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/file_open.h"
27 #include "libavutil/opt.h"
28 #include "libavutil/pixdesc.h"
29 #include "avfilter.h"
30 #include "internal.h"
31 #include "video.h"
32 
33 #include "vidstabutils.h"
34 
35 typedef struct TransformContext {
36  const AVClass *class;
37 
38  VSTransformData td;
39  VSTransformConfig conf;
40 
41  VSTransformations trans; // transformations
42  char *input; // name of transform file
43  int tripod;
44  int debug;
46 
47 #define OFFSET(x) offsetof(TransformContext, x)
48 #define OFFSETC(x) (offsetof(TransformContext, conf)+offsetof(VSTransformConfig, x))
49 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM|AV_OPT_FLAG_VIDEO_PARAM
50 
52  {"input", "set path to the file storing the transforms", OFFSET(input),
53  AV_OPT_TYPE_STRING, {.str = DEFAULT_INPUT_NAME}, .flags = FLAGS },
54  {"smoothing", "set number of frames*2 + 1 used for lowpass filtering", OFFSETC(smoothing),
55  AV_OPT_TYPE_INT, {.i64 = 15}, 0, 1000, FLAGS},
56 
57  {"optalgo", "set camera path optimization algo", OFFSETC(camPathAlgo),
58  AV_OPT_TYPE_INT, {.i64 = VSOptimalL1}, VSOptimalL1, VSAvg, FLAGS, .unit = "optalgo"},
59  { "opt", "global optimization", 0, // from version 1.0 on
60  AV_OPT_TYPE_CONST, {.i64 = VSOptimalL1 }, 0, 0, FLAGS, .unit = "optalgo"},
61  { "gauss", "gaussian kernel", 0,
62  AV_OPT_TYPE_CONST, {.i64 = VSGaussian }, 0, 0, FLAGS, .unit = "optalgo"},
63  { "avg", "simple averaging on motion", 0,
64  AV_OPT_TYPE_CONST, {.i64 = VSAvg }, 0, 0, FLAGS, .unit = "optalgo"},
65 
66  {"maxshift", "set maximal number of pixels to translate image", OFFSETC(maxShift),
67  AV_OPT_TYPE_INT, {.i64 = -1}, -1, 500, FLAGS},
68  {"maxangle", "set maximal angle in rad to rotate image", OFFSETC(maxAngle),
69  AV_OPT_TYPE_DOUBLE, {.dbl = -1.0}, -1.0, 3.14, FLAGS},
70 
71  {"crop", "set cropping mode", OFFSETC(crop),
72  AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, FLAGS, .unit = "crop"},
73  { "keep", "keep border", 0,
74  AV_OPT_TYPE_CONST, {.i64 = VSKeepBorder }, 0, 0, FLAGS, .unit = "crop"},
75  { "black", "black border", 0,
76  AV_OPT_TYPE_CONST, {.i64 = VSCropBorder }, 0, 0, FLAGS, .unit = "crop"},
77 
78  {"invert", "invert transforms", OFFSETC(invert),
79  AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, FLAGS},
80  {"relative", "consider transforms as relative", OFFSETC(relative),
81  AV_OPT_TYPE_INT, {.i64 = 1}, 0, 1, FLAGS},
82  {"zoom", "set percentage to zoom (>0: zoom in, <0: zoom out", OFFSETC(zoom),
83  AV_OPT_TYPE_DOUBLE, {.dbl = 0}, -100, 100, FLAGS},
84  {"optzoom", "set optimal zoom (0: nothing, 1: optimal static zoom, 2: optimal dynamic zoom)", OFFSETC(optZoom),
85  AV_OPT_TYPE_INT, {.i64 = 1}, 0, 2, FLAGS},
86  {"zoomspeed", "for adative zoom: percent to zoom maximally each frame", OFFSETC(zoomSpeed),
87  AV_OPT_TYPE_DOUBLE, {.dbl = 0.25}, 0, 5, FLAGS},
88 
89  {"interpol", "set type of interpolation", OFFSETC(interpolType),
90  AV_OPT_TYPE_INT, {.i64 = 2}, 0, 3, FLAGS, .unit = "interpol"},
91  { "no", "no interpolation", 0,
92  AV_OPT_TYPE_CONST, {.i64 = VS_Zero }, 0, 0, FLAGS, .unit = "interpol"},
93  { "linear", "linear (horizontal)", 0,
94  AV_OPT_TYPE_CONST, {.i64 = VS_Linear }, 0, 0, FLAGS, .unit = "interpol"},
95  { "bilinear","bi-linear", 0,
96  AV_OPT_TYPE_CONST, {.i64 = VS_BiLinear},0, 0, FLAGS, .unit = "interpol"},
97  { "bicubic", "bi-cubic", 0,
98  AV_OPT_TYPE_CONST, {.i64 = VS_BiCubic },0, 0, FLAGS, .unit = "interpol"},
99 
100  {"tripod", "enable virtual tripod mode (same as relative=0:smoothing=0)", OFFSET(tripod),
