FFmpeg
vf_vidstabtransform.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2013 Georg Martius <georg dot martius at web dot de>
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #define DEFAULT_INPUT_NAME "transforms.trf"
22 
23 #include <vid.stab/libvidstab.h>
24 
25 #include "libavutil/common.h"
26 #include "libavutil/opt.h"
27 #include "libavutil/imgutils.h"
28 #include "avfilter.h"
29 #include "internal.h"
30 
31 #include "vidstabutils.h"
32 
33 typedef struct TransformContext {
34  const AVClass *class;
35 
36  VSTransformData td;
37  VSTransformConfig conf;
38 
39  VSTransformations trans; // transformations
40  char *input; // name of transform file
41  int tripod;
42  int debug;
44 
45 #define OFFSET(x) offsetof(TransformContext, x)
46 #define OFFSETC(x) (offsetof(TransformContext, conf)+offsetof(VSTransformConfig, x))
47 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
48 
50  {"input", "set path to the file storing the transforms", OFFSET(input),
51  AV_OPT_TYPE_STRING, {.str = DEFAULT_INPUT_NAME}, .flags = FLAGS },
52  {"smoothing", "set number of frames*2 + 1 used for lowpass filtering", OFFSETC(smoothing),
53  AV_OPT_TYPE_INT, {.i64 = 15}, 0, 1000, FLAGS},
54 
55  {"optalgo", "set camera path optimization algo", OFFSETC(camPathAlgo),
56  AV_OPT_TYPE_INT, {.i64 = VSOptimalL1}, VSOptimalL1, VSAvg, FLAGS, "optalgo"},
57  { "opt", "global optimization", 0, // from version 1.0 on
58  AV_OPT_TYPE_CONST, {.i64 = VSOptimalL1 }, 0, 0, FLAGS, "optalgo"},
59  { "gauss", "gaussian kernel", 0,
60  AV_OPT_TYPE_CONST, {.i64 = VSGaussian }, 0, 0, FLAGS, "optalgo"},
61  { "avg", "simple averaging on motion", 0,
62  AV_OPT_TYPE_CONST, {.i64 = VSAvg }, 0, 0, FLAGS, "optalgo"},
63 
64  {"maxshift", "set maximal number of pixels to translate image", OFFSETC(maxShift),
65  AV_OPT_TYPE_INT, {.i64 = -1}, -1, 500, FLAGS},
66  {"maxangle", "set maximal angle in rad to rotate image", OFFSETC(maxAngle),
67  AV_OPT_TYPE_DOUBLE, {.dbl = -1.0}, -1.0, 3.14, FLAGS},
68 
69  {"crop", "set cropping mode", OFFSETC(crop),
70  AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, FLAGS, "crop"},
71  { "keep", "keep border", 0,
72  AV_OPT_TYPE_CONST, {.i64 = VSKeepBorder }, 0, 0, FLAGS, "crop"},
73  { "black", "black border", 0,
74  AV_OPT_TYPE_CONST, {.i64 = VSCropBorder }, 0, 0, FLAGS, "crop"},
75 
76  {"invert", "invert transforms", OFFSETC(invert),
77  AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, FLAGS},
78  {"relative", "consider transforms as relative", OFFSETC(relative),
79  AV_OPT_TYPE_INT, {.i64 = 1}, 0, 1, FLAGS},
80  {"zoom", "set percentage to zoom (>0: zoom in, <0: zoom out", OFFSETC(zoom),
81  AV_OPT_TYPE_DOUBLE, {.dbl = 0}, -100, 100, FLAGS},
82  {"optzoom", "set optimal zoom (0: nothing, 1: optimal static zoom, 2: optimal dynamic zoom)", OFFSETC(optZoom),
83  AV_OPT_TYPE_INT, {.i64 = 1}, 0, 2, FLAGS},
84  {"zoomspeed", "for adative zoom: percent to zoom maximally each frame", OFFSETC(zoomSpeed),
85  AV_OPT_TYPE_DOUBLE, {.dbl = 0.25}, 0, 5, FLAGS},
86 
87  {"interpol", "set type of interpolation", OFFSETC(interpolType),
