virtualx-engine/drivers/gles2/shaders/canvas.glsl
lawnjelly 5ed0fd067d Batching - use FINAL_MODULATE_ALIAS in shaders
As part of the improvements to batch more cases, batching can store final_modulate as an attribute in the vertex format rather than sending as a uniform. This allows draw calls with different final_modulate to be batched together.

However custom shader code was reading from only the final_modulate uniform, and not the attribute when it was in use. This was leading to visual errors.

This is tricky to solve, because we cannot use the same name for the attribute in the vertex and fragment shaders, because one is an attribute and one a varying, whereas a uniform is accessible anywhere. To get around this, a macro is used which can translate to the most appropriate variable depending on whether uniform or attribute or varying is required.
2021-03-11 17:07:19 +00:00

710 lines
17 KiB
GLSL

/* clang-format off */
[vertex]
#ifdef USE_GLES_OVER_GL
#define lowp
#define mediump
#define highp
#else
precision highp float;
precision highp int;
#endif
uniform highp mat4 projection_matrix;
/* clang-format on */
#include "stdlib.glsl"
uniform highp mat4 modelview_matrix;
uniform highp mat4 extra_matrix;
attribute highp vec2 vertex; // attrib:0
#ifdef USE_ATTRIB_LIGHT_ANGLE
// shared with tangent, not used in canvas shader
attribute highp float light_angle; // attrib:2
#endif
attribute vec4 color_attrib; // attrib:3
attribute vec2 uv_attrib; // attrib:4
#ifdef USE_ATTRIB_MODULATE
attribute highp vec4 modulate_attrib; // attrib:5
#endif
// Usually, final_modulate is passed as a uniform. However during batching
// If larger fvfs are used, final_modulate is passed as an attribute.
// we need to read from the attribute in custom vertex shader
// rather than the uniform. We do this by specifying final_modulate_alias
// in shaders rather than final_modulate directly.
#ifdef USE_ATTRIB_MODULATE
#define final_modulate_alias modulate_attrib
#else
#define final_modulate_alias final_modulate
#endif
#ifdef USE_ATTRIB_LARGE_VERTEX
// shared with skeleton attributes, not used in batched shader
attribute highp vec2 translate_attrib; // attrib:6
attribute highp vec4 basis_attrib; // attrib:7
#endif
#ifdef USE_SKELETON
attribute highp vec4 bone_indices; // attrib:6
attribute highp vec4 bone_weights; // attrib:7
#endif
#ifdef USE_INSTANCING
attribute highp vec4 instance_xform0; //attrib:8
attribute highp vec4 instance_xform1; //attrib:9
attribute highp vec4 instance_xform2; //attrib:10
attribute highp vec4 instance_color; //attrib:11
#ifdef USE_INSTANCE_CUSTOM
attribute highp vec4 instance_custom_data; //attrib:12
#endif
#endif
#ifdef USE_SKELETON
uniform highp sampler2D skeleton_texture; // texunit:-3
uniform highp ivec2 skeleton_texture_size;
uniform highp mat4 skeleton_transform;
uniform highp mat4 skeleton_transform_inverse;
#endif
varying vec2 uv_interp;
varying vec4 color_interp;
#ifdef USE_ATTRIB_MODULATE
// modulate doesn't need interpolating but we need to send it to the fragment shader
varying vec4 modulate_interp;
#endif
#ifdef MODULATE_USED
uniform vec4 final_modulate;
#endif
uniform highp vec2 color_texpixel_size;
#ifdef USE_TEXTURE_RECT
uniform vec4 dst_rect;
uniform vec4 src_rect;
#endif
uniform highp float time;
#ifdef USE_LIGHTING
// light matrices
uniform highp mat4 light_matrix;
uniform highp mat4 light_matrix_inverse;
uniform highp mat4 light_local_matrix;
uniform highp mat4 shadow_matrix;
uniform highp vec4 light_color;
uniform highp vec4 light_shadow_color;
uniform highp vec2 light_pos;
uniform highp float shadowpixel_size;
uniform highp float shadow_gradient;
uniform highp float light_height;
uniform highp float light_outside_alpha;
uniform highp float shadow_distance_mult;
varying vec4 light_uv_interp;
varying vec2 transformed_light_uv;
varying vec4 local_rot;
#ifdef USE_SHADOWS
varying highp vec2 pos;
#endif
const bool at_light_pass = true;
#else
const bool at_light_pass = false;
#endif
/* clang-format off */
VERTEX_SHADER_GLOBALS
/* clang-format on */
vec2 select(vec2 a, vec2 b, bvec2 c) {
vec2 ret;
ret.x = c.x ? b.x : a.x;
ret.y = c.y ? b.y : a.y;
return ret;
}
void main() {
vec4 color = color_attrib;
vec2 uv;
#ifdef USE_INSTANCING
mat4 extra_matrix_instance = extra_matrix * transpose(mat4(instance_xform0, instance_xform1, instance_xform2, vec4(0.0, 0.0, 0.0, 1.0)));
color *= instance_color;
#ifdef USE_INSTANCE_CUSTOM
vec4 instance_custom = instance_custom_data;
#else
vec4 instance_custom = vec4(0.0);
#endif
#else
mat4 extra_matrix_instance = extra_matrix;
vec4 instance_custom = vec4(0.0);
#endif
#ifdef USE_TEXTURE_RECT
if (dst_rect.z < 0.0) { // Transpose is encoded as negative dst_rect.z
uv = src_rect.xy + abs(src_rect.zw) * vertex.yx;
} else {
uv = src_rect.xy + abs(src_rect.zw) * vertex;
}
vec4 outvec = vec4(0.0, 0.0, 0.0, 1.0);
// This is what is done in the GLES 3 bindings and should
// take care of flipped rects.
