Shadow Map Produces Incorrect Results - c++

I'm attempting to implement shadow mapping into my deferred rendering pipeline, but I'm running into a few issues actually generating the shadow map, then shadowing the pixels – pixels that I believe should be shadowed simply aren't.
I have a single directional light, which is the 'sun' in my engine. I have deferred rendering set up for lighting, which works properly thus far. I render the scene again into a depth-only FBO for the shadow map, using the following code to generate the view matrix:
glm::vec3 position = r->getCamera()->getCameraPosition(); // position of level camera
glm::vec3 lightDir = this->sun->getDirection(); // sun direction vector
glm::mat4 depthProjectionMatrix = glm::ortho<float>(-10,10,-10,10,-10,20); // ortho projection
glm::mat4 depthViewMatrix = glm::lookAt(position + (lightDir * 20.f / 2.f), -lightDir, glm::vec3(0,1,0));
glm::mat4 lightSpaceMatrix = depthProjectionMatrix * depthViewMatrix;
Then, in my lighting shader, I use the following code to determine whether a pixel is in shadow or not:
// lightSpaceMatrix is the same as above, FragWorldPos is world position of the texekl
vec4 FragPosLightSpace = lightSpaceMatrix * vec4(FragWorldPos, 1.0f);
// multiply non-ambient light values by ShadowCalculation(FragPosLightSpace)
// ... do more stuff ...
float ShadowCalculation(vec4 fragPosLightSpace) {
// perform perspective divide
vec3 projCoords = fragPosLightSpace.xyz / fragPosLightSpace.w;
// vec3 projCoords = fragPosLightSpace.xyz;
// Transform to [0,1] range
projCoords = projCoords * 0.5 + 0.5;
// Get closest depth value from light's perspective (using [0,1] range fragPosLight as coords)
float closestDepth = texture(gSunShadowMap, projCoords.xy).r;
// Get depth of current fragment from light's perspective
float currentDepth = projCoords.z;
// Check whether current frag pos is in shadow
float bias = 0.005;
float shadow = (currentDepth - bias) > closestDepth ? 1.0 : 0.0;
// Ensure that Z value is no larger than 1
if(projCoords.z > 1.0) {
shadow = 0.0;
}
return shadow;
}
However, that doesn't really get me what I'm after. Here's a screenshot of the output after shadowing, as well as the shadow map half-assedly converted to an image in Photoshop:
Render output
Shadow Map
Since the directional light is the only light in my shader, it seems that the shadow map is being rendered pretty close to correctly, since the perspective/direction roughly match. However, what I don't understand is why none of the teapots actually end up casting a shadow on the others.
I'd appreciate any pointers on what I might be doing wrong. I think that my issue lies either in the calculation of that light space matrix (I'm not sure how to properly calculate that, given a moving camera, such that the stuff that's in view will be updated,) or in the way I determine whether the texel the deferred renderer is shading is in shadow or not. (FWIW, I determine the world position from the depth buffer, but I've proven that this calculation is working correctly.)
Thanks for any help.

Debugging shadow problems can be tricky. Lets start with a few points:
If you look at your render closely, you will actually see a shadow on one of the pots in the top left corner.
Try rotating your sun, this usually helps to see if there are any problems with the light transform matrix. From your output, it seems the sun is very horizontal and might not cast shadows on this setup. (another angle might show more shadows)
It appears as though you are calculating the matrix correctly, but try shrinking your maximum depth in glm::ortho(-10,10,-10,10,-10,20) to tightly fit your scene. If the depth is too large, you will lose precision and shadow will have artifacts.
To visualize where the problem is coming from further, try outputing the result from your shadow map lookup from here:
closestDepth = texture(gSunShadowMap, projCoords.xy).r
If the shadow map is being projected correctly, then you know you have a problem in your depth comparisons. Hope this helps!

