// Set up shared/commonly used variables and macros
void sampler_prelude(struct gl_shader_cache *sc, int tex_num)
{
GLSLF("#undef tex\n");
GLSLF("#define tex texture%d\n", tex_num);
GLSLF("vec2 pos = texcoord%d;\n", tex_num);
GLSLF("vec2 size = texture_size%d;\n", tex_num);
GLSLF("vec2 pt = vec2(1.0) / size;\n");
}
void pass_nnedi3(GL *gl, struct gl_shader_cache *sc, int planes, int tex_num,
int step, float tex_mul, const struct nnedi3_opts *conf,
struct gl_transform *transform)
{
assert(0 <= step && step < 2);
if (!conf)
conf = &nnedi3_opts_def;
const int neurons = nnedi3_neurons[conf->neurons];
const int width = nnedi3_window_width[conf->window];
const int height = nnedi3_window_height[conf->window];
const int offset = nnedi3_weight_offsets[conf->window * 4 + conf->neurons];
const uint32_t *weights = (const int*)(nnedi3_weights + offset * 4);
GLSLF("// nnedi3 (step %d, neurons %d, window %dx%d, mode %d)\n",
step, neurons, width, height, conf->upload);
// This is required since each row will be encoded into vec4s
assert(width % 4 == 0);
const int sample_count = width * height / 4;
if (conf->upload == NNEDI3_UPLOAD_UBO) {
char buf[32];
snprintf(buf, sizeof(buf), "vec4 weights[%d];",
neurons * (sample_count * 2 + 1));
gl_sc_uniform_buffer(sc, "NNEDI3_WEIGHTS", buf, 0);
if (!gl->es && gl->glsl_version < 140)
gl_sc_enable_extension(sc, "GL_ARB_uniform_buffer_object");
} else if (conf->upload == NNEDI3_UPLOAD_SHADER) {
// Somehow necessary for hard coding approach.
GLSLH(#pragma optionNV(fastprecision on))
}
static void pass_sample_separated_get_weights(struct gl_shader_cache *sc,
struct scaler *scaler)
{
gl_sc_uniform_texture(sc, "lut", scaler->lut);
GLSLF("float ypos = LUT_POS(fcoord, %d.0);\n", scaler->lut_size);
int N = scaler->kernel->size;
int width = (N + 3) / 4; // round up
GLSLF("float weights[%d];\n", N);
for (int i = 0; i < N; i++) {
if (i % 4 == 0)
GLSLF("c = texture(lut, vec2(%f, ypos));\n", (i / 4 + 0.5) / width);
GLSLF("weights[%d] = c[%d];\n", i, i % 4);
}
}
static void pass_sample_separated_get_weights(struct gl_shader_cache *sc,
struct scaler *scaler)
{
gl_sc_uniform_sampler(sc, "lut", scaler->gl_target,
TEXUNIT_SCALERS + scaler->index);
// Define a new variable to cache the corrected fcoord.
GLSLF("float fcoord_lut = LUT_POS(fcoord, %d.0);\n", scaler->lut_size);
int N = scaler->kernel->size;
if (N == 2) {
GLSL(vec2 c1 = texture(lut, vec2(0.5, fcoord_lut)).RG;)
static void update_osd(struct vo *vo)
{
struct priv *p = vo->priv;
if (!p->enable_osd)
return;
mpgl_osd_generate(p->osd, p->osd_res, p->osd_pts, 0, 0);
int64_t osd_change_counter = mpgl_get_change_counter(p->osd);
if (p->osd_change_counter == osd_change_counter) {
p->skip_osd = true;
return;
}
p->osd_change_counter = osd_change_counter;
MP_STATS(vo, "start rpi_osd");
p->egl.gl->ClearColor(0, 0, 0, 0);
p->egl.gl->Clear(GL_COLOR_BUFFER_BIT);
for (int n = 0; n < MAX_OSD_PARTS; n++) {
enum sub_bitmap_format fmt = mpgl_osd_get_part_format(p->osd, n);
if (!fmt)
continue;
gl_sc_uniform_sampler(p->sc, "osdtex", GL_TEXTURE_2D, 0);
switch (fmt) {
case SUBBITMAP_RGBA: {
GLSLF("// OSD (RGBA)\n");
GLSL(color = texture(osdtex, texcoord).bgra;)
break;
}
case SUBBITMAP_LIBASS: {
GLSLF("// OSD (libass)\n");
GLSL(color =
vec4(ass_color.rgb, ass_color.a * texture(osdtex, texcoord).r);)
break;
}
default:
abort();
}