Go to the documentation of this file.
26 #define CGROUPS (int [3]){ 32, 32, 1 }
58 C(1, vec2 npos = (vec2(
pos) + 0.5
f) / imageSize(output_img[idx]); )
59 C(1, npos *= crop_range; )
60 C(1, npos += crop_off; )
61 C(1,
return texture(input_img[idx], npos); )
68 C(1,
src *= yuv_matrix; )
69 C(1,
if (fullrange == 1) { )
70 C(2,
src += vec4(0.0, 0.5, 0.5, 0.0); )
72 C(2,
src *= vec4(219.0 / 255.0, 224.0 / 255.0, 224.0 / 255.0, 1.0); )
73 C(2,
src += vec4(16.0 / 255.0, 128.0 / 255.0, 128.0 / 255.0, 0.0); )
82 C(1, imageStore(output_img[0],
pos, vec4(
src.r, 0.0, 0.0, 0.0)); )
83 C(1,
pos /= ivec2(2); )
84 C(1, imageStore(output_img[1],
pos, vec4(
src.g,
src.b, 0.0, 0.0)); )
91 C(1, imageStore(output_img[0],
pos, vec4(
src.r, 0.0, 0.0, 0.0)); )
92 C(1,
pos /= ivec2(2); )
93 C(1, imageStore(output_img[1],
pos, vec4(
src.g, 0.0, 0.0, 0.0)); )
94 C(1, imageStore(output_img[2],
pos, vec4(
src.b, 0.0, 0.0, 0.0)); )
101 C(1, imageStore(output_img[0],
pos, vec4(
src.r, 0.0, 0.0, 0.0)); )
102 C(1, imageStore(output_img[1],
pos, vec4(
src.g, 0.0, 0.0, 0.0)); )
103 C(1, imageStore(output_img[2],
pos, vec4(
src.b, 0.0, 0.0, 0.0)); )
111 VkFilter sampler_mode;
114 int crop_x =
in->crop_left;
115 int crop_y =
in->crop_top;
116 int crop_w =
in->width - (
in->crop_left +
in->crop_right);
117 int crop_h =
in->height - (
in->crop_top +
in->crop_bottom);
119 s->vkctx.queue_family_idx =
s->vkctx.hwctx->queue_family_comp_index;
125 sampler_mode = VK_FILTER_NEAREST;
128 sampler_mode = VK_FILTER_LINEAR;
145 .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
148 .stages = VK_SHADER_STAGE_COMPUTE_BIT,
149 .updater =
s->input_images,
153 .name =
"output_img",
154 .type = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE,
156 .mem_quali =
"writeonly",
159 .stages = VK_SHADER_STAGE_COMPUTE_BIT,
160 .updater =
s->output_images,
166 .type = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
167 .mem_quali =
"readonly",
168 .mem_layout =
"std430",
169 .stages = VK_SHADER_STAGE_COMPUTE_BIT,
170 .updater = &
s->params_desc,
171 .buf_content =
"mat4 yuv_matrix;",
175 VK_SHADER_STAGE_COMPUTE_BIT);
186 if (
s->vkctx.output_format !=
s->vkctx.input_format) {
190 switch (
s->vkctx.output_format) {
200 GLSLC(1, ivec2
pos = ivec2(gl_GlobalInvocationID.xy); );
201 GLSLF(1, vec2 in_d = vec2(%
i, %
i); ,
in->width,
in->height);
202 GLSLF(1, vec2 c_r = vec2(%
i, %
i) / in_d; ,crop_w, crop_h);
203 GLSLF(1, vec2 c_o = vec2(%
i, %
i) / in_d; ,crop_x,crop_y);
206 if (
s->vkctx.output_format ==
s->vkctx.input_format) {
207 for (
int i = 0;
i < desc_i[1].