Related

OpenGL Normal Mapping

I'm trying to implement Normal Mapping, using a simple cube that i created. I followed this tutorial https://learnopengl.com/Advanced-Lighting/Normal-Mapping but i can't really get how normal mapping should be done when drawing 3d objects, since the tutorial is using a 2d object.
In particular, my cube seems almost correctly lighted but there's something i think it's not working how it should be. I'm using a geometry shader that will output green vector normals and red vector tangents, to help me out. Here i post three screenshot of my work.
Directly lighted
Side lighted
Here i actually tried calculating my normals and tangents in a different way. (quite wrong)
In the first image i calculate my cube normals and tangents one face at a time. This seems to work for the face, but if i rotate my cube i think the lighting on the adiacent face is wrong. As you can see in the second image, it's not totally absent.
In the third image, i tried summing all normals and tangents per vertex, as i think it should be done, but the result seems quite wrong, since there is too little lighting.
In the end, my question is how i should calculate normals and tangents.
Should i consider per face calculations or sum vectors per vertex across all relative faces, or else?
EDIT --
I'm passing normal and tangent to the vertex shader and setting up my TBN matrix. But as you can see in the first image, drawing face by face my cube, it seems that those faces adjacent to the one i'm looking directly (that is well lighted) are not correctly lighted and i don't know why. I thought that i wasn't correctly calculating my 'per face' normal and tangent. I thought that calculating some normal and tangent that takes count of the object in general, could be the right way.
If it's right to calculate normal and tangent as visible in the second image (green normal, red tangent) to set up the TBN matrix, why does the right face seems not well lighted?
EDIT 2 --
Vertex shader:
void main(){
texture_coordinates = textcoord;
fragment_position = vec3(model * vec4(position,1.0));
mat3 normalMatrix = transpose(inverse(mat3(model)));
vec3 T = normalize(normalMatrix * tangent);
vec3 N = normalize(normalMatrix * normal);
T = normalize(T - dot(T, N) * N);
vec3 B = cross(N, T);
mat3 TBN = transpose(mat3(T,B,N));
view_position = TBN * viewPos; // camera position
light_position = TBN * lightPos; // light position
fragment_position = TBN * fragment_position;
gl_Position = projection * view * model * vec4(position,1.0);
}
In the VS i set up my TBN matrix and i transform all light, fragment and view vectors to tangent space; doing so i won't have to do any other calculation in the fragment shader.
Fragment shader:
void main() {
vec3 Normal = texture(TextSamplerNormals,texture_coordinates).rgb; // extract normal
Normal = normalize(Normal * 2.0 - 1.0); // correct range
material_color = texture2D(TextSampler,texture_coordinates.st); // diffuse map
vec3 I_amb = AmbientLight.color * AmbientLight.intensity;
vec3 lightDir = normalize(light_position - fragment_position);
vec3 I_dif = vec3(0,0,0);
float DiffusiveFactor = max(dot(lightDir,Normal),0.0);
vec3 I_spe = vec3(0,0,0);
float SpecularFactor = 0.0;
if (DiffusiveFactor>0.0) {
I_dif = DiffusiveLight.color * DiffusiveLight.intensity * DiffusiveFactor;
vec3 vertex_to_eye = normalize(view_position - fragment_position);
vec3 light_reflect = reflect(-lightDir,Normal);
light_reflect = normalize(light_reflect);
SpecularFactor = pow(max(dot(vertex_to_eye,light_reflect),0.0),SpecularLight.power);
if (SpecularFactor>0.0) {
I_spe = DiffusiveLight.color * SpecularLight.intensity * SpecularFactor;
}
}
color = vec4(material_color.rgb * (I_amb + I_dif + I_spe),material_color.a);
}
Handling discontinuity vs continuity
You are thinking about this the wrong way.
Depending on the use case your normal map may be continous or discontinous. For example in your cube, imagine if each face had a different surface type, then the normals would be different depending on which face you are currently in.
Which normal you use is determined by the texture itself and not by any blending in the fragment.
The actual algorithm is
Load rgb values of normal
Convert to -1 to 1 range
Rotate by the model matrix
Use new value in shading calculations
If you want continous normals, then you need to make sure that the charts in the texture space that you use obey that the limits of the texture coordinates agree.
Mathematically that means that if U and V are regions of R^2 that map to the normal field N of your Shape then if the function of the mapping is f it should be that:
If lim S(x_1, x_2) = lim S(y_1, y_2) where {x1,x2} \subset U and {y_1, y_2} \subset V then lim f(x_1, x_2) = lim f(y_1, y_2).
In plain English, if the cooridnates in your chart map to positions that are close in the shape, then the normals they map to should also be close in the normal space.
TL;DR do not belnd in the fragment. This is something that should be done by the normal map itself when its baked, not'by you when rendering.
Handling the tangent space
You have 2 options. Option n1, you pass the tangent T and the normal N to the shader. In which case the binormal B is T X N and the basis {T, N, B} gives you the true space where normals need to be expressed.
Assume that in tangent space, x is side, y is forward z is up. Your transformed normal becomes (xB, yT, zN).
If you do not pass the tangent, you must first create a random vector that is orthogonal to the normal, then use this as the tangent.
(Note N is the model normal, where (x,y,z) is the normal map normal)

