The issue was caused by a prediction algorithm detecting tracking the wrong way. Solved by passing tracking direction explicitly, so that prediction will always happen correctly regardless of the state of the Tracks context.
839 lines
31 KiB
C
839 lines
31 KiB
C
/*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License
|
|
* as published by the Free Software Foundation; either version 2
|
|
* of the License, or (at your option) any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software Foundation,
|
|
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
|
|
*
|
|
* The Original Code is Copyright (C) 2011 Blender Foundation.
|
|
* All rights reserved.
|
|
*/
|
|
|
|
/** \file
|
|
* \ingroup bke
|
|
*/
|
|
|
|
#include "atomic_ops.h"
|
|
#include <stdlib.h>
|
|
|
|
#include "MEM_guardedalloc.h"
|
|
|
|
#include "DNA_movieclip_types.h"
|
|
#include "DNA_object_types.h" /* SELECT */
|
|
|
|
#include "BLI_listbase.h"
|
|
#include "BLI_math.h"
|
|
#include "BLI_task.h"
|
|
#include "BLI_threads.h"
|
|
#include "BLI_utildefines.h"
|
|
#include "PIL_time.h"
|
|
|
|
#include "BKE_global.h"
|
|
#include "BKE_movieclip.h"
|
|
#include "BKE_tracking.h"
|
|
|
|
#include "libmv-capi.h"
|
|
#include "tracking_private.h"
|
|
|
|
typedef struct AutoTrackClip {
|
|
MovieClip *clip;
|
|
|
|
/* Dimensions of movie frame, in pixels.
|
|
*
|
|
* NOTE: All frames within a clip are expected to have match3ed dimensions. */
|
|
int width, height;
|
|
} AutoTrackClip;
|
|
|
|
typedef struct AutoTrackTrack {
|
|
/* Index of a clip from `AutoTrackContext::autotrack_clips` this track belongs to. */
|
|
int clip_index;
|
|
|
|
MovieTrackingTrack *track;
|
|
|
|
/* Options for the region tracker. */
|
|
libmv_TrackRegionOptions track_region_options;
|
|
|
|
/* Denotes whether this track will be tracked.
|
|
* Is usually initialized based on track's selection. Non-trackable tracks are still added to the
|
|
* context to provide AutoTrack all knowledge about what is going on in the scene. */
|
|
bool is_trackable;
|
|
} AutoTrackTrack;
|
|
|
|
typedef struct AutoTrackMarker {
|
|
libmv_Marker libmv_marker;
|
|
} AutoTrackMarker;
|
|
|
|
/* Result of tracking step for a single marker.
|
|
*
|
|
* On success both marker and result are fully initialized to the position on the new frame.
|
|
*
|
|
* On failure marker's frame number is initialized to frame number where it was attempted to be
|
|
* tracked to. The position and other fields of tracked marker are the same as the input. */
|
|
typedef struct AutoTrackTrackingResult {
|
|
struct AutoTrackTrackingResult *next, *prev;
|
|
|
|
bool success;
|
|
libmv_Marker libmv_marker;
|
|
libmv_TrackRegionResult libmv_result;
|
|
} AutoTrackTrackingResult;
|
|
|
|
typedef struct AutoTrackContext {
|
|
/* --------------------------------------------------------------------
|
|
* Invariant part.
|
|
* Stays unchanged during the tracking process.
|
|
* If not the initialization process, all the fields here should be treated as `const`.
|
|
*/
|
|
|
|
/* Frame at which tracking process started.
|
|
* NOTE: Measured in scene time frames, */
|
|
int start_scene_frame;
|
|
|
|
/* True when tracking backwards (from higher frame number to lower frame number.) */
|
|
bool is_backwards;
|
|
|
|
/* Movie clips used during the tracking process. */
|
|
int num_clips;
|
|
AutoTrackClip autotrack_clips[MAX_ACCESSOR_CLIP];
|
|
|
|
/* Tracks for which the context has been created for.
|
|
* This is a flat array of all tracks coming from all clips, regardless of whether track is
|
|
* actually being tracked or not. This allows the AutoTrack to see a big picture of hat is going
|
|
* on in the scene, and request information it needs.
|
|
*
|
|
* Indexed by AutoTrackOptions::track_index. */
|
|
int num_all_tracks;
|
|
AutoTrackTrack *all_autotrack_tracks;
|
|
|
|
/* Accessor for images of clip. Used by the autotrack context. */
|
|
TrackingImageAccessor *image_accessor;
|
|
|
|
/* --------------------------------------------------------------------
|
|
* Variant part.
|
|
* Denotes tracing state and tracking result.
|
|
*/
|
|
|
|
/* Auto-track context.
