glitch pictures anyone? back
Board: Board index ‹ Ray tracing ‹ Visuals
(L) [2008/04/22] [funky] [Re: Glitch pictures anyone?] Wayback!>> toxie wrote:images (C) funky
Thats me [SMILEY :D] ... thanks for the upload toxie!
(L) [2008/04/23] [toxie] [Re: Glitch pictures anyone?] Wayback![IMG #1 Image]
Image (C) hanatos
[IMG #1]:
![[IMG:#0]](images/f838860c953727624b48753e50a2903bfc19c6a0bf9fb1e883cc6631669aa463.jpg)
(L) [2008/04/23] [viergewinnt] [Re: Glitch pictures anyone?] Wayback!A little error within the intersection routine... [SMILEY :)]
(L) [2008/05/01] [toxie] [Re: Glitch pictures anyone?] Wayback![IMG #1 Image]
Strict Aliasing Probs.. (this is meant to be a car [SMILEY ;)])
[IMG #1]:
![[IMG:#0]](images/b941df3b08dca6b0626f5571c05ed138f124e81e4c73c607345afa2858ee9a09.png)
(L) [2008/05/03] [Vilem Otte] [Re: Glitch pictures anyone?] Wayback![IMG #1 Image]
My try to optimise ray-sphere intersection code using ASM and integers (no floats) [SMILEY :lol:] (It's little older issue, but I just found old application with it)
The main issue was in ray-sphere collision code and in traverse. In ray-sphere collision - I was using integer based calculations, every number was 256*number (to get some kind of "better precision") - 256 was just moving number 8-bits to left (multiplying by 256) or 8-bits to right (dividing by 256) - and this kind of operation is very very fast in ASM. But one bad division and collision is bad (and of course I made a mistake in that long-ASM-code [SMILEY :lol:] ). In traverse I used unsigned integer instead of integer, and unsigned int doesn't have a sign (who'd expect that) ... so there are little artefacts on screen (LITTLE!!! [SMILEY :lol:] - whole screen is bad).
Anyway those issues we're repaired long time ago [SMILEY :D], I hope
EDIT: This had to be ONE sphere [SMILEY :lol:]
[IMG #1]:Not scraped:
https://web.archive.org/web/20110212061307im_/http://www.otte.cz/engine/data/RaySphereIntHapr.JPG
(L) [2008/05/09] [Beltane] [Re: Glitch pictures anyone?] Wayback!>> fpsunflower wrote:A failed attempt at optimizing ray/triangle intersection:
[SMILEY :D] I like this! The little fellas look positively Greeble-esque!
(L) [2008/07/23] [quak] [Re: Glitch pictures anyone?] Wayback!My first attempt at a custom built BVH visualization. Although the result is quite unexpected it is nice to look at nonetheless [SMILEY :D]. I can sense the spirit of abstract cubism...
EDIT: Well, the compiler is kind of improving...from cubism to the cartoon age [SMILEY :D]
(L) [2008/07/23] [ingenious] [Re: Glitch pictures anyone?] Wayback!>> wazlaf wrote:Something went terribly wrong inside that cornell box!
Are you blind [SMILEY :?:] [SMILEY :?:] [SMILEY :?:]
This is terribly beautiful! My personal favorite of all pictures here! [SMILEY =D>]
This is one of the best Cornell Box renderings I've ever seen. Respect! You are welcome to code bugs in my ray tracer [SMILEY 8)]
(L) [2008/07/24] [davepermen] [Re: Glitch pictures anyone?] Wayback!>> wazlaf wrote:Something went terribly wrong inside that cornell box!
that'll look awesome in a club!!
it's really really beautiful [SMILEY :)]
but yes, if you have not planned to have lights from the edges of the walls, it's wrong [SMILEY :)]
(L) [2008/07/24] [phresnel] [Re: Glitch pictures anyone?] Wayback!>> tarlack wrote:No no, I'm sure my shadow maps work perfectly.
is that a car from "Mask" (the animated series in the 80's/90's)? but then they had no such models way back then ... [SMILEY :)] cool model!
(L) [2008/07/24] [beason] [Re: Glitch pictures anyone?] Wayback!>> greenhybrid wrote:is that a car from "Mask" (the animated series in the 80's/90's)? but then they had no such models way back then ... cool model!
Wow, side mounted missiles would be great for clearing traffic. What a great idea!
(L) [2008/07/24] [tarlack] [Re: Glitch pictures anyone?] Wayback!>> greenhybrid wrote:is that a car from "Mask" (the animated series in the 80's/90's)? but then they had no such models way back then ... cool model!
Ah, I haven't seen this serie, I made it a while ago without any precise source of inspiration [SMILEY :)] pleased that you like it [SMILEY :)]
>> beason wrote:Wow, side mounted missiles would be great for clearing traffic. What a great idea!