//
// But it doesn't.
// I don't know why, will need to investigate further.
outvec.xy = dst_rect.xy + abs(dst_rect.zw) * select(vertex, vec2(1.0, 1.0) - vertex, lessThan(src_rect.zw, vec2(0.0, 0.0)));
// outvec.xy = dst_rect.xy + abs(dst_rect.zw) * vertex;
#else
vec4 outvec = vec4(vertex.xy, 0.0, 1.0);
uv = uv_attrib;
#endif
float point_size = 1.0;
{
vec2 src_vtx = outvec.xy;
/* clang-format off */
VERTEX_SHADER_CODE
/* clang-format on */
}
gl_PointSize = point_size;
#ifdef USE_ATTRIB_MODULATE
// modulate doesn't need interpolating but we need to send it to the fragment shader
modulate_interp = modulate_attrib;
#endif
#ifdef USE_ATTRIB_LARGE_VERTEX
// transform is in attributes
vec2 temp;
temp = outvec.xy;
temp.x = (outvec.x * basis_attrib.x) + (outvec.y * basis_attrib.z);
temp.y = (outvec.x * basis_attrib.y) + (outvec.y * basis_attrib.w);
temp += translate_attrib;
outvec.xy = temp;
#else
// transform is in uniforms
#if !defined(SKIP_TRANSFORM_USED)
outvec = extra_matrix_instance * outvec;
outvec = modelview_matrix * outvec;
#endif
#endif // not large integer
color_interp = color;
#ifdef USE_PIXEL_SNAP
outvec.xy = floor(outvec + 0.5).xy;
// precision issue on some hardware creates artifacts within texture
// offset uv by a small amount to avoid
uv += 1e-5;
#endif
#ifdef USE_SKELETON
// look up transform from the "pose texture"
if (bone_weights != vec4(0.0)) {
highp mat4 bone_transform = mat4(0.0);
for (int i = 0; i < 4; i++) {
ivec2 tex_ofs = ivec2(int(bone_indices[i]) * 2, 0);
highp mat4 b = mat4(
texel2DFetch(skeleton_texture, skeleton_texture_size, tex_ofs + ivec2(0, 0)),
texel2DFetch(skeleton_texture, skeleton_texture_size, tex_ofs + ivec2(1, 0)),
vec4(0.0, 0.0, 1.0, 0.0),
vec4(0.0, 0.0, 0.0, 1.0));
bone_transform += b * bone_weights[i];
}
mat4 bone_matrix = skeleton_transform * transpose(bone_transform) * skeleton_transform_inverse;
outvec = bone_matrix * outvec;
}
#endif
uv_interp = uv;
gl_Position = projection_matrix * outvec;
#ifdef USE_LIGHTING
light_uv_interp.xy = (light_matrix * outvec).xy;
light_uv_interp.zw = (light_local_matrix * outvec).xy;
transformed_light_uv = (mat3(light_matrix_inverse) * vec3(light_uv_interp.zw, 0.0)).xy; //for normal mapping
#ifdef USE_SHADOWS
pos = outvec.xy;
#endif
#ifdef USE_ATTRIB_LIGHT_ANGLE
// we add a fixed offset because we are using the sign later,
// and don't want floating point error around 0.0
float la = abs(light_angle) - 1.0;
// vector light angle
vec4 vla;
vla.xy = vec2(cos(la), sin(la));
vla.zw = vec2(-vla.y, vla.x);
// vertical flip encoded in the sign
vla.zw *= sign(light_angle);
// apply the transform matrix.