101  AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS},
102  {"debug", "enable debug mode and writer global motions information to file", OFFSET(debug),
103  AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, FLAGS},
104  {NULL}
105 };
106 
107 AVFILTER_DEFINE_CLASS(vidstabtransform);
108 
110 {
111  TransformContext *tc = ctx->priv;
112  ff_vs_init();
113  tc->class = &vidstabtransform_class;
114  av_log(ctx, AV_LOG_VERBOSE, "vidstabtransform filter: init %s\n", LIBVIDSTAB_VERSION);
115  return 0;
116 }
117 
119 {
120  TransformContext *tc = ctx->priv;
121 
122  vsTransformDataCleanup(&tc->td);
123  vsTransformationsCleanup(&tc->trans);
124 }
125 
127 {
128  AVFilterContext *ctx = inlink->dst;
129  TransformContext *tc = ctx->priv;
130  FILE *f;
131 
133  int is_planar = desc->flags & AV_PIX_FMT_FLAG_PLANAR;
134 
135  VSTransformData *td = &(tc->td);
136 
137  VSFrameInfo fi_src;
138  VSFrameInfo fi_dest;
139 
140  if (!vsFrameInfoInit(&fi_src, inlink->w, inlink->h,
141  ff_av2vs_pixfmt(ctx, inlink->format)) ||
142  !vsFrameInfoInit(&fi_dest, inlink->w, inlink->h,
143  ff_av2vs_pixfmt(ctx, inlink->format))) {
144  av_log(ctx, AV_LOG_ERROR, "unknown pixel format: %i (%s)",
145  inlink->format, desc->name);
146  return AVERROR(EINVAL);
147  }
148 
149  if ((!is_planar && fi_src.bytesPerPixel != av_get_bits_per_pixel(desc)/8) ||
150  fi_src.log2ChromaW != desc->log2_chroma_w ||
151  fi_src.log2ChromaH != desc->log2_chroma_h) {
152  av_log(ctx, AV_LOG_ERROR, "pixel-format error: bpp %i<>%i ",
153  fi_src.bytesPerPixel, av_get_bits_per_pixel(desc)/8);
154  av_log(ctx, AV_LOG_ERROR, "chroma_subsampl: w: %i<>%i h: %i<>%i\n",
155  fi_src.log2ChromaW, desc->log2_chroma_w,
156  fi_src.log2ChromaH, desc->log2_chroma_h);
157  return AVERROR(EINVAL);
158  }
159 
160  // set values that are not initializes by the options
161  tc->conf.modName = "vidstabtransform";
162  tc->conf.verbose = 1 + tc->debug;
163  if (tc->tripod) {
164  av_log(ctx, AV_LOG_INFO, "Virtual tripod mode: relative=0, smoothing=0\n");
165  tc->conf.relative = 0;
166  tc->conf.smoothing = 0;
167  }
168  tc->conf.simpleMotionCalculation = 0;
169  tc->conf.storeTransforms = tc->debug;
170  tc->conf.smoothZoom = 0;
171 
172  if (vsTransformDataInit(td, &tc->conf, &fi_src, &fi_dest) != VS_OK) {
173  av_log(ctx, AV_LOG_ERROR, "initialization of vid.stab transform failed, please report a BUG\n");
174  return AVERROR(EINVAL);
175  }
176 
177  vsTransformGetConfig(&tc->conf, td);
178  av_log(ctx, AV_LOG_INFO, "Video transformation/stabilization settings (pass 2/2):\n");
179  av_log(ctx, AV_LOG_INFO, " input = %s\n", tc->input);
180  av_log(ctx, AV_LOG_INFO, " smoothing = %d\n", tc->conf.smoothing);
181  av_log(ctx, AV_LOG_INFO, " optalgo = %s\n",
182  tc->conf.camPathAlgo == VSOptimalL1 ? "opt" :
183  (tc->conf.camPathAlgo == VSGaussian ? "gauss" : "avg"));
184  av_log(ctx, AV_LOG_INFO, " maxshift = %d\n", tc->conf.maxShift);
185  av_log(ctx, AV_LOG_INFO, " maxangle = %f\n", tc->conf.maxAngle);
186  av_log(ctx, AV_LOG_INFO, " crop = %s\n", tc->conf.crop ? "Black" : "Keep");
187  av_log(ctx, AV_LOG_INFO, " relative = %s\n", tc->conf.relative ? "True": "False");
188  av_log(ctx, AV_LOG_INFO, " invert = %s\n", tc->conf.invert ? "True" : "False");
189  av_log(ctx, AV_LOG_INFO, " zoom = %f\n", tc->conf.zoom);
190  av_log(ctx, AV_LOG_INFO, " optzoom = %s\n",
191  tc->conf.optZoom == 1 ? "Static (1)" : (tc->conf.optZoom == 2 ? "Dynamic (2)" : "Off (0)"));
192  if (tc->conf.optZoom == 2)
193  av_log(ctx, AV_LOG_INFO, " zoomspeed = %g\n", tc->conf.zoomSpeed);