I thought to that kind of use when I added them [SMILEY :mrgreen:]
(L) [2008/07/25] [Oatmeal] [Re: Glitch pictures anyone?] Wayback!>> wazlaf wrote:Something went terribly wrong inside that cornell box!
That is a very neat image. Do you know ..um.. how exactly you did that? I sort of want to replicate it..
(L) [2008/07/25] [wenzel] [Re: Glitch pictures anyone?] Wayback!>> Oatmeal wrote:wazlaf wrote:Something went terribly wrong inside that cornell box!
That is a very neat image. Do you know ..um.. how exactly you did that? I sort of want to replicate it..
Sure [SMILEY :)]. This happened in an adjoint formulation path tracer (aka light tracer), which had a bug in the area light sampling code. All emitted particles had their Z component set to zero, so they went straight to the sides. The spheres are black because their BSDFs are delta functions and therefore deterministically creating the sphere->eye link always leads to a BSDF value of zero.
(L) [2008/07/28] [toxie] [Re: Glitch pictures anyone?] Wayback!although beautiful i just edited all posts that quoted the image, so we don't have the same picture over and over again on the same page
(L) [2008/07/28] [lycium] [Re: Glitch pictures anyone?] Wayback!cool pic [SMILEY :D] however,
>> wenzel wrote:The spheres are black because their BSDFs are delta functions and therefore deterministically creating the sphere->eye link always leads to a BSDF value of zero.
finite-area sensor to the rescue! this is similar to how path tracing "wouldn't work" with point lights unless you use a deterministic connection.
(L) [2008/10/24] [phresnel] [Re: Glitch pictures anyone?] Wayback!recently when I was playing with perturbation (including some sine terms and stuff) of input coordinates to the Perlin-Noise function, it happened to show up that the bilinear-interpolation was obviously buggy [SMILEY :)]
tmp-smaller.jpg
(L) [2008/10/28] [toxie] [Re: Glitch pictures anyone?] Wayback!i smell art here
(L) [2008/10/30] [slartybartfast] [Re: Glitch pictures anyone?] Wayback!Art indeed !! This is one of my favourite threads on this forum.
(L) [2008/11/01] [auld] [Re: Glitch pictures anyone?] Wayback!>> ypoissant wrote:When I implemented Photon Mapping for Animation:Master, I made a blooper page I called "[LINK http://www.ypoart.com/tutorials/photon/bloopers.php Different ways to render a Cornell Box]"
What a great series of images. Some of those are almost artistic. I wish I could write shaders like that.
(L) [2008/12/07] [bobmitch] [Re: Glitch pictures anyone?] Wayback![IMG #1 Image]
Making the random seed a function of your x/y coordinates is not always clever... [SMILEY :)]
[IMG #1]:
![[IMG:#0]](images/375af29bfcd2c70103af161fd1e5abeae254e76fa340e405b4963a68d6e7da41.jpg)
(L) [2008/12/08] [toxie] [Re: Glitch pictures anyone?] Wayback!i used exactly this "effect" to get some artistic effect in the "4 million polys and some moonlight" 4k actually.. [SMILEY :)]
(L) [2008/12/08] [phresnel_] [Re: Glitch pictures anyone?] Wayback!I read "seed (x+y)" in that image. And a woman in a red dress.
(L) [2008/12/08] [davepermen] [Re: Glitch pictures anyone?] Wayback!it looks great. like the "pencil" effects in some paint programs.
btw: for random numbers based on x/y, i've just seen
Code: [LINK # Select all]//noise producing function to reduce banding (got it from someone else´s shader):
float rand(vec2 co){
return 0.5+(fract(sin(dot(co.xy ,vec2(12.9898,78.233))) * 43758.5453))*0.5;
}
in a pixel shader on gamedev (for ssao). it looked to behave quite randomly.. at least more so than the glitch picture [SMILEY :)] still it's marvellous.
(L) [2008/12/18] [beason] [Re: Glitch pictures anyone?] Wayback!Adding:
Code: [LINK # Select all]#define double float
to smallpt produces little visible change using gcc-4.2 at home. Using gcc-3.3 at work however, I get this:
bug.png
Maybe I should downgrade!
(L) [2008/12/18] [phresnel] [Re: Glitch pictures anyone?] Wayback!Beason: That's wonderful! I think I'll release a version of picogen with that bug [SMILEY :D]
...
Honestly I am a bit nervous, as I have build that dwarf this morning on windows with gcc33 (which I found out just 5 minutes before I had to leave to work :S), but then, it's using doubles, so it should be okay!
Süper Thread! We shalt somehow conserve all the images to prevent bitrot (web.archive.org is generally an asss on images).