|
|
*
|
|
* NOTE: Is accessed from multiple threads at once. */
|
|
struct libmv_AutoTrack *autotrack;
|
|
|
|
/* Markers from the current frame which will be tracked to the next frame upon the tracking
|
|
* context step.
|
|
*
|
|
* NOTE: This array is re-used across tracking steps, which might make it appear that the array
|
|
* is over-allocated when some tracks has failed to track. */
|
|
int num_autotrack_markers;
|
|
AutoTrackMarker *autotrack_markers;
|
|
|
|
/* Tracking results which are to be synchronized from the AutoTrack context to the Blender's
|
|
* DNA to make the results visible for users. */
|
|
ListBase results_to_sync;
|
|
int synchronized_scene_frame;
|
|
|
|
SpinLock spin_lock;
|
|
} AutoTrackContext;
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Marker coordinate system conversion.
|
|
* \{ */
|
|
|
|
static void normalized_to_libmv_frame(const float normalized[2],
|
|
const int frame_dimensions[2],
|
|
float result[2])
|
|
{
|
|
result[0] = normalized[0] * frame_dimensions[0] - 0.5f;
|
|
result[1] = normalized[1] * frame_dimensions[1] - 0.5f;
|
|
}
|
|
|
|
static void normalized_relative_to_libmv_frame(const float normalized[2],
|
|
const float origin[2],
|
|
const int frame_dimensions[2],
|
|
float result[2])
|
|
{
|
|
result[0] = (normalized[0] + origin[0]) * frame_dimensions[0] - 0.5f;
|
|
result[1] = (normalized[1] + origin[1]) * frame_dimensions[1] - 0.5f;
|
|
}
|
|
|
|
static void libmv_frame_to_normalized(const float frame_coord[2],
|
|
const int frame_dimensions[2],
|
|
float result[2])
|
|
{
|
|
result[0] = (frame_coord[0] + 0.5f) / frame_dimensions[0];
|
|
result[1] = (frame_coord[1] + 0.5f) / frame_dimensions[1];
|
|
}
|
|
|
|
static void libmv_frame_to_normalized_relative(const float frame_coord[2],
|
|
const float origin[2],
|
|
const int frame_dimensions[2],
|
|
float result[2])
|
|
{
|
|
result[0] = (frame_coord[0] - origin[0]) / frame_dimensions[0];
|
|
result[1] = (frame_coord[1] - origin[1]) / frame_dimensions[1];
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Conversion of markers between Blender's DNA and Libmv.
|
|
* \{ */
|
|
|
|
static void dna_marker_to_libmv_marker(/*const*/ MovieTrackingTrack *track,
|
|
const MovieTrackingMarker *marker,
|
|
int clip,
|
|
int track_index,
|
|
int frame_width,
|
|
int frame_height,
|
|
bool backwards,
|
|
libmv_Marker *libmv_marker)
|
|
{
|
|
const int frame_dimensions[2] = {frame_width, frame_height};
|
|
libmv_marker->clip = clip;
|
|
libmv_marker->frame = marker->framenr;
|
|
libmv_marker->track = track_index;
|
|
|
|
normalized_to_libmv_frame(marker->pos, frame_dimensions, libmv_marker->center);
|
|
for (int i = 0; i < 4; i++) {
|
|
normalized_relative_to_libmv_frame(
|
|
marker->pattern_corners[i], marker->pos, frame_dimensions, libmv_marker->patch[i]);
|
|
}
|
|
|
|
normalized_relative_to_libmv_frame(
|
|
marker->search_min, marker->pos, frame_dimensions, libmv_marker->search_region_min);
|
|
|
|
normalized_relative_to_libmv_frame(
|
|
marker->search_max, marker->pos, frame_dimensions, libmv_marker->search_region_max);
|
|
|
|
/* NOTE: All the markers does have 1.0 weight.
