Go to the documentation of this file.
66 #define OFFSET(x) offsetof(VagueDenoiserContext, x)
67 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_FILTERING_PARAM
88 0.037828455506995f, -0.023849465019380f, -0.110624404418423f, 0.377402855612654f,
89 0.852698679009403f, 0.377402855612654f, -0.110624404418423f, -0.023849465019380f, 0.037828455506995f
93 -0.064538882628938f, 0.040689417609558f, 0.418092273222212f, -0.788485616405664f,
94 0.418092273222212f, 0.040689417609558f, -0.064538882628938f
98 -0.064538882628938f, -0.040689417609558f, 0.418092273222212f, 0.788485616405664f,
99 0.418092273222212f, -0.040689417609558f, -0.064538882628938f
103 -0.037828455506995f, -0.023849465019380f, 0.110624404418423f, 0.377402855612654f,
104 -0.852698679009403f, 0.377402855612654f, 0.110624404418423f, -0.023849465019380f, -0.037828455506995f
137 int p,
i, nsteps_width, nsteps_height, nsteps_max;
139 s->depth =
desc->comp[0].depth;
140 s->bpc = (
s->depth + 7) / 8;
141 s->nb_planes =
desc->nb_components;
144 s->planeheight[0] =
s->planeheight[3] =
inlink->h;
146 s->planewidth[0] =
s->planewidth[3] =
inlink->w;
153 if (!
s->block || !
s->in || !
s->out || !
s->tmp)
156 s->threshold *= 1 << (
s->depth - 8);
157 s->peak = (1 <<
s->depth) - 1;
159 nsteps_width = ((
s->planes & 2 ||
s->planes & 4) &&
s->nb_planes > 1) ?
s->planewidth[1] :
s->planewidth[0];
160 nsteps_height = ((
s->planes & 2 ||
s->planes & 4) &&
s->nb_planes > 1) ?
s->planeheight[1] :
s->planeheight[0];
162 for (nsteps_max = 1; nsteps_max < 15; nsteps_max++) {
163 if (pow(2, nsteps_max) >= nsteps_width || pow(2, nsteps_max) >= nsteps_height)
167 s->nsteps =
FFMIN(
s->nsteps, nsteps_max - 2);
169 for (p = 0; p < 4; p++) {
170 s->hlowsize[p][0] = (
s->planewidth[p] + 1) >> 1;
171 s->hhighsize[p][0] =
s->planewidth[p] >> 1;
172 s->vlowsize[p][0] = (
s->planeheight[p] + 1) >> 1;
173 s->vhighsize[p][0] =
s->planeheight[p] >> 1;
175 for (
i = 1;
i <
s->nsteps;
i++) {
176 s->hlowsize[p][
i] = (
s->hlowsize[p][
i - 1] + 1) >> 1;
177 s->hhighsize[p][
i] =
s->hlowsize[p][
i - 1] >> 1;
178 s->vlowsize[p][
i] = (
s->vlowsize[p][
i - 1] + 1) >> 1;
179 s->vhighsize[p][
i] =
s->vlowsize[p][
i - 1] >> 1;
186 static inline void copy(
const float *p1,
float *p2,
const int length)
188 memcpy(p2, p1, length *
sizeof(
float));
191 static inline void copyv(
const float *p1,
const int stride1,
float *p2,
const int length)
195 for (
i = 0;
i < length;
i++) {
201 static inline void copyh(
const float *p1,
float *p2,
const int stride2,
const int length)
205 for (
i = 0;
i < length;
i++) {
221 const int originalLast = last;
231 for (
i = 0;
i < nextend;
i++)
237 nextend = idx - last;
238 for (
i = 0;
i < nextend;
i++)
277 const int low_size = (
size + 1) >> 1;
278 const int high_size =
size >> 1;
284 right_ext = (
size % 2 == 0) ? 2 : 1;
288 findex = (
size + 2) >> 1;
290 for (
i = 9;
i < findex + 11;
i++) {
308 right_ext = (
size % 2 == 0) ? 1 : 2;
311 for (
i = 8;
i < findex + 11;
i++) {
331 const int stride,
const float threshold,
334 const float frac = 1.f - percent * 0.01f;