elems;
i++) {
214 GLSLF(2, imageStore(output_img[%
i],
pos, res); ,
i);
222 switch (
s->vkctx.output_format) {
226 default:
return AVERROR(EINVAL);
238 if (
s->vkctx.output_format !=
s->vkctx.input_format) {
240 double tmp_mat[3][3];
243 float yuv_matrix[4][4];
254 VK_BUFFER_USAGE_STORAGE_BUFFER_BIT,
255 VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT);
265 memset(par, 0,
sizeof(*par));
267 for (
int y = 0; y < 3; y++)
268 for (
int x = 0; x < 3; x++)
269 par->yuv_matrix[x][y] = tmp_mat[x][y];
271 par->yuv_matrix[3][3] = 1.0;
277 s->params_desc.buffer =
s->params_buf.buf;
278 s->params_desc.range = VK_WHOLE_SIZE;
297 VkCommandBuffer cmd_buf;
302 int barrier_count = 0;
314 s->input_images[
i].imageLayout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL;
323 s->output_images[
i].imageLayout = VK_IMAGE_LAYOUT_GENERAL;
329 VkImageMemoryBarrier bar = {
330 .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
332 .dstAccessMask = VK_ACCESS_SHADER_READ_BIT,
333 .oldLayout =
in->layout[
i],
334 .newLayout =
s->input_images[
i].imageLayout,
335 .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
336 .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
338 .subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
339 .subresourceRange.levelCount = 1,
340 .subresourceRange.layerCount = 1,
343 memcpy(&barriers[barrier_count++], &bar,
sizeof(VkImageMemoryBarrier));
345 in->layout[
i] = bar.newLayout;
346 in->access[
i] = bar.dstAccessMask;
350 VkImageMemoryBarrier bar = {
351 .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
353 .dstAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
354 .oldLayout =
out->layout[
i],
355 .newLayout =
s->output_images[
i].imageLayout,
356 .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
357 .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
358 .image =
out->img[
i],
359 .subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
360 .subresourceRange.levelCount = 1,
361 .subresourceRange.layerCount = 1,
364 memcpy(&barriers[barrier_count++], &bar,
sizeof(VkImageMemoryBarrier));
366 out->layout[
i] = bar.newLayout;
367 out->access[
i] = bar.dstAccessMask;
370 vkCmdPipelineBarrier(cmd_buf, VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT,
371 VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0,
372 0,
NULL, 0,
NULL, barrier_count, barriers);
376 vkCmdDispatch(cmd_buf,
417 out->color_range =
s->out_range;
418 if (
s->vkctx.output_format !=
s->vkctx.input_format)
439 &
s->vkctx.output_width,
440 &
s->vkctx.output_height);
444 if (
s->out_format_string) {
451 s->vkctx.output_format =
s->vkctx.input_format;
454 if (
s->vkctx.output_format !=
s->vkctx.input_format) {
487 #define OFFSET(x) offsetof(ScaleVulkanContext, x)
488 #define FLAGS (AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM)
528 .
name =
"scale_vulkan",
536 .priv_class = &scale_vulkan_class,
int ff_vk_filter_query_formats(AVFilterContext *avctx)
General lavfi IO functions.
AVFrame * ff_get_video_buffer(AVFilterLink *link, int w, int h)
Request a picture buffer with a specific set of permissions.
static int process_frames(AVFilterContext *avctx, AVFrame *out_f, AVFrame *in_f)
static av_cold int init(AVCodecContext *avctx)
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
#define FF_FILTER_FLAG_HWFRAME_AWARE
The filter is aware of hardware frames, and any hardware frame context should not be automatically pr...
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
AVFilter ff_vf_scale_vulkan
The exact code depends on how similar the blocks are and how related they are to the and needs to apply these operations to the correct inlink or outlink if there are several Macros are available to factor that when no extra processing is inlink
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
This structure describes decoded (raw) audio or video data.
@ AVCOL_RANGE_JPEG
the normal 2^n-1 "JPEG" YUV ranges
static const AVFilterPad scale_vulkan_outputs[]
int ff_vk_mt_is_np_rgb(enum AVPixelFormat pix_fmt)
Returns 1 if the image is any sort of supported RGB.
static const AVOption scale_vulkan_options[]
static const char rgb2yuv[]
static const char write_nv12[]
VkSampler * ff_vk_init_sampler(AVFilterContext *avctx, int unnorm_coords, VkFilter filt)
Create a Vulkan sampler, will be auto-freed in ff_vk_filter_uninit()
int ff_scale_eval_dimensions(void *log_ctx, const char *w_expr, const char *h_expr, AVFilterLink *inlink, AVFilterLink *outlink, int *ret_w, int *ret_h)
Parse and evaluate string expressions for width and height.
const char * name
Filter name.
A link between two filters.
static const char write_420[]
#define DUP_SAMPLER_ARRAY4(x)
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
uint32_t av_get_random_seed(void)
Get a seed to use in conjunction with random functions.
AVFILTER_DEFINE_CLASS(scale_vulkan)
int av_pix_fmt_count_planes(enum AVPixelFormat pix_fmt)
SPIRVShader * ff_vk_init_shader(AVFilterContext *avctx, VulkanPipeline *pl, const char *name, VkShaderStageFlags stage)
Inits a shader for a specific pipeline.
void * priv
private data for use by the filter
VulkanPipeline * ff_vk_create_pipeline(AVFilterContext *avctx)
Inits a pipeline.
int ff_vk_add_exec_dep(AVFilterContext *avctx, FFVkExecContext *e, AVFrame *frame, VkPipelineStageFlagBits in_wait_dst_flag)
Adds a frame as a queue dependency.
