diff options
author | Clément Foucault <foucault.clem@gmail.com> | 2022-02-17 19:03:23 +0300 |
---|---|---|
committer | Clément Foucault <foucault.clem@gmail.com> | 2022-02-19 00:30:45 +0300 |
commit | 16d565836588de15b739ae5965d12d7794b87f3e (patch) | |
tree | 935a55de98502bb17fe765bf4db338451333ed22 /intern/opencolorio/gpu_shader_display_transform.glsl | |
parent | e81dda4f3865827f3b3039caa57d9e97e2019781 (diff) |
OCIO: Port shader creation logic to use GPUShaderCreateInfo
This commit should suffice to make the shader API agnostic now (given that
all users of it use the GPU API).
This makes the shaders not trigger a false positive error anymore since
the binding slots are now garanteed by the backend and not changed at
after compilation.
This also bundles all uniforms into UBOs. Making them extendable without
limitations of push constants. The generated uniforms from OCIO are not
densely packed in the UBO to avoid complexity. Another approach would be to
use GPU_uniformbuf_create_from_list but this requires converting uniforms
to GPUInputs which is too complex for what it is.
Reviewed by: brecht, jbakker
Differential Revision: https://developer.blender.org/D14123
Diffstat (limited to 'intern/opencolorio/gpu_shader_display_transform.glsl')
-rw-r--r-- | intern/opencolorio/gpu_shader_display_transform.glsl | 210 |
1 files changed, 0 insertions, 210 deletions
diff --git a/intern/opencolorio/gpu_shader_display_transform.glsl b/intern/opencolorio/gpu_shader_display_transform.glsl deleted file mode 100644 index f5a7a7bf45d..00000000000 --- a/intern/opencolorio/gpu_shader_display_transform.glsl +++ /dev/null @@ -1,210 +0,0 @@ -/* Blender OpenColorIO implementation */ - -uniform sampler2D image_texture; -uniform sampler2D overlay_texture; - -uniform float dither; -uniform float scale; -uniform float exponent; -uniform bool predivide; -uniform bool overlay; - -#ifdef USE_CURVE_MAPPING -uniform sampler1D curve_mapping_texture; - -layout(std140) uniform OCIO_GPUCurveMappingParameters -{ - /* Curve mapping parameters - * - * See documentation for OCIO_CurveMappingSettings to get fields descriptions. - * (this ones pretty much copies stuff from C structure.) - */ - vec4 curve_mapping_mintable; - vec4 curve_mapping_range; - vec4 curve_mapping_ext_in_x; - vec4 curve_mapping_ext_in_y; - vec4 curve_mapping_ext_out_x; - vec4 curve_mapping_ext_out_y; - vec4 curve_mapping_first_x; - vec4 curve_mapping_first_y; - vec4 curve_mapping_last_x; - vec4 curve_mapping_last_y; - vec4 curve_mapping_black; - vec4 curve_mapping_bwmul; - int curve_mapping_lut_size; - int curve_mapping_use_extend_extrapolate; -}; - -float read_curve_mapping(int table, int index) -{ - return texelFetch(curve_mapping_texture, index, 0)[table]; -} - -float curvemap_calc_extend(int table, float x, vec2 first, vec2 last) -{ - if (x <= first[0]) { - if (curve_mapping_use_extend_extrapolate == 0) { - /* horizontal extrapolation */ - return first[1]; - } - else { - float fac = (curve_mapping_ext_in_x[table] != 0.0) ? - ((x - first[0]) / curve_mapping_ext_in_x[table]) : - 10000.0; - return first[1] + curve_mapping_ext_in_y[table] * fac; - } - } - else if (x >= last[0]) { - if (curve_mapping_use_extend_extrapolate == 0) { - /* horizontal extrapolation */ - return last[1]; - } - else { - float fac = (curve_mapping_ext_out_x[table] != 0.0) ? - ((x - last[0]) / curve_mapping_ext_out_x[table]) : - -10000.0; - return last[1] + curve_mapping_ext_out_y[table] * fac; - } - } - return 0.0; -} - -float curvemap_evaluateF(int table, float value) -{ - float mintable_ = curve_mapping_mintable[table]; - float range = curve_mapping_range[table]; - float mintable = 0.0; - int CM_TABLE = curve_mapping_lut_size - 1; - - float fi; - int i; - - /* index