337 for (y = 0; y <
height; y++) {
338 for (x = 0; x <
width; x++) {
347 const float threshold,
const float percent)
349 const float frac = 1.f - percent * 0.01f;
350 const float shift = threshold * 0.01f * percent;
353 for (y = 0; y <
height; y++) {
354 for (x = 0; x <
width; x++) {
356 if (
temp <= threshold)
366 const int stride,
const float threshold,
369 const float percent01 = percent * 0.01f;
370 const float tr2 = threshold * threshold * percent01;
371 const float frac = 1.f - percent01;
374 for (y = 0; y <
height; y++) {
375 for (x = 0; x <
width; x++) {
377 if (
temp <= threshold) {
381 block[x] *= (tp2 - tr2) / tp2;
389 const int stride,
const float threshold)
393 for (
int y = 0; y <
height; y++) {
394 for (
int x = 0; x <
width; x++) {
402 return threshold * threshold / (
FFMAX(
sqrtf(
mean - threshold), FLT_EPSILON));
409 for (p = 0; p <
s->nb_planes; p++) {
410 const int height =
s->planeheight[p];
411 const int width =
s->planewidth[p];
412 const uint8_t *srcp8 = in->
data[p];
413 const uint16_t *srcp16 = (
const uint16_t *)in->
data[p];
414 uint8_t *dstp8 =
out->data[p];
415 uint16_t *dstp16 = (uint16_t *)
out->data[p];
417 int h_low_size0 =
width;
419 int nsteps_transform =
s->nsteps;
420 int nsteps_invert =
s->nsteps;
421 const float *
input =
s->block;
423 if (!((1 << p) &
s->planes)) {
425 s->planewidth[p] *
s->bpc,
s->planeheight[p]);
430 for (y = 0; y <
height; y++) {
431 for (x = 0; x <
width; x++)
437 for (y = 0; y <
height; y++) {
438 for (x = 0; x <
width; x++)
445 while (nsteps_transform--) {
446 int low_size = (h_low_size0 + 1) >> 1;
448 for (j = 0; j < v_low_size0; j++) {
455 low_size = (v_low_size0 + 1) >> 1;
457 for (j = 0; j < h_low_size0; j++) {
464 h_low_size0 = (h_low_size0 + 1) >> 1;
465 v_low_size0 = (v_low_size0 + 1) >> 1;
471 for (
int n = 0; n <
s->nsteps; n++) {
475 if (n ==
s->nsteps - 1) {
477 s->thresholding(
s->block,
s->hlowsize[p][n],
s->vlowsize[p][n],
width, threshold,
s->percent);
479 block =
s->block +
s->hlowsize[p][n];
481 s->thresholding(
block,
s->hhighsize[p][n],
s->vlowsize[p][n],
width, threshold,
s->percent);
484 s->thresholding(
block,
s->hlowsize[p][n],
s->vhighsize[p][n],
width, threshold,
s->percent);
485 block =
s->block +
s->hlowsize[p][n] +
s->vlowsize[p][n] *
width;
487 s->thresholding(
block,
s->hhighsize[p][n],
s->vhighsize[p][n],
width, threshold,
s->percent);
491 while (nsteps_invert--) {
492 const int idx =
s->vlowsize[p][nsteps_invert] +
s->vhighsize[p][nsteps_invert];
493 const int idx2 =
s->hlowsize[p][nsteps_invert] +
s->hhighsize[p][nsteps_invert];
494 float * idx3 =
s->block;
495 for (
i = 0;
i < idx2;
i++) {
503 for (
i = 0;
i < idx;
i++) {
512 for (y = 0; y <
height; y++) {
513 for (x = 0; x <
width; x++)
516 dstp8 +=
out->linesize[p];
519 for (y = 0; y <
height; y++) {
520 for (x = 0; x <
width; x++)
523 dstp16 +=
out->linesize[p] / 2;
597 .
name =
"vaguedenoiser",
600 .priv_class = &vaguedenoiser_class,
AVFrame * ff_get_video_buffer(AVFilterLink *link, int w, int h)
Request a picture buffer with a specific set of permissions.
