Tuesday, November 20, 2012

Project5

Part1 Globe:

Implemented

  • Bump mapped terrain
  •     float center = texture2D( u_Bump, v_Texcoord + vec2( -u_time*0.8, 0.0 ));
        float right = texture2D( u_Bump, v_Texcoord+ vec2( -u_time*0.8, 0.0 ) + vec2(1.0/1000.0, 0.0) );
        float top = texture2D( u_Bump, v_Texcoord + vec2( -u_time*0.8, 0.0 )+ vec2(0.0, 1.0/500.0) );

        vec3 perturbedNorm = normalize( vec3(center - right, center - top, 0 .2) );
        vec3 bumpNorm = normalize(eastNorthUpToEyeCoordinates(v_positionMC, normal)* perturbedNorm);

        float BumpDiffuse = max( dot(u_CameraSpaceDirLight, bumpNorm), 0.0 );
  • Rim lighting to simulate atmosphere
    float rimFactor = dot(v_Normal, v_Position) + 1;
  • Nighttime lights on the dark side of the globe
  • Specular mapping
  • Moving clouds(from west to east)
  • Orbiting Moon with texture mapping



Part2 SSAO(Screen Space Ambient Occlusion):
gatherOcclusion:
        float distance =  length( occluder_position - pt_position );
        float overhead = dot( pt_normal, normalize( occluder_position - pt_position ) );
        float planar = 1.0 - abs( dot( pt_normal, occluder_normal ) );
        return planar*max( 0.0, overhead )*( 1.0/(1.0 + distance) ) ;

regularSample:
        vec2 occludePos = texcoord + vec2( i, j )*REGULAR_SAMPLE_STEP;
        vec3 occPos = samplePos(occludePos);
        vec3 occNorm = normalize( sampleNrm(occludePos) );
        accumOcclusion += gatherOcclusion( normalize(normal), position, occNorm, occPos ) ;


Part3 Vertex Pulsing
float displacement = scaleFactor * (0.5 * sin(Position.y * u_frequency * u_time) + 1);
vec3 newPosition = Position + displacement * Normal;

Friday, November 9, 2012

Image Processing/ Vertex Shading

Video


Pat1
Original picture:





Fetures implemeted:
  •  Image negative: vec3(1.0) - rgb

  •  Gaussian blur: GaussianBlurMatrix = 1/16[[1 2 1][2 4 2][1 2 1]]

  •  Grayscale: vec3 w = vec3(0.2125, 0.7154, 0.0721); luminace = dot(rgb, W);

  •  Edge Detection: Sobel-horizontal = [[-1 -2 -1][0 0 0 ][1 2 1]]; Sobel-vertical = [[-1 0 1][-2 0 2 ][-1 0 1]].

  • Toon shading


Optional features:
  • Pixelate: define the pixel size(by using the picture size); Get the new coordinates by6 multiply the pixel size with the pixel index. Pixel indices are calculated by dived the original texture coordinates by pixel size.

  • Brightness: u = (rgb.r + rgb.g + rgb.b)/3;

  • Contrast: (m-a)/(n-a) vs m/n

  • Night Vision: Only multiplied by green color;



Part2:
Sea wave:
float s = sin(pos.x*2PI + time);
        float t = cos(pos.y*2PI + time);
        height = sin(sqrt(s^2 + t^2 ))/sqrt(s^2 +t^2 );



Tuesday, November 6, 2012

Raterizer -- some triangles are missing



Optional feature I chose are:
  • backface culling
  • interactive camera
The rasterization doesn't looks right, some faces are missing:




Back-face culling



Video:

Friday, October 12, 2012

Path Tracer--Got basic features right

  • Anti-aliasing:
Added a random offset when each ray is generated.


   
I don't know why there are white lines on the white sphere.


  • Depth of Field:
 Get the focused plane and shake the camera certain random amount in X and Y direction.