Omnidirectional Lighting in OpenGL/GLSL 4.1

I've gotten shadows working properly for my Directional Lights, but I'm a little stumped when it comes to Point Lights. My idea is to use a cube map to render the depth from all six sides surrounding the light. So far, that's all working and good. I have verified this step by rendering each face of my cube to a 2D image, and it appears to be correct.
Now I'm trying to get the shadows to show up in the world. To do so, I am using GLSL's samplerCubeShadow data type. With it, I do:
vec3 lightToFrag = light.position - fragPos
float lenLightToFrag = length(lightToFrag)
vec3 normLightToFrag = normalize(lightToFrag)
float shadow = texture(depthTexture, vec4(normLightToFrag, lightToFrag))
I've tried a number of configurations, and this always renders my scene in black. Any ideas? My fragPos is just the model matrix times the vertex position. Should I be applying the light's model-view matrix to it? Or, similarly, should I be applying the world's model-view matrix to the light? Any feedback is really appreciated!
Assuming you are storing depth values in cubemap;
AFAIK cubemap is an AABB in world space, so you need to do calculations in world space. In your case light.position and fragPos must be in world space, or provide alternative variables/members if you use these names in view space in somewhere else e.g. per-fragment light calculations
Also you need to convert lightToFrag to depth value before pass to texture.
This answer shows how to convert lightToFrag to depth value: Omnidirectional shadow mapping with depth cubemap
Here my implementation (I removed #ifdef SHAD_CUBE because others use same name):
uniform samplerCubeShadow uShadMap;
uniform vec2 uFarNear;
float depthValue(const in vec3 v) {
vec3 absv = abs(v);
float z = max(absv.x, max(absv.y, absv.z));
return uFarNear.x + uFarNear.y / z;
}
float shadowCoef() {
vec3 L;
float d;
L = vPosWS - light.position_ws;
d = depthValue(L);
return texture(uShadMap, vec4(L, d));
}
This may require uniform model matrix if you only have ModelViewProjection (MVP)
Here how to calculate uNearFar at client side:
float n, f, nfsub, nf[2];
n = sm->near;
f = sm->far;
nfsub = f - n;
nf[0] = (f + n) / nfsub * 0.5f + 0.5f;
nf[1] =-(f * n) / nfsub;
glUniform2f(gkUniformLoc(prog, "uFarNear"), nf[0], nf[1]);
this is just optimization but you don't have to use this and follow the link which I mentioned before.
You may need bias value, related answer uses bias but I'm not sure how to apply it to cubemap correctly. I'm not sure d -+ 0.0001 is correct way or not.
If you want to store world distances in cubemap then this tutorial seems god one: https://learnopengl.com/Advanced-Lighting/Shadows/Point-Shadows

OpenGL shadow mapping with deferred rendering, position transformation

I am using deferred rendering where i store the eye space position in a texture accordingly:
vertex:
gl_Position = vec4(vertex_position, 1.0);
geometry:
vertexOut.position = vec3(viewMatrix * modelMatrix * gl_in[i].gl_Position);
fragment:
positionOut = vec3(vertexIn.position);
Now, in the second pass (lighting pass) I am trying to sample my shadow map, using UV coordinates calculated from this vec4
vec4 lightSpacePos = lightProjectionMatrix * lightViewMatrix * lightModelMatrix * vec4(position, 1.0);
The position used is the same position stored and sampled from the position texture.
Do I need to transfrom the position with the inverse camera view matrix before doing this calculation? To bring it back to world space or how should I proceed?
Typically shadow mapping is done by comparing the window-space Z coordinate (this is what a depth texture stores) of your current fragment vs. your light. This must be done using a common reference orientation, so that involves re-projecting your current fragment's position from the perspective of your light.
You have the view-space position right now, which is relative to your current camera and not particularly useful. To do this effectively you want world-space position. You can get that if you transform the view-space position by the inverse view matrix.
Given world-space position, transform into clip-space from light's perspective:
// This will be in clip-space
vec4 lightSpacePos = lightProjectionMatrix * lightViewMatrix * vec4 (worldPos);
// Transform it into NDC-space by dividing by w
lightSpacePos /= lightSpacePos.w;
// Range is now [-1.0, 1.0], but you need [0.0, 1.0]
lightSpacePos = lightSpacePos * vec4 (0.5) + vec4 (0.5);
Assuming default depth range, lightSpacePos is now ready for use. xy contains the texture coordinates to sample from your shadow map and z contains the depth to use for comparison.
For a more thorough explanation, see the following answer.
Incidentally, you will want to eliminate your position texture from your G-Buffer to achieve reasonable performance. It is very easy to reconstruct world- or view-space position given only the depth and the projection and view matrices and the arithmetic involved is much quicker than an extra texture fetch. Storing an additional texture with adequate precision to represent position in 3D space will burn through tons of memory bandwidth each frame and is completely unnecessary.
This article from the OpenGL Wiki explains how to do this. You can take it one step farther and work back to world-space, which is more desirable than view-space. You may need to tweak your depth buffer a little bit to get adequate precision, but it will still be quicker than storing position separately.