// The rotate will be encoded in the transform matrix for single rects,
// and just the flips in the light angle.
// For batching we will encode the rotation and the flips
// in the light angle, and can use the same shader.
local_rot.xy = normalize((modelview_matrix * (extra_matrix_instance * vec4(vla.xy, 0.0, 0.0))).xy);
local_rot.zw = normalize((modelview_matrix * (extra_matrix_instance * vec4(vla.zw, 0.0, 0.0))).xy);
#else
local_rot.xy = normalize((modelview_matrix * (extra_matrix_instance * vec4(1.0, 0.0, 0.0, 0.0))).xy);
local_rot.zw = normalize((modelview_matrix * (extra_matrix_instance * vec4(0.0, 1.0, 0.0, 0.0))).xy);
#ifdef USE_TEXTURE_RECT
local_rot.xy *= sign(src_rect.z);
local_rot.zw *= sign(src_rect.w);
#endif
#endif // not using light angle
#endif
}
/* clang-format off */
[fragment]
// texture2DLodEXT and textureCubeLodEXT are fragment shader specific.
// Do not copy these defines in the vertex section.
#ifndef USE_GLES_OVER_GL
#ifdef GL_EXT_shader_texture_lod
#extension GL_EXT_shader_texture_lod : enable
#define texture2DLod(img, coord, lod) texture2DLodEXT(img, coord, lod)
#define textureCubeLod(img, coord, lod) textureCubeLodEXT(img, coord, lod)
#endif
#endif // !USE_GLES_OVER_GL
#ifdef GL_ARB_shader_texture_lod
#extension GL_ARB_shader_texture_lod : enable
#endif
#if !defined(GL_EXT_shader_texture_lod) && !defined(GL_ARB_shader_texture_lod)
#define texture2DLod(img, coord, lod) texture2D(img, coord, lod)
#define textureCubeLod(img, coord, lod) textureCube(img, coord, lod)
#endif
#ifdef USE_GLES_OVER_GL
#define lowp
#define mediump
#define highp
#else
#if defined(USE_HIGHP_PRECISION)
precision highp float;
precision highp int;
#else
precision mediump float;
precision mediump int;
#endif
#endif
#include "stdlib.glsl"
uniform sampler2D color_texture; // texunit:-1
/* clang-format on */
uniform highp vec2 color_texpixel_size;
uniform mediump sampler2D normal_texture; // texunit:-2
varying mediump vec2 uv_interp;
varying mediump vec4 color_interp;
#ifdef USE_ATTRIB_MODULATE
varying mediump vec4 modulate_interp;
#endif
uniform highp float time;
uniform vec4 final_modulate;
#ifdef SCREEN_TEXTURE_USED
uniform sampler2D screen_texture; // texunit:-4
#endif
#ifdef SCREEN_UV_USED
uniform vec2 screen_pixel_size;
#endif
#ifdef USE_LIGHTING
uniform highp mat4 light_matrix;
uniform highp mat4 light_local_matrix;
uniform highp mat4 shadow_matrix;
uniform highp vec4 light_color;
uniform highp vec4 light_shadow_color;
uniform highp vec2 light_pos;
uniform highp float shadowpixel_size;
uniform highp float shadow_gradient;
uniform highp float light_height;
uniform highp float light_outside_alpha;
uniform highp float shadow_distance_mult;
uniform lowp sampler2D light_texture; // texunit:-6
varying vec4 light_uv_interp;
varying vec2 transformed_light_uv;
varying vec4 local_rot;
#ifdef USE_SHADOWS
uniform highp sampler2D shadow_texture; // texunit:-5
varying highp vec2 pos;
#endif
const bool at_light_pass = true;
#else
const bool at_light_pass = false;
#endif
uniform bool use_default_normal;
/* clang-format off */
FRAGMENT_SHADER_GLOBALS
/* clang-format on */
void light_compute(
inout vec4 light,
inout vec2 light_vec,
inout float light_height,
inout vec4 light_color,
vec2 light_uv,
inout vec4 shadow_color,
inout vec2 shadow_vec,
vec3 normal,
vec2 uv,
#if defined(SCREEN_UV_USED)
vec2 screen_uv,
#endif
vec4 color) {
#if defined(USE_LIGHT_SHADER_CODE)
/* clang-format off */
LIGHT_SHADER_CODE
/* clang-format on */
#endif
}
void main() {
vec4 color = color_interp;
vec2 uv = uv_interp;
#ifdef USE_FORCE_REPEAT
//needs to use this to workaround GLES2/WebGL1 forcing tiling that textures that don't support it
uv = mod(uv, vec2(1.0, 1.0));
#endif
#if !defined(COLOR_USED)
//default behavior, texture by color
color *= texture2D(color_texture, uv);
#endif
#ifdef SCREEN_UV_USED
vec2 screen_uv = gl_FragCoord.xy * screen_pixel_size;
#endif
vec3 normal;
#if defined(NORMAL_USED)
bool normal_used = true;
#else
bool normal_used = false;
#endif
if (use_default_normal) {
normal.xy = texture2D(normal_texture, uv).xy * 2.0 - 1.0;
normal.z = sqrt(max(0.0, 1.0 - dot(normal.xy, normal.xy)));
normal_used = true;
} else {
normal = vec3(0.0, 0.0, 1.0);
}
{
float normal_depth = 1.0;
#if defined(NORMALMAP_USED)
vec3 normal_map = vec3(0.0, 0.0, 1.0);
normal_used = true;
#endif
// If larger fvfs are used, final_modulate is passed as an attribute.