88  AV_OPT_TYPE_INT, {.i64 = 2}, 0, 3, FLAGS, "interpol"},
89  { "no", "no interpolation", 0,
90  AV_OPT_TYPE_CONST, {.i64 = VS_Zero }, 0, 0, FLAGS, "interpol"},
91  { "linear", "linear (horizontal)", 0,
92  AV_OPT_TYPE_CONST, {.i64 = VS_Linear }, 0, 0, FLAGS, "interpol"},
93  { "bilinear","bi-linear", 0,
94  AV_OPT_TYPE_CONST, {.i64 = VS_BiLinear},0, 0, FLAGS, "interpol"},
95  { "bicubic", "bi-cubic", 0,
96  AV_OPT_TYPE_CONST, {.i64 = VS_BiCubic },0, 0, FLAGS, "interpol"},
97 
98  {"tripod", "enable virtual tripod mode (same as relative=0:smoothing=0)", OFFSET(tripod),
99  AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS},
100  {"debug", "enable debug mode and writer global motions information to file", OFFSET(debug),
101  AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS},
102  {NULL}
103 };
104 
105 AVFILTER_DEFINE_CLASS(vidstabtransform);
106 
108 {
109  TransformContext *tc = ctx->priv;
110  ff_vs_init();
111  tc->class = &vidstabtransform_class;
112  av_log(ctx, AV_LOG_VERBOSE, "vidstabtransform filter: init %s\n", LIBVIDSTAB_VERSION);
113  return 0;
114 }
115 
117 {
118  TransformContext *tc = ctx->priv;
119 
120  vsTransformDataCleanup(&tc->td);
121  vsTransformationsCleanup(&tc->trans);
122 }
123 
125 {
126  AVFilterContext *ctx = inlink->dst;
127  TransformContext *tc = ctx->priv;
128  FILE *f;
129 
131  int is_planar = desc->flags & AV_PIX_FMT_FLAG_PLANAR;
132 
133  VSTransformData *td = &(tc->td);
134 
135  VSFrameInfo fi_src;
136  VSFrameInfo fi_dest;
137 
138  if (!vsFrameInfoInit(&fi_src, inlink->w, inlink->h,
139  ff_av2vs_pixfmt(ctx, inlink->format)) ||
140  !vsFrameInfoInit(&fi_dest, inlink->w, inlink->h,
141  ff_av2vs_pixfmt(ctx, inlink->format))) {
142  av_log(ctx, AV_LOG_ERROR, "unknown pixel format: %i (%s)",
143  inlink->format, desc->name);
144  return AVERROR(EINVAL);
145  }
146 
147  if ((!is_planar && fi_src.bytesPerPixel != av_get_bits_per_pixel(desc)/8) ||
148  fi_src.log2ChromaW != desc->log2_chroma_w ||
149  fi_src.log2ChromaH != desc->log2_chroma_h) {
150  av_log(ctx, AV_LOG_ERROR, "pixel-format error: bpp %i<>%i ",
151  fi_src.bytesPerPixel, av_get_bits_per_pixel(desc)/8);
152  av_log(ctx, AV_LOG_ERROR, "chroma_subsampl: w: %i<>%i h: %i<>%i\n",
153  fi_src.log2ChromaW, desc->log2_chroma_w,
154  fi_src.log2ChromaH, desc->log2_chroma_h);
155  return AVERROR(EINVAL);
156  }
157 
158  // set values that are not initializes by the options
159  tc->conf.modName = "vidstabtransform";
160  tc->conf.verbose = 1 + tc->debug;
161  if (tc->tripod) {
162  av_log(ctx, AV_LOG_INFO, "Virtual tripod mode: relative=0, smoothing=0\n");
163  tc->conf.relative = 0;
164  tc->conf.smoothing = 0;
165  }
166  tc->conf.simpleMotionCalculation = 0;
167  tc->conf.storeTransforms = tc->debug;
168  tc->conf.smoothZoom = 0;
169 
170  if (vsTransformDataInit(td, &tc->conf, &fi_src, &fi_dest) != VS_OK) {
171  av_log(ctx, AV_LOG_ERROR, "initialization of vid.stab transform failed, please report a BUG\n");
172  return AVERROR(EINVAL);
173  }
174 
175  vsTransformGetConfig(&tc->conf, td);
176  av_log(ctx, AV_LOG_INFO, "Video transformation/stabilization settings (pass 2/2):\n");