194  av_log(ctx, AV_LOG_INFO, " interpol = %s\n", getInterpolationTypeName(tc->conf.interpolType));
195 
196  f = avpriv_fopen_utf8(tc->input, "rb");
197  if (!f) {
198  int ret = AVERROR(errno);
199  av_log(ctx, AV_LOG_ERROR, "cannot open input file %s\n", tc->input);
200  return ret;
201  } else {
202  VSManyLocalMotions mlms;
203  if (vsReadLocalMotionsFile(f, &mlms) == VS_OK) {
204  // calculate the actual transforms from the local motions
205  if (vsLocalmotions2Transforms(td, &mlms, &tc->trans) != VS_OK) {
206  av_log(ctx, AV_LOG_ERROR, "calculating transformations failed\n");
207  return AVERROR(EINVAL);
208  }
209  } else { // try to read old format
210  if (!vsReadOldTransforms(td, f, &tc->trans)) { /* read input file */
211  av_log(ctx, AV_LOG_ERROR, "error parsing input file %s\n", tc->input);
212  return AVERROR(EINVAL);
213  }
214  }
215  }
216  fclose(f);
217 
218  if (vsPreprocessTransforms(td, &tc->trans) != VS_OK) {
219  av_log(ctx, AV_LOG_ERROR, "error while preprocessing transforms\n");
220  return AVERROR(EINVAL);
221  }
222 
223  // TODO: add sharpening, so far the user needs to call the unsharp filter manually
224  return 0;
225 }
226 
227 
229 {
230  AVFilterContext *ctx = inlink->dst;
231  TransformContext *tc = ctx->priv;
232  VSTransformData* td = &(tc->td);
233 
234  AVFilterLink *outlink = inlink->dst->outputs[0];
235  int direct = 0;
236  AVFrame *out;
237  VSFrame inframe;
238  int plane;
239 
240  if (av_frame_is_writable(in)) {
241  direct = 1;
242  out = in;
243  } else {
244  out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
245  if (!out) {
246  av_frame_free(&in);
247  return AVERROR(ENOMEM);
248  }
250  }
251 
252  for (plane = 0; plane < vsTransformGetSrcFrameInfo(td)->planes; plane++) {
253  inframe.data[plane] = in->data[plane];
254  inframe.linesize[plane] = in->linesize[plane];
255  }
256  if (direct) {
257  vsTransformPrepare(td, &inframe, &inframe);
258  } else { // separate frames
259  VSFrame outframe;
260  for (plane = 0; plane < vsTransformGetDestFrameInfo(td)->planes; plane++) {
261  outframe.data[plane] = out->data[plane];
262  outframe.linesize[plane] = out->linesize[plane];
263  }
264  vsTransformPrepare(td, &inframe, &outframe);
265  }
266 
267  vsDoTransform(td, vsGetNextTransform(td, &tc->trans));
268 
269  vsTransformFinish(td);
270 
271  if (!direct)
272  av_frame_free(&in);
273 
274  return ff_filter_frame(outlink, out);
275 }
276 
278  {
279  .name = "default",
280  .type = AVMEDIA_TYPE_VIDEO,
281  .filter_frame = filter_frame,
282  .config_props = config_input,
283  },
284 };
285 
287  .name = "vidstabtransform",
288  .description = NULL_IF_CONFIG_SMALL("Transform the frames, "
289  "pass 2 of 2 for stabilization "
290  "(see vidstabdetect for pass 1)."),
291  .priv_size = sizeof(TransformContext),
292  .init = init,
293  .uninit = uninit,
297  .priv_class = &vidstabtransform_class,
298 };
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:112
td
#define td
Definition: regdef.h:70
uninit
static av_cold void uninit(AVFilterContext *ctx)
Definition: vf_vidstabtransform.c:118
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:2962
FILTER_PIXFMTS_ARRAY
#define FILTER_PIXFMTS_ARRAY(array)
Definition: internal.h:162
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:1696
OFFSETC
#define OFFSETC(x)
Definition: vf_vidstabtransform.c:48
av_frame_free
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:160
config_input
static int config_input(AVFilterLink *inlink)
Definition: vf_vidstabtransform.c:126
AVFrame
This structure describes decoded (raw) audio or video data.