|
|
* Might support in the future, but will require more elaborated process which will involve
|
|
* F-Curve evaluation. */
|
|
libmv_marker->weight = 1.0f;
|
|
|
|
if (marker->flag & MARKER_TRACKED) {
|
|
libmv_marker->source = LIBMV_MARKER_SOURCE_TRACKED;
|
|
}
|
|
else {
|
|
libmv_marker->source = LIBMV_MARKER_SOURCE_MANUAL;
|
|
}
|
|
libmv_marker->status = LIBMV_MARKER_STATUS_UNKNOWN;
|
|
libmv_marker->model_type = LIBMV_MARKER_MODEL_TYPE_POINT;
|
|
libmv_marker->model_id = 0;
|
|
|
|
/* NOTE: We currently don't support reference marker from different clip. */
|
|
libmv_marker->reference_clip = clip;
|
|
|
|
if (track->pattern_match == TRACK_MATCH_KEYFRAME) {
|
|
const MovieTrackingMarker *keyframe_marker = tracking_get_keyframed_marker(
|
|
track, marker->framenr, backwards);
|
|
libmv_marker->reference_frame = keyframe_marker->framenr;
|
|
}
|
|
else {
|
|
libmv_marker->reference_frame = backwards ? marker->framenr - 1 : marker->framenr;
|
|
}
|
|
|
|
libmv_marker->disabled_channels =
|
|
((track->flag & TRACK_DISABLE_RED) ? LIBMV_MARKER_CHANNEL_R : 0) |
|
|
((track->flag & TRACK_DISABLE_GREEN) ? LIBMV_MARKER_CHANNEL_G : 0) |
|
|
((track->flag & TRACK_DISABLE_BLUE) ? LIBMV_MARKER_CHANNEL_B : 0);
|
|
}
|
|
|
|
static void libmv_marker_to_dna_marker(libmv_Marker *libmv_marker,
|
|
int frame_width,
|
|
int frame_height,
|
|
MovieTrackingMarker *marker)
|
|
{
|
|
const int frame_dimensions[2] = {frame_width, frame_height};
|
|
marker->framenr = libmv_marker->frame;
|
|
|
|
libmv_frame_to_normalized(libmv_marker->center, frame_dimensions, marker->pos);
|
|
for (int i = 0; i < 4; i++) {
|
|
libmv_frame_to_normalized_relative(libmv_marker->patch[i],
|
|
libmv_marker->center,
|
|
frame_dimensions,
|
|
marker->pattern_corners[i]);
|
|
}
|
|
|
|
libmv_frame_to_normalized_relative(
|
|
libmv_marker->search_region_min, libmv_marker->center, frame_dimensions, marker->search_min);
|
|
|
|
libmv_frame_to_normalized_relative(
|
|
libmv_marker->search_region_max, libmv_marker->center, frame_dimensions, marker->search_max);
|
|
|
|
marker->flag = 0;
|
|
if (libmv_marker->source == LIBMV_MARKER_SOURCE_TRACKED) {
|
|
marker->flag |= MARKER_TRACKED;
|
|
}
|
|
else {
|
|
marker->flag &= ~MARKER_TRACKED;
|
|
}
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name General helpers.
|
|
*
|
|
* TODO(sergey): Should be moved to tracking_util.c
|
|
*
|
|
* \{ */
|
|
|
|
/* Returns false if marker crossed margin area from frame bounds. */
|
|
static bool tracking_check_marker_margin(const libmv_Marker *libmv_marker,
|
|
int margin,
|
|
int frame_width,
|
|
int frame_height)
|
|
{
|
|
float patch_min[2], patch_max[2];
|
|
float margin_left, margin_top, margin_right, margin_bottom;
|
|
|
|
INIT_MINMAX2(patch_min, patch_max);
|
|
minmax_v2v2_v2(patch_min, patch_max, libmv_marker->patch[0]);
|
|
minmax_v2v2_v2(patch_min, patch_max, libmv_marker->patch[1]);
|
|
minmax_v2v2_v2(patch_min, patch_max, libmv_marker->patch[2]);
|
|
minmax_v2v2_v2(patch_min, patch_max, libmv_marker->patch[3]);
|
|
|
|
margin_left = max_ff(libmv_marker->center[0] - patch_min[0], margin);
|
|
margin_top = max_ff(patch_max[1] - libmv_marker->center[1], margin);
|
|
margin_right = max_ff(patch_max[0] - libmv_marker->center[0], margin);
|
|
margin_bottom = max_ff(libmv_marker->center[1] - patch_min[1], margin);
|
|
|
|
if (libmv_marker->center[0] < margin_left ||
|
|
libmv_marker->center[0] > frame_width - margin_right ||
|
|
libmv_marker->center[1] < margin_bottom ||
|
|
libmv_marker->center[1] > frame_height - margin_top) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Autotrack context initialization.