177  av_log(ctx, AV_LOG_INFO, " input = %s\n", tc->input);
178  av_log(ctx, AV_LOG_INFO, " smoothing = %d\n", tc->conf.smoothing);
179  av_log(ctx, AV_LOG_INFO, " optalgo = %s\n",
180  tc->conf.camPathAlgo == VSOptimalL1 ? "opt" :
181  (tc->conf.camPathAlgo == VSGaussian ? "gauss" : "avg"));
182  av_log(ctx, AV_LOG_INFO, " maxshift = %d\n", tc->conf.maxShift);
183  av_log(ctx, AV_LOG_INFO, " maxangle = %f\n", tc->conf.maxAngle);
184  av_log(ctx, AV_LOG_INFO, " crop = %s\n", tc->conf.crop ? "Black" : "Keep");
185  av_log(ctx, AV_LOG_INFO, " relative = %s\n", tc->conf.relative ? "True": "False");
186  av_log(ctx, AV_LOG_INFO, " invert = %s\n", tc->conf.invert ? "True" : "False");
187  av_log(ctx, AV_LOG_INFO, " zoom = %f\n", tc->conf.zoom);
188  av_log(ctx, AV_LOG_INFO, " optzoom = %s\n",
189  tc->conf.optZoom == 1 ? "Static (1)" : (tc->conf.optZoom == 2 ? "Dynamic (2)" : "Off (0)"));
190  if (tc->conf.optZoom == 2)
191  av_log(ctx, AV_LOG_INFO, " zoomspeed = %g\n", tc->conf.zoomSpeed);
192  av_log(ctx, AV_LOG_INFO, " interpol = %s\n", getInterpolationTypeName(tc->conf.interpolType));
193 
194  f = fopen(tc->input, "r");
195  if (!f) {
196  int ret = AVERROR(errno);
197  av_log(ctx, AV_LOG_ERROR, "cannot open input file %s\n", tc->input);
198  return ret;
199  } else {
200  VSManyLocalMotions mlms;
201  if (vsReadLocalMotionsFile(f, &mlms) == VS_OK) {
202  // calculate the actual transforms from the local motions
203  if (vsLocalmotions2Transforms(td, &mlms, &tc->trans) != VS_OK) {
204  av_log(ctx, AV_LOG_ERROR, "calculating transformations failed\n");
205  return AVERROR(EINVAL);
206  }
207  } else { // try to read old format
208  if (!vsReadOldTransforms(td, f, &tc->trans)) { /* read input file */
209  av_log(ctx, AV_LOG_ERROR, "error parsing input file %s\n", tc->input);
210  return AVERROR(EINVAL);
211  }
212  }
213  }
214  fclose(f);
215 
216  if (vsPreprocessTransforms(td, &tc->trans) != VS_OK) {
217  av_log(ctx, AV_LOG_ERROR, "error while preprocessing transforms\n");
218  return AVERROR(EINVAL);
219  }
220 
221  // TODO: add sharpening, so far the user needs to call the unsharp filter manually
222  return 0;
223 }
224 
225 
227 {
228  AVFilterContext *ctx = inlink->dst;
229  TransformContext *tc = ctx->priv;
230  VSTransformData* td = &(tc->td);
231 
232  AVFilterLink *outlink = inlink->dst->outputs[0];
233  int direct = 0;
234  AVFrame *out;
235  VSFrame inframe;
236  int plane;
237 
238  if (av_frame_is_writable(in)) {
239  direct = 1;
240  out = in;
241  } else {
242  out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
243  if (!out) {
244  av_frame_free(&in);
245  return AVERROR(ENOMEM);
246  }
248  }
249 
250  for (plane = 0; plane < vsTransformGetSrcFrameInfo(td)->planes; plane++) {
251  inframe.data[plane] = in->data[plane];
252  inframe.linesize[plane] = in->linesize[plane];
253  }
254  if (direct) {
255  vsTransformPrepare(td, &inframe, &inframe);
256  } else { // separate frames
257  VSFrame outframe;
258  for (plane = 0; plane < vsTransformGetDestFrameInfo(td)->planes; plane++) {
259  outframe.data[plane] = out->data[plane];