#define AV_PIX_FMT_YUVA422P16
#define AV_PIX_FMT_GBRAP16
AVPixelFormat
Pixel format.
static void transform_step(float *input, float *output, const int size, const int low_size, VagueDenoiserContext *s)
static void filter(VagueDenoiserContext *s, AVFrame *in, AVFrame *out)
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 FILTER_PIXFMTS_ARRAY(array)
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
static void copyv(const float *p1, const int stride1, float *p2, const int length)
filter_frame For filters that do not use the this method is called when a frame is pushed to the filter s input It can be called at any time except in a reentrant way If the input frame is enough to produce output
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.
#define AV_PIX_FMT_YUVA422P9
#define FILTER_INPUTS(array)
This structure describes decoded (raw) audio or video data.
#define AV_PIX_FMT_YUVA420P16
#define AV_PIX_FMT_YUVA420P10
static const struct @467 planes[]
#define AV_PIX_FMT_YUV420P10
@ AV_PIX_FMT_YUV440P
planar YUV 4:4:0 (1 Cr & Cb sample per 1x2 Y samples)
const char * name
Filter name.
static void copyh(const float *p1, float *p2, const int stride2, const int length)
A link between two filters.
#define AV_PIX_FMT_YUVA422P10
static void symmetric_extension(float *output, const int size, const int left_ext, const int right_ext)
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
void av_image_copy_plane(uint8_t *dst, int dst_linesize, const uint8_t *src, int src_linesize, int bytewidth, int height)
Copy image plane from src to dst.
#define AV_PIX_FMT_YUVA420P9
#define AV_PIX_FMT_GBRP14
@ AV_PIX_FMT_GBRAP
planar GBRA 4:4:4:4 32bpp
static const float analysis_high[7]
#define AV_PIX_FMT_GBRP10
#define AV_PIX_FMT_YUVA444P16
#define AV_PIX_FMT_YUV422P9
static void invert_step(const float *input, float *output, float *temp, const int size, VagueDenoiserContext *s)
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf type
#define AV_PIX_FMT_GRAY16
static void qian_thresholding(float *block, const int width, const int height, const int stride, const float threshold, const float percent)
A filter pad used for either input or output.
#define AV_PIX_FMT_YUV444P10
@ AV_PIX_FMT_YUVJ411P
planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples) full scale (JPEG), deprecated in favor ...
trying all byte sequences megabyte in length and selecting the best looking sequence will yield cases to try But first
#define AV_PIX_FMT_YUV422P16
const AVFilterPad ff_video_default_filterpad[1]
An AVFilterPad array whose only entry has name "default" and is of type AVMEDIA_TYPE_VIDEO.
void(* thresholding)(float *block, const int width, const int height, const int stride, const float threshold, const float percent)
@ AV_PIX_FMT_YUVJ422P
planar YUV 4:2:2, 16bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV422P and setting col...
#define AV_PIX_FMT_GBRAP10
static const float synthesis_low[7]
#define AV_PIX_FMT_GBRAP12
@ AV_PIX_FMT_YUVA420P
planar YUV 4:2:0, 20bpp, (1 Cr & Cb sample per 2x2 Y & A samples)
#define AV_PIX_FMT_YUV444P16
#define AV_CEIL_RSHIFT(a, b)
#define AV_PIX_FMT_YUVA444P12
#define AV_PIX_FMT_YUV420P9
#define AV_PIX_FMT_YUV420P16
#define AV_PIX_FMT_GRAY14
static const float synthesis_high[9]
@ AV_PIX_FMT_YUV420P
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
#define FILTER_OUTPUTS(array)
@ AV_PIX_FMT_YUVJ444P
planar YUV 4:4:4, 24bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV444P and setting col...
static float bayes_threshold(float *block, const int width, const int height, const int stride, const float threshold)
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
#define AV_PIX_FMT_GRAY10
#define AV_PIX_FMT_GBRP16
Describe the class of an AVClass context structure.