|
|
* \{ */
|
|
|
|
static bool autotrack_is_marker_usable(const MovieTrackingMarker *marker)
|
|
{
|
|
if ((marker->flag & MARKER_DISABLED)) {
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
static bool autotrack_is_track_trackable(const AutoTrackContext *context,
|
|
const AutoTrackTrack *autotrack_track)
|
|
{
|
|
/*const*/ MovieTrackingTrack *track = autotrack_track->track;
|
|
if (TRACK_SELECTED(track) && (track->flag & (TRACK_LOCKED | TRACK_HIDDEN)) == 0) {
|
|
const AutoTrackClip *autotrack_clip = &context->autotrack_clips[autotrack_track->clip_index];
|
|
MovieClip *clip = autotrack_clip->clip;
|
|
const int clip_frame_number = BKE_movieclip_remap_scene_to_clip_frame(
|
|
clip, context->start_scene_frame);
|
|
|
|
const MovieTrackingMarker *marker = BKE_tracking_marker_get(track, clip_frame_number);
|
|
return autotrack_is_marker_usable(marker);
|
|
}
|
|
return false;
|
|
}
|
|
|
|
static void autotrack_context_init_clips(AutoTrackContext *context,
|
|
MovieClip *clip,
|
|
MovieClipUser *user)
|
|
{
|
|
/* NOTE: Currently only tracking within a single clip. */
|
|
|
|
context->num_clips = 1;
|
|
|
|
context->autotrack_clips[0].clip = clip;
|
|
BKE_movieclip_get_size(
|
|
clip, user, &context->autotrack_clips[0].width, &context->autotrack_clips[0].height);
|
|
}
|
|
|
|
/* Initialize flat list of tracks for quick index-based access for the specified clip.
|
|
* All the tracks from this clip are added at the end of the array of already-collected tracks.
|
|
*
|
|
* NOTE: Clips should be initialized first. */
|
|
static void autotrack_context_init_tracks_for_clip(AutoTrackContext *context, int clip_index)
|
|
{
|
|
BLI_assert(clip_index >= 0);
|
|
BLI_assert(clip_index < context->num_clips);
|
|
|
|
const AutoTrackClip *autotrack_clip = &context->autotrack_clips[clip_index];
|
|
MovieClip *clip = autotrack_clip->clip;
|
|
MovieTracking *tracking = &clip->tracking;
|
|
ListBase *tracks_base = BKE_tracking_get_active_tracks(tracking);
|
|
|
|
const int num_clip_tracks = BLI_listbase_count(tracks_base);
|
|
if (num_clip_tracks == 0) {
|
|
return;
|
|
}
|
|
|
|
context->all_autotrack_tracks = MEM_reallocN(context->all_autotrack_tracks,
|
|
(context->num_all_tracks + num_clip_tracks) *
|
|
sizeof(AutoTrackTrack));
|
|
|
|
LISTBASE_FOREACH (MovieTrackingTrack *, track, tracks_base) {
|
|
AutoTrackTrack *autotrack_track = &context->all_autotrack_tracks[context->num_all_tracks++];
|
|
autotrack_track->clip_index = clip_index;
|
|
autotrack_track->track = track;
|
|
autotrack_track->is_trackable = autotrack_is_track_trackable(context, autotrack_track);
|
|
|
|
tracking_configure_tracker(
|
|
track, NULL, context->is_backwards, &autotrack_track->track_region_options);
|
|
}
|
|
}
|
|
|
|
/* Initialize flat list of tracks for quick index-based access for all clips used for tracking.
|
|
*
|
|
* NOTE: Clips should be initialized first. */
|
|
static void autotrack_context_init_tracks(AutoTrackContext *context)
|
|
{
|
|
BLI_assert(context->num_clips >= 1);
|
|
|
|
for (int clip_index = 0; clip_index < context->num_clips; ++clip_index) {
|
|
autotrack_context_init_tracks_for_clip(context, clip_index);
|
|
}
|
|
}
|
|
|
|
/* NOTE: Clips should be initialized first. */
|
|
static void autotrack_context_init_image_accessor(AutoTrackContext *context)
|
|
{
|
|
BLI_assert(context->num_clips >= 1);
|
|
|
|
/* Planarize arrays of clips and tracks, storing pointers to their base "objects".
|
|
* This allows image accessor to be independent, but adds some overhead here. Could be solved
|
|
* by either more strongly coupling accessor API with the AutoTrack, or by giving some functors
|
|
* to the accessor to access clip/track from their indices. */
|
|
|
|
MovieClip *clips[MAX_ACCESSOR_CLIP];
|
|
for (int i = 0; i < context->num_clips; ++i) {
|
|
clips[i] = context->autotrack_clips[i].clip;
|
|
}
|
|
|
|
MovieTrackingTrack **tracks = MEM_malloc_arrayN(
|
|
context->num_all_tracks, sizeof(MovieTrackingTrack *), "image accessor init tracks");
|
|
for (int i = 0; i < context->num_all_tracks; ++i) {
|
|
tracks[i] = context->all_autotrack_tracks[i].track;
|
|
}
|
|
|
|
context->image_accessor = tracking_image_accessor_new(clips, 1, tracks, context->num_all_tracks);
|
|
|
|
MEM_freeN(tracks);
|
|
}
|
|
|
|
/* Count markers which are usable to be passed to the AutoTrack context. */
|
|
static size_t autotrack_count_all_usable_markers(AutoTrackContext *context)
|
|
{
|
|
size_t num_usable_markers = 0;
|
|
for (int track_index = 0; track_index < context->num_all_tracks; ++track_index) {
|
|
const MovieTrackingTrack *track = context->all_autotrack_tracks[track_index].track;
|
|
for (int marker_index = 0; marker_index < track->markersnr; ++marker_index) {
|
|
const MovieTrackingMarker *marker = &track->markers[marker_index];
|
|
if (!autotrack_is_marker_usable(marker)) {
|
|
continue;
|
|
}
|
|
num_usable_markers++;
|
|
}
|
|
}
|
|
return num_usable_markers;
|
|
}
|
|
|
|
static int autotrack_count_trackable_markers(AutoTrackContext *context)
|
|
{
|
|
int num_trackable_markers = 0;
|
|
for (int track_index = 0; track_index < context->num_all_tracks; ++track_index) {
|
|
const AutoTrackTrack *autotrack_track = &context->all_autotrack_tracks[track_index];
|
|
if (!autotrack_track->is_trackable) {
|
|
continue;
|
|
}
|
|
num_trackable_markers++;
|
|
}
|
|
return num_trackable_markers;
|
|
}
|
|
|
|
/* Provide Libmv side of auto track all information about given tracks.
