Go to the documentation of this file.
45 C(0, shared vec4 cache[DIR(gl_WorkGroupSize) + FILTER_RADIUS*2 + 1]; )
47 C(0,
void distort(
const ivec2
pos,
const int idx) )
49 C(1,
const uint cp = DIR(gl_LocalInvocationID) + FILTER_RADIUS; )
51 C(1, cache[cp] = texture(input_img[idx],
pos); )
53 C(1,
const ivec2 loc_l =
pos - INC(FILTER_RADIUS); )
54 C(1, cache[cp - FILTER_RADIUS] = texture(input_img[idx], loc_l); )
56 C(1,
const ivec2 loc_h =
pos + INC(DIR(gl_WorkGroupSize)); )
57 C(1, cache[cp + DIR(gl_WorkGroupSize)] = texture(input_img[idx], loc_h); )
61 C(1, vec4 sum = vec4(0); )
62 C(1,
for (
int p = -FILTER_RADIUS; p <= FILTER_RADIUS; p++) )
63 C(2, sum += cache[cp + p]; )
65 C(1, sum /= vec4(FILTER_RADIUS*2 + 1); )
66 C(1, imageStore(output_img[idx],
pos, sum); )
81 .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
84 .stages = VK_SHADER_STAGE_COMPUTE_BIT,
89 .type = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE,
91 .mem_quali =
"writeonly",
94 .stages = VK_SHADER_STAGE_COMPUTE_BIT,
101 s->vkctx.queue_family_idx =
s->vkctx.hwctx->queue_family_comp_index;
106 desc_i[0].
updater =
s->input_images;
114 VK_SHADER_STAGE_COMPUTE_BIT);
120 GLSLF(0, #define FILTER_RADIUS (%
i) ,
s->size_x - 1);
121 GLSLC(0, #define INC(
x) (ivec2(
x, 0)) );
122 GLSLC(0, #define DIR(var) (var.x) );
127 GLSLC(1,
const ivec2
pos = ivec2(gl_GlobalInvocationID.xy); );
132 if (
s->planes & (1 <<
i)) {
135 GLSLF(2, vec4 res = texture(input_img[%
i],
pos); ,
i);
136 GLSLF(2, imageStore(output_img[%
i],
pos, res); ,
i);
150 desc_i[1].
updater =
s->output_images;
157 VK_SHADER_STAGE_COMPUTE_BIT);
163 GLSLF(0, #define FILTER_RADIUS (%
i) ,
s->size_y - 1);
164 GLSLC(0, #define INC(
x) (ivec2(0,
x)) );
165 GLSLC(0, #define DIR(var) (var.y) );
170 GLSLC(1,
const ivec2
pos = ivec2(gl_GlobalInvocationID.xy); );
175 if (
s->planes & (1 <<
i)) {
178 GLSLF(2, vec4 res = texture(input_img[%
i],
pos); ,
i);
179 GLSLF(2, imageStore(output_img[%
i],
pos, res); ,
i);
205 VkCommandBuffer cmd_buf;
232 s->input_images[
i].imageLayout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL;
233 s->tmp_images[
i].imageLayout = VK_IMAGE_LAYOUT_GENERAL;
234 s->output_images[
i].imageLayout = VK_IMAGE_LAYOUT_GENERAL;
241 VkImageMemoryBarrier bar[] = {
243 .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
245 .dstAccessMask = VK_ACCESS_SHADER_READ_BIT,
246 .oldLayout =
in->layout[
i],
247 .newLayout =
s->input_images[
i].imageLayout,
248 .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
249 .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
251 .subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
252 .subresourceRange.levelCount = 1,
253 .subresourceRange.layerCount = 1,
256 .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
258 .dstAccessMask = VK_ACCESS_SHADER_WRITE_BIT | VK_ACCESS_SHADER_READ_BIT,
259 .oldLayout =
tmp->layout[
i],
260 .newLayout =
s->tmp_images[
i].imageLayout,
261 .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
262 .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
263 .image =
tmp->img[
i],
264 .subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
265 .subresourceRange.levelCount = 1,
266 .subresourceRange.layerCount = 1,
269 .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
271 .dstAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
272 .oldLayout =
out->layout[
i],
273 .newLayout =
s->output_images[
i].imageLayout,
274 .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
275 .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
276 .image =
out->img[
i],
277 .subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
278 .subresourceRange.levelCount = 1,
279 .subresourceRange.layerCount = 1,
283 vkCmdPipelineBarrier(cmd_buf, VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT,
284 VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0,
287 in->layout[
i] = bar[0].newLayout;
288 in->access[
i] = bar[0].dstAccessMask;
290 tmp->layout[
i] = bar[1].newLayout;
291 tmp->access[
i] = bar[1].dstAccessMask;
293 out->layout[
i] = bar[2].newLayout;
294 out->access[
i] = bar[2].dstAccessMask;
300 s->vkctx.output_height, 1);
304 vkCmdDispatch(cmd_buf,
s->vkctx.output_width,
371 #define OFFSET(x) offsetof(AvgBlurVulkanContext, x)
372 #define FLAGS (AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM)
402 .
name =
"avgblur_vulkan",
410 .priv_class = &avgblur_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 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
static const AVFilterPad avgblur_vulkan_outputs[]
#define FF_FILTER_FLAG_HWFRAME_AWARE
The filter is aware of hardware frames, and any hardware frame context should not be automatically pr...
static const AVFilterPad avgblur_vulkan_inputs[]
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
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.