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
int av_frame_copy_props(AVFrame *dst, const AVFrame *src)
Copy only "metadata" fields from src to dst.
@ AV_PIX_FMT_YUVJ420P
planar YUV 4:2:0, 12bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV420P and setting col...
static void hard_thresholding(float *block, const int width, const int height, const int stride, const float threshold, const float percent)
#define AV_PIX_FMT_YUV440P10
static __device__ float sqrtf(float a)
static const AVFilterPad vaguedenoiser_inputs[]
#define AV_PIX_FMT_YUV422P10
@ AV_PIX_FMT_GRAY8
Y , 8bpp.
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
static av_cold int init(AVFilterContext *ctx)
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
static void copy(const float *p1, float *p2, const int length)
static int shift(int a, int b)
AVFILTER_DEFINE_CLASS(vaguedenoiser)
const AVFilter ff_vf_vaguedenoiser
#define AV_PIX_FMT_YUV422P12
#define AV_PIX_FMT_YUV444P12
int av_frame_is_writable(AVFrame *frame)
Check if the frame data is writable.
static av_cold void uninit(AVFilterContext *ctx)
The reader does not expect b to be semantically here and if the code is changed by maybe adding a a division or other the signedness will almost certainly be mistaken To avoid this confusion a new type was SUINT is the C unsigned type but it holds a signed int to use the same example SUINT a
static int config_input(AVFilterLink *inlink)
@ AV_PIX_FMT_YUVA444P
planar YUV 4:4:4 32bpp, (1 Cr & Cb sample per 1x1 Y & A samples)
#define AV_PIX_FMT_YUVA444P10
and forward the test the status of outputs and forward it to the corresponding return FFERROR_NOT_READY If the filters stores internally one or a few frame for some input
#define AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC
Some filters support a generic "enable" expression option that can be used to enable or disable a fil...
@ AV_OPT_TYPE_FLOAT
Underlying C type is float.
#define i(width, name, range_min, range_max)
int w
agreed upon image width
#define av_malloc_array(a, b)
#define AV_PIX_FMT_GBRP12
@ AV_PIX_FMT_YUVJ440P
planar YUV 4:4:0 full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV440P and setting color_range
const char * name
Pad name.
#define AV_PIX_FMT_YUV444P9
static const AVOption vaguedenoiser_options[]
static const float analysis_low[9]
#define AV_PIX_FMT_YUVA444P9
static int left_ext(int wavelet_length, int levels, uint64_t sn)
#define AV_PIX_FMT_YUV420P12
#define AV_PIX_FMT_YUV422P14
static void soft_thresholding(float *block, const int width, const int height, const int stride, const float threshold, const float percent)
int h
agreed upon image height
#define AV_PIX_FMT_YUVA422P12
@ AV_OPT_TYPE_INT
Underlying C type is int.
static float mean(const float *input, int size)
@ AV_PIX_FMT_YUV444P
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
@ AV_PIX_FMT_GBRP
planar GBR 4:4:4 24bpp
@ AV_PIX_FMT_YUV422P
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Descriptor that unambiguously describes how the bits of a pixel are stored in the up to 4 data planes...
@ AV_PIX_FMT_YUV411P
planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples)
int linesize[AV_NUM_DATA_POINTERS]
For video, a positive or negative value, which is typically indicating the size in bytes of each pict...
@ AV_PIX_FMT_YUV410P
planar YUV 4:1:0, 9bpp, (1 Cr & Cb sample per 4x4 Y samples)
The exact code depends on how similar the blocks are and how related they are to the block
#define AV_PIX_FMT_YUV440P12
#define AV_PIX_FMT_YUV444P14
#define AV_PIX_FMT_GRAY12
@ AV_OPT_TYPE_CONST
Special option type for declaring named constants.
@ AV_PIX_FMT_YUVA422P
planar YUV 4:2:2 24bpp, (1 Cr & Cb sample per 2x1 Y & A samples)
#define AV_PIX_FMT_YUV420P14
static enum AVPixelFormat pix_fmts[]