(L) [2008/12/18] [tbp] [Re: Glitch pictures anyone?] Wayback!>> beason wrote:Adding '#define double float' to smallpt produces little visible change
... because most computations still happen in double precision due to promotion rules; throw in '-fsingle-precision-constant' for more radical - but not definitive - effect [SMILEY :P]
(L) [2008/12/18] [BlindSide] [Re: Glitch pictures anyone?] Wayback!Not a beautiful picture, decided to go for the more traditional "wtf is that!?!" glitch.
(L) [2008/12/18] [beason] [Re: Glitch pictures anyone?] Wayback!phresnel: Thanks [SMILEY :)]
tbp: Thanks for the tip. I suspected it was something like that.
BlindSide: Looks like some type of Combine apparatus from Half-Life 2.
(L) [2008/12/30] [playmesumch00ns] [Re: Glitch pictures anyone?] Wayback!Those last ones are quite interesting toxie. What caused them?
(L) [2008/12/31] [monkeyman] [Re: Glitch pictures anyone?] Wayback!>> wazlaf wrote:
Something went terribly wrong inside that cornell box!
this one is nice...
it looks like an inverse of the lighting except for the shadows [SMILEY :)]
(L) [2009/01/06] [squeen] [Re: Glitch pictures anyone?] Wayback!"Trouble with Surface Normals?" -or- "The Real Reason Dinosaurs Became Extinct: Shingles"
dinos.jpg
(L) [2009/01/06] [Michael77] [Re: Glitch pictures anyone?] Wayback!>> squeen wrote:"Trouble with Surface Normals" -or- "The Real Reason Dinosaurs Became Extinct: Shingles"
Cool, I like that. Just a question: The background: Is this a displacement mapped sphere or is the something else?
(L) [2009/01/07] [davepermen] [Re: Glitch pictures anyone?] Wayback!that's the asteroid destroying all dinosaurs!! i have to save my pleo!!!!
but yes, that's one cool looking sphere (the dinos are cool, too).
(L) [2009/01/07] [squeen] [Re: Glitch pictures anyone?] Wayback!>> Michael77 wrote:...Is this a displacement mapped sphere or is the something else?
Unfortunately no, adding true displacement mapping to the ray tracer is still on my wish list. It's just a highly tessellated "asteroid" I had lying around.
(L) [2009/01/07] [davepermen] [Re: Glitch pictures anyone?] Wayback!nice stuff you have lying around... dinosaurs, asteroids.. i have spheres.. sometimes even cubes.. oh and planes, huuuge planes.. [SMILEY :)]
i found out what to call my source code from now on.. it's a dinosource.. i know.. awesome joke.. [SMILEY :)]
(L) [2009/01/08] [squeen] [Re: Glitch pictures anyone?] Wayback!You can get the killeroo model here:
[LINK http://www.headus.com/au/samples/killeroo/index.html]
The triceratops I found on the web as well (don't remember where). It's featured in the OpenGL Shading Language (orange) book.
(L) [2009/03/11] [squeen] [Re: Glitch pictures anyone?] Wayback!Honestly, I don't know what-the-h*ll went wrong here (byte alignment?)
RGBmarble.jpg
(L) [2009/03/13] [avo] [Re: Glitch pictures anyone?] Wayback!Anyone can guess which scene:
glitch.png
(L) [2009/03/14] [lycium] [Re: Glitch pictures anyone?] Wayback!conference room? (or maybe that "auditorium" model from keller's papers)
ps. i recently made an [LINK http://lyc.deviantart.com/art/affine-glitch-114630175 intentionally glitchy dualscreen wallpaper] if anyone's interested; three cheers again for this awesome thread! [SMILEY =D>]
(L) [2009/03/14] [avo] [Re: Glitch pictures anyone?] Wayback!>> lycium wrote:conference room?
Yes... My renderer decided that it is not impressionistic enough [SMILEY :mrgreen:]
(L) [2009/03/17] [spacerat] [Re: Glitch pictures anyone?] Wayback!Instead of softening the silhouette, anything except the silhouette was removed ...
(L) [2009/03/18] [jay79] [Re: Glitch pictures anyone?] Wayback!lastrender.jpg
I'm working on Irradiance Caching, and here's a bug with determining whether to place sample or not [SMILEY :)]
(L) [2009/04/17] [phat shades] [Re: Glitch pictures anyone?] Wayback!disco disco
(L) [2009/06/20] [toxie] [Re: Glitch pictures anyone?] Wayback!nice one! looks really a bit like painted (with a weird raster on top of it)..
(L) [2009/06/24] [toxie] [Re: Glitch pictures anyone?] Wayback!importance sampling locations visualized + horrible floating point precision issues + branching structure =
[IMG #1 Image]
[IMG #1]:
![[IMG:#0]](images/0ec28f526de8cb10c01d948493ed6ff731d049ccf79416268874d0b627425a92.png)
back