  • Short Demo: 
video

Path tracer--Got the color right

I made mistakes in the Accumulate function for color accumulation.
Here is the corrected one rendered scene I got:

Tuesday, October 9, 2012

Path Tracer -- Always get flat shading


I implemented:
  • calculateBSDF
  • calculateFresnel
  • calculateTransmissionDirection.
  • stream compaction
But the rendered picture I got is always flat shading. And the color doesn't looks right.

The problem is that in remove_if function I accidentally remove all the rays that need to continue tracing. For the colors, that is because I used plus instead of multiply.

Correctted that I got this:

Error : expected an identifier

 error : expected an identifier    C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v4.0\include\thrust\device_ptr.h    286

the problem turned out to be with the inclusion of the thrust headers themselves. Even just including thrust/device_ptr.h threw up errors. For some reason, the thrust headers have to be included before all the other 'C' headers in this project.

Sunday, September 30, 2012

Pushing to Git returning Error Code 403 fatal

Pushing to Git returning Error Code 403 fatal: HTTP request failed

Why:
Github seems only supports ssh way to read&write the repo, although https way also displayed 'Read&Write'. So you need to change your repo config on your PC to ssh way:

Solution:
  1.  Manually edit .git/config file under your repo directory. Find url=entry under section [remote "origin"]. Change all the texts before @ symbol to ssh://git. Save config file and quit. now you could use git push origin master to sync your repo on GitHub
  2.  Another solution is just use shell command:
git remote set-url origin ssh://git@github.com/username/projectName.git

Interactive Camera and specular reflection

  • Interactive Camera
Interactive camera is done. Now the camera position can be controled by keyboard input.
Keys:
      'W': move forward(in Z direction)
      'S': move backward(in Z direction)
      'A': move to left
      'D': move to right
      'U': up
      'I': down

  •  Specular Relection
Specular not correct, still working on it.


Reflection not right:

Without reflection:

Wierd Pattern on the Back Wall

Feautures implemented:
  • Raycasting from a camera into a scenen through a pixel grid
  • Phong lighting for one point light source
  • Diffuse lambertian surfaces
  • Raytraced shadows
  • Cube intersection testing
  • Sphere surface point sampling
I don't know what happened to that wall. The diffuse and  ambient


Finally, I found out why. It is because I used EPSILON in the box-ray intersection set when the ray parrelle with one plane. EPSILON is defined as 0.000000001, which seems too small for a float number camparation. I have changed it to 0.001. 
Now, the picutre looks like this:



Friday, September 28, 2012

Raycast and box-ray intersection


Basic raycast from camera and box-ray intersection finished. At this point, I can get a 2D picture now.
  • Raycast:
Use the caculation steps from Computer Graphic course.
Raycast direction test:

Here is a screen shot of one of the images I got,flat shading: 

  •  Box-ray intersection
I used the slab algorithm. The link is posted on the previous blog.

Thursday, September 27, 2012

CUDA GPU Ray Tracer

This is a course project for CIS 565.
  •  Ray Tracing

Basic Algorithm: For each pixel, shoot a ray into the scene. Check intersections for the ray. If  intersection happens, cast a shadow ray to light source to see if the light source is visible and shade the current puxel accordingly. If the surface is diffuse, the ray will stop there. If it is reflective, shoot a new ray reflected across the normal from teh incident ray. And repeat over until reached the tracing depth or the ray hits a light or diffuse surface.

Folowing is a picture demonstrates how ray tracing algorithm works.(grab from wikipedia)

 Since CUDA does not support recursion, we need to use Iterative Ray-tracing.
Useful link: http://en.wikipedia.org/wiki/Ray_tracing_%28graphics%29
  • Box-ray intersection
I use slab method to check box-ray intersection.
Useful link: http://www.siggraph.org/education/materials/HyperGraph/raytrace/rtinter3.htm
  • Sphere surface point sampling
I used spherical coordinates for sphere surface point sampling.
Useful link: Wolfrat Math World: http://mathworld.wolfram.com/SpherePointPicking.html