|
|
* Information from all clips is passed to the auto tracker.
|
|
*
|
|
* NOTE: Clips and all tracks are to be initialized before calling this. */
|
|
static void autotrack_context_init_autotrack(AutoTrackContext *context)
|
|
{
|
|
context->autotrack = libmv_autoTrackNew(context->image_accessor->libmv_accessor);
|
|
|
|
/* Count number of markers to be put to a context. */
|
|
const size_t num_trackable_markers = autotrack_count_all_usable_markers(context);
|
|
if (num_trackable_markers == 0) {
|
|
return;
|
|
}
|
|
|
|
/* Allocate memory for all the markers. */
|
|
libmv_Marker *libmv_markers = MEM_malloc_arrayN(
|
|
sizeof(libmv_Marker), num_trackable_markers, "libmv markers array");
|
|
|
|
/* Fill in markers array. */
|
|
int num_filled_libmv_markers = 0;
|
|
for (int track_index = 0; track_index < context->num_all_tracks; ++track_index) {
|
|
const AutoTrackTrack *autotrack_track = &context->all_autotrack_tracks[track_index];
|
|
/*const*/ MovieTrackingTrack *track = autotrack_track->track;
|
|
for (int marker_index = 0; marker_index < track->markersnr; ++marker_index) {
|
|
/*const*/ MovieTrackingMarker *marker = &track->markers[marker_index];
|
|
if (!autotrack_is_marker_usable(marker)) {
|
|
continue;
|
|
}
|
|
const AutoTrackClip *autotrack_clip = &context->autotrack_clips[autotrack_track->clip_index];
|
|
dna_marker_to_libmv_marker(track,
|
|
marker,
|
|
autotrack_track->clip_index,
|
|
track_index,
|
|
autotrack_clip->width,
|
|
autotrack_clip->height,
|
|
context->is_backwards,
|
|
&libmv_markers[num_filled_libmv_markers++]);
|
|
}
|
|
}
|
|
|
|
/* Add all markers to autotrack. */
|
|
libmv_autoTrackSetMarkers(context->autotrack, libmv_markers, num_trackable_markers);
|
|
|
|
/* Free temporary memory. */
|
|
MEM_freeN(libmv_markers);
|
|
}
|
|
|
|
static void autotrack_context_init_markers(AutoTrackContext *context)
|
|
{
|
|
/* Count number of trackable tracks. */
|
|
context->num_autotrack_markers = autotrack_count_trackable_markers(context);
|
|
if (context->num_autotrack_markers == 0) {
|
|
return;
|
|
}
|
|
|
|
/* Allocate required memory. */
|
|
context->autotrack_markers = MEM_calloc_arrayN(
|
|
sizeof(AutoTrackMarker), context->num_autotrack_markers, "auto track options");
|
|
|
|
/* Fill in all the markers. */
|
|
int autotrack_marker_index = 0;
|
|
for (int track_index = 0; track_index < context->num_all_tracks; ++track_index) {
|
|
const AutoTrackTrack *autotrack_track = &context->all_autotrack_tracks[track_index];
|
|
if (!autotrack_track->is_trackable) {
|
|
continue;
|
|
}
|
|
|
|
const AutoTrackClip *autotrack_clip = &context->autotrack_clips[autotrack_track->clip_index];
|
|
MovieClip *clip = autotrack_clip->clip;
|
|
const int clip_frame_number = BKE_movieclip_remap_scene_to_clip_frame(
|
|
clip, context->start_scene_frame);
|
|
|
|
/*const*/ MovieTrackingTrack *track = context->all_autotrack_tracks[track_index].track;
|
|
const MovieTrackingMarker *marker = BKE_tracking_marker_get(track, clip_frame_number);
|
|
|
|
AutoTrackMarker *autotrack_marker = &context->autotrack_markers[autotrack_marker_index++];
|
|
dna_marker_to_libmv_marker(track,
|
|
marker,
|
|
autotrack_track->clip_index,
|
|
track_index,
|
|
autotrack_clip->width,
|
|
autotrack_clip->height,
|
|
context->is_backwards,
|
|
&autotrack_marker->libmv_marker);
|
|
}
|
|
}
|
|
|
|