in table */ - fi = (value - mintable) * range; - i = int(fi); - - /* fi is table float index and should check against table range i.e. [0.0 CM_TABLE] */ - if (fi < 0.0 || fi > float(CM_TABLE)) { - return curvemap_calc_extend(table, - value, - vec2(curve_mapping_first_x[table], curve_mapping_first_y[table]), - vec2(curve_mapping_last_x[table], curve_mapping_last_y[table])); - } - else { - if (i < 0) { - return read_curve_mapping(table, 0); - } - if (i >= CM_TABLE) { - return read_curve_mapping(table, CM_TABLE); - } - fi = fi - float(i); - float cm1 = read_curve_mapping(table, i); - float cm2 = read_curve_mapping(table, i + 1); - return mix(cm1, cm2, fi); - } -} - -vec4 curvemapping_evaluate_premulRGBF(vec4 col) -{ - col.rgb = (col.rgb - curve_mapping_black.rgb) * curve_mapping_bwmul.rgb; - - vec4 result; - result.r = curvemap_evaluateF(0, col.r); - result.g = curvemap_evaluateF(1, col.g); - result.b = curvemap_evaluateF(2, col.b); - result.a = col.a; - return result; -} -#endif /* USE_CURVE_MAPPING */ - -/* Using a triangle distribution which gives a more final uniform noise. - * See Banding in Games:A Noisy Rant(revision 5) Mikkel Gjøl, Playdead (slide 27) */ -/* GPUs are rounding before writing to framebuffer so we center the distribution around 0.0. */ -/* Return triangle noise in [-1..1[ range */ -float dither_random_value(vec2 co) -{ - /* Original code from https://www.shadertoy.com/view/4t2SDh */ - /* Uniform noise in [0..1[ range */ - float nrnd0 = fract(sin(dot(co.xy, vec2(12.9898, 78.233))) * 43758.5453); - /* Convert uniform distribution into triangle-shaped distribution. */ - float orig = nrnd0 * 2.0 - 1.0; - nrnd0 = orig * inversesqrt(abs(orig)); - nrnd0 = max(-1.0, nrnd0); /* Removes nan's */ - return nrnd0 - sign(orig); -} - -vec2 round_to_pixel(sampler2D tex, vec2 uv) -{ - vec2 size = textureSize(tex, 0); - return vec2(ivec2(uv * size)) / size; -} - -vec4 apply_dither(vec4 col, vec2 uv) -{ - col.rgb += dither_random_value(uv) * 0.0033 * dither; - return col; -} - -vec4 OCIO_ProcessColor(vec4 col, vec4 col_overlay, vec2 noise_uv) -{ -#ifdef USE_CURVE_MAPPING - col = curvemapping_evaluate_premulRGBF(col); -#endif - - if (predivide) { - if (col.a > 0.0 && col.a < 1.0) { - col.rgb *= 1.0 / col.a; - } - } - - /* NOTE: This is true we only do de-premul here and NO premul - * and the reason is simple -- opengl is always configured - * for straight alpha at this moment - */ - - /* Convert to scene linear (usually a no-op). */ - col = OCIO_to_scene_linear(col); - - /* Apply exposure in scene linear. */ - col.rgb *= scale; - - /* Convert to display space. */ - col = OCIO_to_display(col); - - /* Blend with overlay in UI colorspace. - * - * UI colorspace here refers to the display linear color space, - * i.e: The linear color space w.r.t. display chromaticity and radiometry. - * We separate the colormanagement process into two steps to be able to - * merge UI using alpha blending in the correct color space. */ - if (overlay) { - col.rgb = pow(col.rgb, vec3(exponent * 2.2)); - col = clamp(col, 0.0, 1.0); - col *= 1.0 - col_overlay.a; - col += col_overlay; /* Assumed unassociated alpha. */ - col.rgb = pow(col.rgb, vec3(1.0 / 2.2)); - } - else { - col.rgb = pow(col.rgb, vec3(exponent)); - } - - if (dither > 0.0) { - col = apply_dither(col, noise_uv); - } - - return col; -} - -/* ------------------------------------------------------------------------ */ - -in vec2 texCoord_interp; -out vec4 fragColor; - -void main() -{ - vec4 col = texture(image_texture, texCoord_interp.st); - vec4 col_overlay = texture(overlay_texture, texCoord_interp.st); - vec2 noise_uv = round_to_pixel(image_texture, texCoord_interp.st); - - fragColor = OCIO_ProcessColor(col, col_overlay, noise_uv); -} |