260  outframe.linesize[plane] = out->linesize[plane];
261  }
262  vsTransformPrepare(td, &inframe, &outframe);
263  }
264 
265  vsDoTransform(td, vsGetNextTransform(td, &tc->trans));
266 
267  vsTransformFinish(td);
268 
269  if (!direct)
270  av_frame_free(&in);
271 
272  return ff_filter_frame(outlink, out);
273 }
274 
276  {
277  .name = "default",
278  .type = AVMEDIA_TYPE_VIDEO,
279  .filter_frame = filter_frame,
280  .config_props = config_input,
281  },
282 };
283 
285  {
286  .name = "default",
287  .type = AVMEDIA_TYPE_VIDEO,
288  },
289 };
290 
292  .name = "vidstabtransform",
293  .description = NULL_IF_CONFIG_SMALL("Transform the frames, "
294  "pass 2 of 2 for stabilization "
295  "(see vidstabdetect for pass 1)."),
296  .priv_size = sizeof(TransformContext),
297  .init = init,
298  .uninit = uninit,
302  .priv_class = &vidstabtransform_class,
303 };
ff_get_video_buffer
AVFrame * ff_get_video_buffer(AVFilterLink *link, int w, int h)
Request a picture buffer with a specific set of permissions.
Definition: video.c:98
direct
static void direct(const float *in, const FFTComplex *ir, int len, float *out)
Definition: af_afir.c:61
td
#define td
Definition: regdef.h:70
uninit
static av_cold void uninit(AVFilterContext *ctx)
Definition: vf_vidstabtransform.c:116
AVERROR
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
opt.h
vidstabutils.h
out
FILE * out
Definition: movenc.c:54
ff_filter_frame
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1018
av_pix_fmt_desc_get
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:2564
FILTER_PIXFMTS_ARRAY
#define FILTER_PIXFMTS_ARRAY(array)
Definition: internal.h:171
inlink
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
Definition: filter_design.txt:212
zoom
static void zoom(float *u, float *v, float amount)
Definition: vf_xfade.c:1629
OFFSETC
#define OFFSETC(x)
Definition: vf_vidstabtransform.c:46
av_frame_free
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:112
config_input
static int config_input(AVFilterLink *inlink)
Definition: vf_vidstabtransform.c:124
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:303
AVOption
AVOption.
Definition: opt.h:247
AV_LOG_VERBOSE
#define AV_LOG_VERBOSE
Detailed information.
Definition: log.h:196
TransformContext
Definition: vf_vidstabtransform.c:33
av_get_bits_per_pixel
int av_get_bits_per_pixel(const AVPixFmtDescriptor *pixdesc)
Return the number of bits per pixel used by the pixel format described by pixdesc.
Definition: pixdesc.c:2516
AVFilter::name
const char * name
Filter name.
Definition: avfilter.h:153
AVFILTER_DEFINE_CLASS
AVFILTER_DEFINE_CLASS(vidstabtransform)
TransformContext::input
char * input
Definition: vf_vidstabtransform.c:40
AVFrame::data
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:317
avfilter_vf_vidstabtransform_inputs
static const AVFilterPad avfilter_vf_vidstabtransform_inputs[]
Definition: vf_vidstabtransform.c:275
init
static av_cold int init(AVFilterContext *ctx)
Definition: vf_vidstabtransform.c:107
DEFAULT_INPUT_NAME
#define DEFAULT_INPUT_NAME
Definition: vf_vidstabtransform.c:21
AVFilterPad
A filter pad used for either input or output.