AutoTrackContext *BKE_autotrack_context_new(MovieClip *clip,
|
|
MovieClipUser *user,
|
|
const bool is_backwards)
|
|
{
|
|
AutoTrackContext *context = MEM_callocN(sizeof(AutoTrackContext), "autotrack context");
|
|
|
|
context->start_scene_frame = user->framenr;
|
|
context->is_backwards = is_backwards;
|
|
context->synchronized_scene_frame = context->start_scene_frame;
|
|
|
|
autotrack_context_init_clips(context, clip, user);
|
|
autotrack_context_init_tracks(context);
|
|
autotrack_context_init_image_accessor(context);
|
|
autotrack_context_init_autotrack(context);
|
|
autotrack_context_init_markers(context);
|
|
|
|
BLI_spin_init(&context->spin_lock);
|
|
|
|
return context;
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Threaded context step (tracking process).
|
|
* \{ */
|
|
|
|
/* NOTE: This is a TLS in a sense that this struct is never accessed from multiple threads, and
|
|
* that threads are re-using the struct as much as possible. */
|
|
typedef struct AutoTrackTLS {
|
|
ListBase results; /* Elements of `AutoTrackTrackingResult`. */
|
|
} AutoTrackTLS;
|
|
|
|
static void autotrack_context_step_cb(void *__restrict userdata,
|
|
const int marker_index,
|
|
const TaskParallelTLS *__restrict tls)
|
|
{
|
|
AutoTrackContext *context = userdata;
|
|
AutoTrackTLS *autotrack_tls = (AutoTrackTLS *)tls->userdata_chunk;
|
|
|
|
const AutoTrackMarker *autotrack_marker = &context->autotrack_markers[marker_index];
|
|
const libmv_Marker *libmv_current_marker = &autotrack_marker->libmv_marker;
|
|
|
|
const int frame_delta = context->is_backwards ? -1 : 1;
|
|
const int clip_index = libmv_current_marker->clip;
|
|
const int track_index = libmv_current_marker->track;
|
|
|
|
const AutoTrackClip *autotrack_clip = &context->autotrack_clips[clip_index];
|
|
const AutoTrackTrack *autotrack_track = &context->all_autotrack_tracks[track_index];
|
|
const MovieTrackingTrack *track = autotrack_track->track;
|
|
|
|
/* Check whether marker is going outside of allowed frame margin. */
|
|
if (!tracking_check_marker_margin(
|
|
libmv_current_marker, track->margin, autotrack_clip->width, autotrack_clip->height)) {
|
|
return;
|
|
}
|
|
|
|
const int new_marker_frame = libmv_current_marker->frame + frame_delta;
|
|
|
|
AutoTrackTrackingResult *autotrack_result = MEM_mallocN(sizeof(AutoTrackTrackingResult),
|
|
"autotrack result");
|
|
autotrack_result->libmv_marker = *libmv_current_marker;
|
|
autotrack_result->libmv_marker.frame = new_marker_frame;
|
|
|
|
/* Update reference frame. */
|
|
libmv_Marker libmv_reference_marker;
|
|
if (track->pattern_match == TRACK_MATCH_KEYFRAME) {
|
|
autotrack_result->libmv_marker.reference_frame = libmv_current_marker->reference_frame;
|
|
libmv_autoTrackGetMarker(context->autotrack,
|
|
clip_index,
|
|
autotrack_result->libmv_marker.reference_frame,
|
|
track_index,
|
|
&libmv_reference_marker);
|
|
}
|
|
else {
|
|
BLI_assert(track->pattern_match == TRACK_MATCH_PREVIOS_FRAME);
|
|
autotrack_result->libmv_marker.reference_frame = libmv_current_marker->frame;
|
|
libmv_reference_marker = *libmv_current_marker;
|
|
}
|
|
|
|
/* Perform actual tracking. */
|
|
autotrack_result->success = libmv_autoTrackMarker(context->autotrack,
|
|
&autotrack_track->track_region_options,
|
|
&autotrack_result->libmv_marker,
|
|
&autotrack_result->libmv_result);
|
|
|
|
/* If tracking failed restore initial position.
