Shadow map (3 lights with shadow can be created at the same time) #447
Conversation
| LIGHT_LENGTH = num of lights(3) * num of total vec4 in one light(7) | ||
| LIGHT_TRANSFORM_LENGTH = num of lights(3) * max num of total mat4 of light transform for different kind of light(4) |
| void main() | ||
| { | ||
|
|
||
| } |
There was a problem hiding this comment.
If you don't need to use fragment shader, it is OK to create a program with only vertex shader. Then pipeline should help to optimize. Did you have a try?
|
|
||
| void AddLightViewProjMatrix(cd::Matrix4x4 lightViewProjMatrix) { m_lightViewProjMatrices.push_back(lightViewProjMatrix); } | ||
| const std::vector<cd::Matrix4x4>& GetLightViewProjMatrix() { return m_lightViewProjMatrices; } | ||
| const std::vector<cd::Matrix4x4> GetLightViewProjMatrix() const { return m_lightViewProjMatrices; } |
There was a problem hiding this comment.
| const std::vector<cd::Matrix4x4> GetLightViewProjMatrix() const { return m_lightViewProjMatrices; } | |
| std::vector<cd::Matrix4x4> GetLightViewProjMatrix() const { return m_lightViewProjMatrices; } |
It is good to use const but here is not suitable for the return value to use const. The reason is that people who call this method can decide to accept return value as const or not. So in the API side, it is better not to limit const.
| @@ -13,7 +13,7 @@ class RenderContext; | |||
| namespace | |||
| { | |||
|
|
|||
| constexpr uint16_t MAX_LIGHT_COUNT = 64; | |||
| constexpr uint16_t MAX_LIGHT_COUNT = 3; | |||
There was a problem hiding this comment.
So after merging this PR, the number of math lights will be limited to 3? I am OK with a small number because game usually uses baking to simulate many "lights". It can be a TODO item to expand lighting limits.
| namespace | ||
| { | ||
| // uniform name | ||
| constexpr const char* cameraPos = "u_cameraPos"; |
There was a problem hiding this comment.
It is strange that github pr page shows the first constexpr black. Actually, it should look red as a key word..
| for(float x = -totalOffset; x <= totalOffset; x += stepOffset) | ||
| { | ||
| for(float y = -totalOffset; y <= totalOffset; y += stepOffset) | ||
| { | ||
| for(float z = -totalOffset; z <= totalOffset; z += stepOffset) | ||
| { | ||
| float closestDepth = textureIndex(lightIndex, lightToFrag + vec3(x, y, z)); | ||
| closestDepth *= far_plane; | ||
| shadow += step(closestDepth, currentDepth - bias); | ||
| } | ||
| } | ||
| } |
There was a problem hiding this comment.
| for(float x = -totalOffset; x <= totalOffset; x += stepOffset) | |
| { | |
| for(float y = -totalOffset; y <= totalOffset; y += stepOffset) | |
| { | |
| for(float z = -totalOffset; z <= totalOffset; z += stepOffset) | |
| { | |
| float closestDepth = textureIndex(lightIndex, lightToFrag + vec3(x, y, z)); | |
| closestDepth *= far_plane; | |
| shadow += step(closestDepth, currentDepth - bias); | |
| } | |
| } | |
| } | |
| for(float x = -totalOffset; x <= totalOffset; x += stepOffset) | |
| { | |
| for(float y = -totalOffset; y <= totalOffset; y += stepOffset) | |
| { | |
| for(float z = -totalOffset; z <= totalOffset; z += stepOffset) | |
| { | |
| float closestDepth = textureIndex(lightIndex, lightToFrag + vec3(x, y, z)); | |
| closestDepth *= far_plane; | |
| shadow += step(closestDepth, currentDepth - bias); | |
| } | |
| } | |
| } |
|
Create https://github.com/CatDogEngine/CatDogEngine/issues/456 to track future improvements. |
There's still some bugs in point light shadow.