Variance Shadow Map Depth Issue

I have been trying to get variance shadow mapping to work in my webgl application, but I seem to be having an issue that I could use some help with. In short, my shadows seem to vary over a much smaller distance than the examples I have seen out there. I.e. the shadow range is from 0 to 500 units, but the shadow is black 5 units away and almost non-existent 10 units away. The examples I am following are based on these two links:
VSM from Florian Boesch
VSM from Fabian Sanglard
In both of those examples, the authors are using spot light perspective projection to map the variance values to a floating point texture. In my engine, I have so far tried to use the same logic except I am using a directional light and orthographic projection. I tried both techniques and the result seems to always be the same for me. I'm not sure if its the because of me using an orthographic matrix to do projection - I suspect it might be. Here is a picture of the problem:
Notice how the box is only a few units away from the circle but the shadow is much darker even though the camera shadow is 0.1 to 500 units.
In the light shadow pass my code looks like this:
// viewMatrix is a uniform of the inverse world matrix of the camera
// vWorldPosition is the varying vec4 of the vertex position x world matrix
vec3 lightPos = (viewMatrix * vWorldPosition).xyz;
depth = clamp(length(lightPos) / 40.0, 0.0, 1.0);
float moment1 = depth;
float moment2 = depth * depth;
// Adjusting moments (this is sort of bias per pixel) using partial derivative
float dx = dFdx(depth);
float dy = dFdy(depth);
moment2 += pow(depth, 2.0) + 0.25 * (dx * dx + dy * dy) ;
gl_FragColor = vec4(moment1, moment2, 0.0, 1.0);
Then in my shadow pass:
// lightViewMatrix is the light camera's inverse world matrix
// vertWorldPosition is the attribute position x world matrix
vec3 lightViewPos = lightViewMatrix * vertWorldPosition;
float lightDepth2 = clamp(length(lightViewPos) / 40.0, 0.0, 1.0);
float illuminated = vsm( shadowMap[i], shadowCoord.xy, lightDepth2, shadowBias[i] );
shadowColor = shadowColor * illuminated
Firstly, should I be doing anything differently with Orthographic projection (Its probably not this, but I don't know what it might be as it happens using both techniques above :( )? If not, what might I be able to do to get a more even spread of the shadow?
Many thanks

reconstructed world position from depth is wrong

I'm trying to implement deferred shading/lighting. In order to reduce the number/size of the buffers I use I wanted to use the depth texture to reconstruct world position later on.
I do this by multiplying the pixel's coordinates with the inverse of the projection matrix and the inverse of the camera matrix. This sort of works, but the position is a bit off. Here's the absolute difference with a sampled world position texture:
For reference, this is the code I use in the second pass fragment shader:
vec2 screenPosition_texture = vec2((gl_FragCoord.x)/WIDTH, (gl_FragCoord.y)/HEIGHT);
float pixelDepth = texture2D(depth, screenPosition_texture).x;
vec4 worldPosition = pMatInverse*vec4(VertexIn.position, pixelDepth, 1.0);
worldPosition = vec4(worldPosition.xyz/worldPosition.w, 1.0);
//worldPosition /= 1.85;
worldPosition = cMatInverse*worldPosition_byDepth;
If I uncomment worldPosition /= 1.85, the position is reconstructed a lot better (on my geometry/range of depth values). I just got this value by messing around after comparing my output with what it should be (stored in a third texture).
I'm using 0.1 near, 100.0 far and my geometries are up to about 15 away.
I know there may be precision errors, but this seems a bit too big of an error too close to the camera.
Did I miss anything here?
As mentioned in a comment:
I didn't convert the depth value from NDC space to clip space.
I should have added this line:
pixelDepth = pixelDepth * 2.0 - 1.0;