|
|
* This is how Blender side is currently expecting failed track to be handled. Without this the
|
|
* marker is left in an arbitrary position which did not provide good correlation. */
|
|
if (!autotrack_result->success) {
|
|
autotrack_result->libmv_marker = *libmv_current_marker;
|
|
autotrack_result->libmv_marker.frame = new_marker_frame;
|
|
}
|
|
|
|
BLI_addtail(&autotrack_tls->results, autotrack_result);
|
|
}
|
|
|
|
static void autotrack_context_reduce(const void *__restrict UNUSED(userdata),
|
|
void *__restrict chunk_join,
|
|
void *__restrict chunk)
|
|
{
|
|
AutoTrackTLS *autotrack_tls = (AutoTrackTLS *)chunk;
|
|
if (BLI_listbase_is_empty(&autotrack_tls->results)) {
|
|
/* Nothing to be joined from. */
|
|
return;
|
|
}
|
|
|
|
AutoTrackTLS *autotrack_tls_join = (AutoTrackTLS *)chunk_join;
|
|
BLI_movelisttolist(&autotrack_tls_join->results, &autotrack_tls->results);
|
|
}
|
|
|
|
bool BKE_autotrack_context_step(AutoTrackContext *context)
|
|
{
|
|
if (context->num_autotrack_markers == 0) {
|
|
return false;
|
|
}
|
|
|
|
AutoTrackTLS tls;
|
|
BLI_listbase_clear(&tls.results);
|
|
|
|
TaskParallelSettings settings;
|
|
BLI_parallel_range_settings_defaults(&settings);
|
|
settings.use_threading = (context->num_autotrack_markers > 1);
|
|
settings.userdata_chunk = &tls;
|
|
settings.userdata_chunk_size = sizeof(AutoTrackTLS);
|
|
settings.func_reduce = autotrack_context_reduce;
|
|
|
|
BLI_task_parallel_range(
|
|
0, context->num_autotrack_markers, context, autotrack_context_step_cb, &settings);
|
|
|
|
/* Prepare next tracking step by updating the AutoTrack context with new markers and moving
|
|
* tracked markers as an input for the next iteration. */
|
|
context->num_autotrack_markers = 0;
|
|
LISTBASE_FOREACH (AutoTrackTrackingResult *, autotrack_result, &tls.results) {
|
|
if (!autotrack_result->success) {
|
|
continue;
|
|
}
|
|
|
|
/* Insert tracking results to the AutoTrack context to make them usable for the next frame
|
|
* tracking iteration. */
|
|
libmv_autoTrackAddMarker(context->autotrack, &autotrack_result->libmv_marker);
|
|
|
|
/* Update the list of markers which will be tracked on the next iteration. */
|
|
context->autotrack_markers[context->num_autotrack_markers++].libmv_marker =
|
|
autotrack_result->libmv_marker;
|
|
}
|
|
|
|
BLI_spin_lock(&context->spin_lock);
|
|
BLI_movelisttolist(&context->results_to_sync, &tls.results);
|
|
BLI_spin_unlock(&context->spin_lock);
|
|
|
|
return true;
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Context data synchronization.
|
|
*
|
|
* Used to copy tracking result to Blender side, while the tracking is still happening in its
|
|
* thread.
|
|
*
|
|
* \{ */
|
|
|
|
void BKE_autotrack_context_sync(AutoTrackContext *context)
|
|
{
|
|
const int frame_delta = context->is_backwards ? -1 : 1;
|
|
|
|
BLI_spin_lock(&context->spin_lock);
|
|
ListBase results_to_sync = context->results_to_sync;
|
|
BLI_listbase_clear(&context->results_to_sync);
|
|
BLI_spin_unlock(&context->spin_lock);
|
|
|
|
LISTBASE_FOREACH_MUTABLE (AutoTrackTrackingResult *, autotrack_result, &results_to_sync) {
|
|
const libmv_Marker *libmv_marker = &autotrack_result->libmv_marker;
|
|
const int clip_index = libmv_marker->clip;
|
|
const int track_index = libmv_marker->track;
|
|
const AutoTrackClip *autotrack_clip = &context->autotrack_clips[clip_index];
|
|
const MovieClip *clip = autotrack_clip->clip;
|
|
const AutoTrackTrack *autotrack_track = &context->all_autotrack_tracks[track_index];
|
|
MovieTrackingTrack *track = autotrack_track->track;
|
|
|
|
const int start_clip_frame = BKE_movieclip_remap_scene_to_clip_frame(
|
|
clip, context->start_scene_frame);
|
|
const int first_result_frame = start_clip_frame + frame_delta;
|
|
|
|
/* Insert marker which corresponds to the tracking result. */
|
|
MovieTrackingMarker marker;
|
|
libmv_marker_to_dna_marker(
|
|
&autotrack_result->libmv_marker, autotrack_clip->width, autotrack_clip->height, &marker);
|
|
if (!autotrack_result->success) {
|
|
marker.flag |= MARKER_DISABLED;
|
|
}
|
|
BKE_tracking_marker_insert(track, &marker);
|
|
|
|
/* Insert disabled marker at the end of tracked segment.