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()
const char * name
Filter name.
static void avgblur_vulkan_uninit(AVFilterContext *avctx)
static const char blur_kernel[]
A link between two filters.
AVFILTER_DEFINE_CLASS(avgblur_vulkan)
#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.
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.
static const AVOption avgblur_vulkan_options[]
FFmpeg Automated Testing Environment ************************************Introduction Using FATE from your FFmpeg source directory Submitting the results to the FFmpeg result aggregation server Uploading new samples to the fate suite FATE makefile targets and variables Makefile targets Makefile variables Examples Introduction **************FATE is an extended regression suite on the client side and a means for results aggregation and presentation on the server side The first part of this document explains how you can use FATE from your FFmpeg source directory to test your ffmpeg binary The second part describes how you can run FATE to submit the results to FFmpeg’s FATE server In any way you can have a look at the publicly viewable FATE results by visiting this as it can be seen if some test on some platform broke with their recent contribution This usually happens on the platforms the developers could not test on The second part of this document describes how you can run FATE to submit your results to FFmpeg’s FATE server If you want to submit your results be sure to check that your combination of OS and compiler is not already listed on the above mentioned website In the third part you can find a comprehensive listing of FATE makefile targets and variables Using FATE from your FFmpeg source directory **********************************************If you want to run FATE on your machine you need to have the samples in place You can get the samples via the build target fate rsync Use this command from the top level source this will cause FATE to fail NOTE To use a custom wrapper to run the pass ‘ target exec’ to ‘configure’ or set the TARGET_EXEC Make variable Submitting the results to the FFmpeg result aggregation server ****************************************************************To submit your results to the server you should run fate through the shell script ‘tests fate sh’ from the FFmpeg sources This script needs to be invoked with a configuration file as its first argument tests fate sh path to fate_config A configuration file template with comments describing the individual configuration variables can be found at ‘doc fate_config sh template’ Create a configuration that suits your based on the configuration template The ‘slot’ configuration variable can be any string that is not yet but it is suggested that you name it adhering to the following pattern ‘ARCH OS COMPILER COMPILER VERSION’ The configuration file itself will be sourced in a shell therefore all shell features may be used This enables you to setup the environment as you need it for your build For your first test runs the ‘fate_recv’ variable should be empty or commented out This will run everything as normal except that it will omit the submission of the results to the server The following files should be present in $workdir as specified in the configuration it may help to try out the ‘ssh’ command with one or more ‘ v’ options You should get detailed output concerning your SSH configuration and the authentication process The only thing left is to automate the execution of the fate sh script and the synchronisation of the samples directory Uploading new samples to the fate suite *****************************************If you need a sample uploaded send a mail to samples request This is for developers who have an account on the fate suite server If you upload new please make sure they are as small as space on each network bandwidth and so on benefit from smaller test cases Also keep in mind older checkouts use existing sample that means in practice generally do not remove or overwrite files as it likely would break older checkouts or releases Also all needed samples for a commit should be ideally before the push If you need an account for frequently uploading samples or you wish to help others by doing that send a mail to ffmpeg devel rsync vauL Duo x
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".
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)
int ff_vk_filter_config_input(AVFilterLink *inlink)
const VkComponentMapping ff_comp_identity_map
VkDescriptorImageInfo input_images[3]
static const AVFilterPad outputs[]
static int process_frames(AVFilterContext *avctx, AVFrame *out_f, AVFrame *tmp_f, AVFrame *in_f)
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.
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.
VkDescriptorImageInfo output_images[3]
VulkanFilterContext vkctx
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
#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)
VkDescriptorImageInfo tmp_images[3]
static av_cold int init_filter(AVFilterContext *ctx, AVFrame *in)
#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.
static const struct @315 planes[]
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
const char * name
Pad name.
int ff_vk_start_exec_recording(AVFilterContext *avctx, FFVkExecContext *e)
Begin recording to the command buffer.
#define FF_ARRAY_ELEMS(a)
int ff_vk_submit_exec_queue(AVFilterContext *avctx, FFVkExecContext *e)
Submits a command buffer to the queue for execution.
const char * ff_vk_shader_rep_fmt(enum AVPixelFormat pixfmt)
Gets the glsl format string for a pixel format.
AVFilter ff_vf_avgblur_vulkan
int h
agreed upon image height
static int avgblur_vulkan_filter_frame(AVFilterLink *link, AVFrame *in)
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)
int ff_vk_init_compute_pipeline(AVFilterContext *avctx, VulkanPipeline *pl)
Initializes a compute pipeline.
static av_cold int uninit(AVCodecContext *avctx)
int ff_vk_create_imageview(AVFilterContext *avctx, FFVkExecContext *e, VkImageView *v, VkImage img, VkFormat fmt, const VkComponentMapping map)
Create an imageview.
int ff_vk_filter_config_output(AVFilterLink *outlink)