A filter pad used for either input or output.
int ff_vk_compile_shader(AVFilterContext *avctx, SPIRVShader *shd, const char *entrypoint)
Compiles the shader, entrypoint must be set to "main".
const struct LumaCoefficients * ff_get_luma_coefficients(enum AVColorSpace csp)
s EdgeDetect Foobar g libavfilter vf_edgedetect c libavfilter vf_foobar c edit libavfilter and add an entry for foobar following the pattern of the other filters edit libavfilter allfilters and add an entry for foobar following the pattern of the other filters configure make j< whatever > ffmpeg ffmpeg i you should get a foobar png with Lena edge detected That s your new playground is ready Some little details about what s going which in turn will define variables for the build system and the C
void ff_vk_filter_uninit(AVFilterContext *avctx)
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
int ff_vk_filter_config_input(AVFilterLink *inlink)
const VkComponentMapping ff_comp_identity_map
void ff_fill_rgb2yuv_table(const struct LumaCoefficients *coeffs, double rgb2yuv[3][3])
static int scale_vulkan_filter_frame(AVFilterLink *link, AVFrame *in)
static const AVFilterPad outputs[]
int ff_vk_map_buffers(AVFilterContext *avctx, FFVkBuffer *buf, uint8_t *mem[], int nb_buffers, int invalidate)
Maps the buffer to userspace.
int ff_vk_init_pipeline_layout(AVFilterContext *avctx, VulkanPipeline *pl)
Initializes the pipeline layout after all shaders and descriptor sets have been finished.
int ff_vk_add_descriptor_set(AVFilterContext *avctx, VulkanPipeline *pl, SPIRVShader *shd, VulkanDescriptorSetBinding *desc, int num, int only_print_to_shader)
Adds a descriptor set to the shader and registers them in the pipeline.
@ AV_PIX_FMT_YUV420P
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a link
void ff_vk_set_compute_shader_sizes(AVFilterContext *avctx, SPIRVShader *shd, int local_size[3])
Writes the workgroup size for a shader.
void ff_vk_bind_pipeline_exec(AVFilterContext *avctx, FFVkExecContext *e, VulkanPipeline *pl)
Add a command to bind the completed pipeline and its descriptor sets.
const VkFormat * av_vkfmt_from_pixfmt(enum AVPixelFormat p)
Returns the format of each image up to the number of planes for a given sw_format.
int main(int argc, char *argv[])
int av_frame_copy_props(AVFrame *dst, const AVFrame *src)
Copy only "metadata" fields from src to dst.
static void scale_vulkan_uninit(AVFilterContext *avctx)
@ AVCHROMA_LOC_TOPLEFT
ITU-R 601, SMPTE 274M 296M S314M(DV 4:1:1), mpeg2 4:2:2.
AVFilterLink ** inputs
array of pointers to input links
VkCommandBuffer ff_vk_get_exec_buf(AVFilterContext *avctx, FFVkExecContext *e)
Gets the command buffer to use for this submission from the exe context.
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several inputs
enum AVColorRange out_range
@ AVCOL_RANGE_UNSPECIFIED
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
#define GET_QUEUE_COUNT(hwctx, graph, comp, tx)
#define AV_NUM_DATA_POINTERS
AVFilterContext * src
source filter
#define AVERROR_EXTERNAL
Generic error in an external library.
int ff_vk_filter_init(AVFilterContext *avctx)
void ff_vk_update_descriptor_set(AVFilterContext *avctx, VulkanPipeline *pl, int set_id)
Updates a descriptor set via the updaters defined.