// we need to read from this in custom fragment shaders or applying in the post step,
// rather than using final_modulate directly.
#if defined(final_modulate_alias)
#undef final_modulate_alias
#endif
#ifdef USE_ATTRIB_MODULATE
#define final_modulate_alias modulate_interp
#else
#define final_modulate_alias final_modulate
#endif
/* clang-format off */
FRAGMENT_SHADER_CODE
/* clang-format on */
#if defined(NORMALMAP_USED)
normal = mix(vec3(0.0, 0.0, 1.0), normal_map * vec3(2.0, -2.0, 1.0) - vec3(1.0, -1.0, 0.0), normal_depth);
#endif
}
#if !defined(MODULATE_USED)
color *= final_modulate_alias;
#endif
#ifdef USE_LIGHTING
vec2 light_vec = transformed_light_uv;
vec2 shadow_vec = transformed_light_uv;
if (normal_used) {
normal.xy = mat2(local_rot.xy, local_rot.zw) * normal.xy;
}
float att = 1.0;
vec2 light_uv = light_uv_interp.xy;
vec4 light = texture2D(light_texture, light_uv);
if (any(lessThan(light_uv_interp.xy, vec2(0.0, 0.0))) || any(greaterThanEqual(light_uv_interp.xy, vec2(1.0, 1.0)))) {
color.a *= light_outside_alpha; //invisible
} else {
float real_light_height = light_height;
vec4 real_light_color = light_color;
vec4 real_light_shadow_color = light_shadow_color;
#if defined(USE_LIGHT_SHADER_CODE)
//light is written by the light shader
light_compute(
light,
light_vec,
real_light_height,
real_light_color,
light_uv,
real_light_shadow_color,
shadow_vec,
normal,
uv,
#if defined(SCREEN_UV_USED)
screen_uv,
#endif
color);
#endif
light *= real_light_color;
if (normal_used) {
vec3 light_normal = normalize(vec3(light_vec, -real_light_height));
light *= max(dot(-light_normal, normal), 0.0);
}
color *= light;
#ifdef USE_SHADOWS
#ifdef SHADOW_VEC_USED
mat3 inverse_light_matrix = mat3(light_matrix);
inverse_light_matrix[0] = normalize(inverse_light_matrix[0]);
inverse_light_matrix[1] = normalize(inverse_light_matrix[1]);
inverse_light_matrix[2] = normalize(inverse_light_matrix[2]);
shadow_vec = (inverse_light_matrix * vec3(shadow_vec, 0.0)).xy;
#else
shadow_vec = light_uv_interp.zw;
#endif
float angle_to_light = -atan(shadow_vec.x, shadow_vec.y);
float PI = 3.14159265358979323846264;
/*int i = int(mod(floor((angle_to_light+7.0*PI/6.0)/(4.0*PI/6.0))+1.0, 3.0)); // +1 pq os indices estao em ordem 2,0,1 nos arrays
float ang*/
float su, sz;
float abs_angle = abs(angle_to_light);
vec2 point;
float sh;
if (abs_angle < 45.0 * PI / 180.0) {
point = shadow_vec;
sh = 0.0 + (1.0 / 8.0);
} else if (abs_angle > 135.0 * PI / 180.0) {
point = -shadow_vec;
sh = 0.5 + (1.0 / 8.0);
} else if (angle_to_light > 0.0) {
point = vec2(shadow_vec.y, -shadow_vec.x);
sh = 0.25 + (1.0 / 8.0);
} else {
point = vec2(-shadow_vec.y, shadow_vec.x);
sh = 0.75 + (1.0 / 8.0);
}
highp vec4 s = shadow_matrix * vec4(point, 0.0, 1.0);