Definition: internal.h:50
filter_frame
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
Definition: vf_vidstabtransform.c:226
AV_LOG_ERROR
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:180
av_cold
#define av_cold
Definition: attributes.h:90
avfilter_vf_vidstabtransform_outputs
static const AVFilterPad avfilter_vf_vidstabtransform_outputs[]
Definition: vf_vidstabtransform.c:284
TransformContext::conf
VSTransformConfig conf
Definition: vf_vidstabtransform.c:37
ff_av2vs_pixfmt
VSPixelFormat ff_av2vs_pixfmt(AVFilterContext *ctx, enum AVPixelFormat pf)
convert AV's pixelformat to vid.stab pixelformat
Definition: vidstabutils.c:32
AV_OPT_TYPE_DOUBLE
@ AV_OPT_TYPE_DOUBLE
Definition: opt.h:226
ctx
AVFormatContext * ctx
Definition: movenc.c:48
ff_vidstab_pix_fmts
enum AVPixelFormat ff_vidstab_pix_fmts[]
Definition: vidstabutils.c:23
f
#define f(width, name)
Definition: cbs_vp9.c:255
FILTER_INPUTS
#define FILTER_INPUTS(array)
Definition: internal.h:191
OFFSET
#define OFFSET(x)
Definition: vf_vidstabtransform.c:45
AVClass
Describe the class of an AVClass context structure.
Definition: log.h:66
ff_vf_vidstabtransform
const AVFilter ff_vf_vidstabtransform
Definition: vf_vidstabtransform.c:291
NULL
#define NULL
Definition: coverity.c:32
av_frame_copy_props
int av_frame_copy_props(AVFrame *dst, const AVFrame *src)
Copy only "metadata" fields from src to dst.
Definition: frame.c:537
NULL_IF_CONFIG_SMALL
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:117
TransformContext::td
VSTransformData td
Definition: vf_vidstabtransform.c:36
av_frame_is_writable
int av_frame_is_writable(AVFrame *frame)
Check if the frame data is writable.
Definition: frame.c:473
input
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
Definition: filter_design.txt:172
AV_LOG_INFO
#define AV_LOG_INFO
Standard information.
Definition: log.h:191
internal.h
FLAGS
#define FLAGS
Definition: vf_vidstabtransform.c:47
TransformContext::debug
int debug
Definition: vf_vidstabtransform.c:42
invert
static void invert(float *h, int n)
Definition: asrc_sinc.c:195
vidstabtransform_options
static const AVOption vidstabtransform_options[]
Definition: vf_vidstabtransform.c:49
common.h
TransformContext::trans
VSTransformations trans
Definition: vf_vidstabtransform.c:39
AVFilterPad::name
const char * name
Pad name.
Definition: internal.h:56
AVFilter
Filter definition.
Definition: avfilter.h:149
ret
ret
Definition: filter_design.txt:187
ff_vs_init
void ff_vs_init(void)
sets the memory allocation function and logging constants to av versions
Definition: vidstabutils.c:77
AV_OPT_TYPE_INT
@ AV_OPT_TYPE_INT
Definition: opt.h:224
avfilter.h
AV_PIX_FMT_FLAG_PLANAR
#define AV_PIX_FMT_FLAG_PLANAR
At least one pixel component is not in the first data plane.
Definition: pixdesc.h:132
AVFilterContext
An instance of a filter.
Definition: avfilter.h:386
tc
#define tc
Definition: regdef.h:69
desc
const char * desc
Definition: libsvtav1.c:79
AVMEDIA_TYPE_VIDEO
@ AVMEDIA_TYPE_VIDEO
Definition: avutil.h:201
AVPixFmtDescriptor
Descriptor that unambiguously describes how the bits of a pixel are stored in the up to 4 data planes...
Definition: pixdesc.h:69
AV_OPT_TYPE_BOOL
@ AV_OPT_TYPE_BOOL
Definition: opt.h:241
FILTER_OUTPUTS
#define FILTER_OUTPUTS(array)
Definition: internal.h:192
imgutils.h
AVFrame::linesize
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
Definition: frame.h:334
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:28
AV_OPT_TYPE_STRING
@ AV_OPT_TYPE_STRING
Definition: opt.h:228
AV_OPT_TYPE_CONST
@ AV_OPT_TYPE_CONST
Definition: opt.h:233
TransformContext::tripod
int tripod
Definition: vf_vidstabtransform.c:41