|
|
* When tracking forward the disabled marker is added at the next frame from the result,
|
|
* when tracking backwards the marker is added at the previous frame. */
|
|
tracking_marker_insert_disabled(track, &marker, context->is_backwards, false);
|
|
|
|
if (marker.framenr == first_result_frame) {
|
|
MovieTrackingMarker *prev_marker = BKE_tracking_marker_get_exact(
|
|
track, marker.framenr - frame_delta);
|
|
BLI_assert(prev_marker != NULL);
|
|
|
|
tracking_marker_insert_disabled(track, prev_marker, !context->is_backwards, false);
|
|
}
|
|
|
|
/* Update synchronized frame to the latest tracked fame from the current results. */
|
|
const int marker_scene_frame = BKE_movieclip_remap_clip_to_scene_frame(clip, marker.framenr);
|
|
if (context->is_backwards) {
|
|
context->synchronized_scene_frame = min_ii(context->synchronized_scene_frame,
|
|
marker_scene_frame);
|
|
}
|
|
else {
|
|
context->synchronized_scene_frame = max_ii(context->synchronized_scene_frame,
|
|
marker_scene_frame);
|
|
}
|
|
|
|
MEM_freeN(autotrack_result);
|
|
}
|
|
|
|
for (int clip_index = 0; clip_index < context->num_clips; clip_index++) {
|
|
MovieTracking *tracking = &context->autotrack_clips[clip_index].clip->tracking;
|
|
BKE_tracking_dopesheet_tag_update(tracking);
|
|
}
|
|
}
|
|
|
|
/* TODO(sergey): Find a way to avoid this, somehow making all needed logic in
|
|
* BKE_autotrack_context_sync(). */
|
|
void BKE_autotrack_context_sync_user(AutoTrackContext *context, MovieClipUser *user)
|
|
{
|
|
user->framenr = context->synchronized_scene_frame;
|
|
}
|
|
|
|
/** \} */
|
|
|
|
/* -------------------------------------------------------------------- */
|
|
/** \name Finalization.
|
|
* \{ */
|
|
|
|
void BKE_autotrack_context_finish(AutoTrackContext *context)
|
|
{
|
|
for (int clip_index = 0; clip_index < context->num_clips; clip_index++) {
|
|
const AutoTrackClip *autotrack_clip = &context->autotrack_clips[clip_index];
|
|
MovieClip *clip = autotrack_clip->clip;
|
|
ListBase *plane_tracks_base = BKE_tracking_get_active_plane_tracks(&clip->tracking);
|
|
const int start_clip_frame = BKE_movieclip_remap_scene_to_clip_frame(
|
|
clip, context->start_scene_frame);
|
|
|
|
LISTBASE_FOREACH (MovieTrackingPlaneTrack *, plane_track, plane_tracks_base) {
|
|
if ((plane_track->flag & PLANE_TRACK_AUTOKEY)) {
|
|
continue;
|
|
}
|
|
for (int track_index = 0; track_index < context->num_all_tracks; track_index++) {
|
|
const AutoTrackTrack *autotrack_track = &context->all_autotrack_tracks[track_index];
|
|
if (!autotrack_track->is_trackable) {
|
|
continue;
|
|
}
|
|
MovieTrackingTrack *track = autotrack_track->track;
|
|
if (BKE_tracking_plane_track_has_point_track(plane_track, track)) {
|
|
BKE_tracking_track_plane_from_existing_motion(plane_track, start_clip_frame);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
void BKE_autotrack_context_free(AutoTrackContext *context)
|
|
{
|
|
if (context->autotrack != NULL) {
|
|
libmv_autoTrackDestroy(context->autotrack);
|
|
}
|
|
|
|
if (context->image_accessor != NULL) {
|
|
tracking_image_accessor_destroy(context->image_accessor);
|
|
}
|
|
|
|
MEM_SAFE_FREE(context->all_autotrack_tracks);
|
|
MEM_SAFE_FREE(context->autotrack_markers);
|
|
|
|
BLI_freelistN(&context->results_to_sync);
|
|
|
|
BLI_spin_end(&context->spin_lock);
|
|
|
|
MEM_freeN(context);
|
|
}
|
|
|
|
/** \} */
|