s.xyz /= s.w;
su = s.x * 0.5 + 0.5;
sz = s.z * 0.5 + 0.5;
//sz=lightlength(light_vec);
highp float shadow_attenuation = 0.0;
#ifdef USE_RGBA_SHADOWS
#define SHADOW_DEPTH(m_tex, m_uv) dot(texture2D((m_tex), (m_uv)), vec4(1.0 / (255.0 * 255.0 * 255.0), 1.0 / (255.0 * 255.0), 1.0 / 255.0, 1.0))
#else
#define SHADOW_DEPTH(m_tex, m_uv) (texture2D((m_tex), (m_uv)).r)
#endif
#ifdef SHADOW_USE_GRADIENT
/* clang-format off */
/* GLSL es 100 doesn't support line continuation characters(backslashes) */
#define SHADOW_TEST(m_ofs) { highp float sd = SHADOW_DEPTH(shadow_texture, vec2(m_ofs, sh)); shadow_attenuation += 1.0 - smoothstep(sd, sd + shadow_gradient, sz); }
#else
#define SHADOW_TEST(m_ofs) { highp float sd = SHADOW_DEPTH(shadow_texture, vec2(m_ofs, sh)); shadow_attenuation += step(sz, sd); }
/* clang-format on */
#endif
#ifdef SHADOW_FILTER_NEAREST
SHADOW_TEST(su);
#endif
#ifdef SHADOW_FILTER_PCF3
SHADOW_TEST(su + shadowpixel_size);
SHADOW_TEST(su);
SHADOW_TEST(su - shadowpixel_size);
shadow_attenuation /= 3.0;
#endif
#ifdef SHADOW_FILTER_PCF5
SHADOW_TEST(su + shadowpixel_size * 2.0);
SHADOW_TEST(su + shadowpixel_size);
SHADOW_TEST(su);
SHADOW_TEST(su - shadowpixel_size);
SHADOW_TEST(su - shadowpixel_size * 2.0);
shadow_attenuation /= 5.0;
#endif
#ifdef SHADOW_FILTER_PCF7
SHADOW_TEST(su + shadowpixel_size * 3.0);
SHADOW_TEST(su + shadowpixel_size * 2.0);
SHADOW_TEST(su + shadowpixel_size);
SHADOW_TEST(su);
SHADOW_TEST(su - shadowpixel_size);
SHADOW_TEST(su - shadowpixel_size * 2.0);
SHADOW_TEST(su - shadowpixel_size * 3.0);
shadow_attenuation /= 7.0;
#endif
#ifdef SHADOW_FILTER_PCF9
SHADOW_TEST(su + shadowpixel_size * 4.0);
SHADOW_TEST(su + shadowpixel_size * 3.0);
SHADOW_TEST(su + shadowpixel_size * 2.0);
SHADOW_TEST(su + shadowpixel_size);
SHADOW_TEST(su);
SHADOW_TEST(su - shadowpixel_size);
SHADOW_TEST(su - shadowpixel_size * 2.0);
SHADOW_TEST(su - shadowpixel_size * 3.0);
SHADOW_TEST(su - shadowpixel_size * 4.0);
shadow_attenuation /= 9.0;
#endif
#ifdef SHADOW_FILTER_PCF13
SHADOW_TEST(su + shadowpixel_size * 6.0);
SHADOW_TEST(su + shadowpixel_size * 5.0);
SHADOW_TEST(su + shadowpixel_size * 4.0);
SHADOW_TEST(su + shadowpixel_size * 3.0);
SHADOW_TEST(su + shadowpixel_size * 2.0);
SHADOW_TEST(su + shadowpixel_size);
SHADOW_TEST(su);
SHADOW_TEST(su - shadowpixel_size);
SHADOW_TEST(su - shadowpixel_size * 2.0);
SHADOW_TEST(su - shadowpixel_size * 3.0);
SHADOW_TEST(su - shadowpixel_size * 4.0);
SHADOW_TEST(su - shadowpixel_size * 5.0);
SHADOW_TEST(su - shadowpixel_size * 6.0);
shadow_attenuation /= 13.0;
#endif
//color *= shadow_attenuation;
color = mix(real_light_shadow_color, color, shadow_attenuation);
//use shadows
#endif
}
//use lighting
#endif
gl_FragColor = color;
}