VkDescriptorImageInfo input_images[3]
uint8_t pi<< 24) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0f/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8, uint8_t,(*(const uint8_t *) pi - 0x80) *(1.0/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16, int16_t,(*(const int16_t *) pi >> 8)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, int16_t, *(const int16_t *) pi *(1.0/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32, int32_t,(*(const int32_t *) pi >> 24)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, int32_t, *(const int32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, float, av_clip_uint8(lrintf(*(const float *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, float, av_clip_int16(lrintf(*(const float *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, float, av_clipl_int32(llrintf(*(const float *) pi *(1U<< 31)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, double, av_clip_uint8(lrint(*(const double *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, double, av_clip_int16(lrint(*(const double *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, double, av_clipl_int32(llrint(*(const double *) pi *(1U<< 31)))) #define SET_CONV_FUNC_GROUP(ofmt, ifmt) static void set_generic_function(AudioConvert *ac) { } void ff_audio_convert_free(AudioConvert **ac) { if(! *ac) return;ff_dither_free(&(*ac) ->dc);av_freep(ac);} AudioConvert *ff_audio_convert_alloc(AVAudioResampleContext *avr, enum AVSampleFormat out_fmt, enum AVSampleFormat in_fmt, int channels, int sample_rate, int apply_map) { AudioConvert *ac;int in_planar, out_planar;ac=av_mallocz(sizeof(*ac));if(!ac) return NULL;ac->avr=avr;ac->out_fmt=out_fmt;ac->in_fmt=in_fmt;ac->channels=channels;ac->apply_map=apply_map;if(avr->dither_method !=AV_RESAMPLE_DITHER_NONE &&av_get_packed_sample_fmt(out_fmt)==AV_SAMPLE_FMT_S16 &&av_get_bytes_per_sample(in_fmt) > 2) { ac->dc=ff_dither_alloc(avr, out_fmt, in_fmt, channels, sample_rate, apply_map);if(!ac->dc) { av_free(ac);return NULL;} return ac;} in_planar=ff_sample_fmt_is_planar(in_fmt, channels);out_planar=ff_sample_fmt_is_planar(out_fmt, channels);if(in_planar==out_planar) { ac->func_type=CONV_FUNC_TYPE_FLAT;ac->planes=in_planar ? ac->channels :1;} else if(in_planar) ac->func_type=CONV_FUNC_TYPE_INTERLEAVE;else ac->func_type=CONV_FUNC_TYPE_DEINTERLEAVE;set_generic_function(ac);if(ARCH_AARCH64) ff_audio_convert_init_aarch64(ac);if(ARCH_ARM) ff_audio_convert_init_arm(ac);if(ARCH_X86) ff_audio_convert_init_x86(ac);return ac;} int ff_audio_convert(AudioConvert *ac, AudioData *out, AudioData *in) { int use_generic=1;int len=in->nb_samples;int p;if(ac->dc) { av_log(ac->avr, AV_LOG_TRACE, "%d samples - audio_convert: %s to %s (dithered)\n", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));return ff_convert_dither(ac-> in
#define i(width, name, range_min, range_max)
int w
agreed upon image width
static const char scale_bilinear[]
static const char write_444[]
static const AVFilterPad scale_vulkan_inputs[]
const char * name
Pad name.
@ AVCOL_RANGE_MPEG
the normal 219*2^(n-8) "MPEG" YUV ranges
@ AV_PIX_FMT_NV12
planar YUV 4:2:0, 12bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (firs...
int ff_vk_start_exec_recording(AVFilterContext *avctx, FFVkExecContext *e)
Begin recording to the command buffer.
int ff_vk_create_buf(AVFilterContext *avctx, FFVkBuffer *buf, size_t size, VkBufferUsageFlags usage, VkMemoryPropertyFlagBits flags)
Create a VkBuffer with the specified parameters.
int ff_vk_submit_exec_queue(AVFilterContext *avctx, FFVkExecContext *e)
Submits a command buffer to the queue for execution.
enum AVPixelFormat av_get_pix_fmt(const char *name)
Return the pixel format corresponding to name.
const char * ff_vk_shader_rep_fmt(enum AVPixelFormat pixfmt)
Gets the glsl format string for a pixel format.
void ff_vk_free_buf(AVFilterContext *avctx, FFVkBuffer *buf)
Frees a buffer.
int ff_vk_unmap_buffers(AVFilterContext *avctx, FFVkBuffer *buf, int nb_buffers, int flush)
Unmaps the buffer from userspace.
int h
agreed upon image height
VkDescriptorImageInfo output_images[3]
@ AV_PIX_FMT_YUV444P
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
int ff_vk_create_exec_ctx(AVFilterContext *avctx, FFVkExecContext **ctx)
Init an execution context for command recording and queue submission.
void ff_vk_discard_exec_deps(AVFilterContext *avctx, FFVkExecContext *e)
Discards all queue dependencies.
static int query_formats(AVFilterContext *ctx)
static int scale_vulkan_config_output(AVFilterLink *outlink)
int ff_vk_init_compute_pipeline(AVFilterContext *avctx, VulkanPipeline *pl)
Initializes a compute pipeline.
static av_cold int uninit(AVCodecContext *avctx)
static av_cold int init_filter(AVFilterContext *ctx, AVFrame *in)
AVColorRange
MPEG vs JPEG YUV range.
int ff_vk_create_imageview(AVFilterContext *avctx, FFVkExecContext *e, VkImageView *v, VkImage img, VkFormat fmt, const VkComponentMapping map)
Create an imageview.
VkDescriptorBufferInfo params_desc
int ff_vk_filter_config_output(AVFilterLink *outlink)
VulkanFilterContext vkctx