Definition: frame.h:375
pixdesc.h
AVOption
AVOption.
Definition: opt.h:346
AV_LOG_VERBOSE
#define AV_LOG_VERBOSE
Detailed information.
Definition: log.h:196
TransformContext
Definition: vf_vidstabtransform.c:35
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:2914
AVFilter::name
const char * name
Filter name.
Definition: avfilter.h:170
video.h
AVFILTER_DEFINE_CLASS
AVFILTER_DEFINE_CLASS(vidstabtransform)
TransformContext::input
char * input
Definition: vf_vidstabtransform.c:42
AVFrame::data
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:396
avfilter_vf_vidstabtransform_inputs
static const AVFilterPad avfilter_vf_vidstabtransform_inputs[]
Definition: vf_vidstabtransform.c:277
init
static av_cold int init(AVFilterContext *ctx)
Definition: vf_vidstabtransform.c:109
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:33
filter_frame
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
Definition: vf_vidstabtransform.c:228
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
ff_video_default_filterpad
const AVFilterPad ff_video_default_filterpad[1]
An AVFilterPad array whose only entry has name "default" and is of type AVMEDIA_TYPE_VIDEO.
Definition: video.c:37
TransformContext::conf
VSTransformConfig conf
Definition: vf_vidstabtransform.c:39
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:237
ctx
AVFormatContext * ctx
Definition: movenc.c:48
ff_vidstab_pix_fmts
enum AVPixelFormat ff_vidstab_pix_fmts[]
Definition: vidstabutils.c:23
FILTER_INPUTS
#define FILTER_INPUTS(array)
Definition: internal.h:182
file_open.h
OFFSET
#define OFFSET(x)
Definition: vf_vidstabtransform.c:47
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:286
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:709
f
f
Definition: af_crystalizer.c:121
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:106
TransformContext::td
VSTransformData td
Definition: vf_vidstabtransform.c:38
av_frame_is_writable
int av_frame_is_writable(AVFrame *frame)
Check if the frame data is writable.
Definition: frame.c:645
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:49
TransformContext::debug
int debug
Definition: vf_vidstabtransform.c:44
invert
static void invert(float *h, int n)
Definition: asrc_sinc.c:184
vidstabtransform_options
static const AVOption vidstabtransform_options[]
Definition: vf_vidstabtransform.c:51
common.h
TransformContext::trans
VSTransformations trans
Definition: vf_vidstabtransform.c:41
AVFilterPad::name
const char * name
Pad name.
Definition: internal.h:39
avpriv_fopen_utf8
FILE * avpriv_fopen_utf8(const char *path, const char *mode)
Open a file using a UTF-8 filename.
Definition: file_open.c:159
AVFilter
Filter definition.
Definition: avfilter.h:166
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:235
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:407
tc
#define tc
Definition: regdef.h:69
desc
const char * desc
Definition: libsvtav1.c:75
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:251
FILTER_OUTPUTS
#define FILTER_OUTPUTS(array)
Definition: internal.h:183
AVFrame::linesize
int linesize[AV_NUM_DATA_POINTERS]
For video, a positive or negative value, which is typically indicating the size in bytes of each pict...
Definition: frame.h:420
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:27
AV_OPT_TYPE_STRING
@ AV_OPT_TYPE_STRING
Definition: opt.h:239
AV_OPT_TYPE_CONST
@ AV_OPT_TYPE_CONST
Definition: opt.h:244
TransformContext::tripod
int tripod
Definition: vf_vidstabtransform.c:43