mpv/video/filter/vf.c

688 lines
21 KiB
C

/*
* This file is part of MPlayer.
*
* MPlayer 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.
*
* MPlayer 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 MPlayer; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include <sys/types.h>
#include <libavutil/common.h>
#include <libavutil/mem.h>
#include "config.h"
#include "mpvcore/mp_msg.h"
#include "mpvcore/m_option.h"
#include "mpvcore/m_config.h"
#include "mpvcore/options.h"
#include "video/img_format.h"
#include "video/mp_image.h"
#include "video/mp_image_pool.h"
#include "vf.h"
#include "video/memcpy_pic.h"
extern const vf_info_t vf_info_vo;
extern const vf_info_t vf_info_crop;
extern const vf_info_t vf_info_expand;
extern const vf_info_t vf_info_pp;
extern const vf_info_t vf_info_scale;
extern const vf_info_t vf_info_format;
extern const vf_info_t vf_info_noformat;
extern const vf_info_t vf_info_flip;
extern const vf_info_t vf_info_rotate;
extern const vf_info_t vf_info_mirror;
extern const vf_info_t vf_info_noise;
extern const vf_info_t vf_info_eq;
extern const vf_info_t vf_info_gradfun;
extern const vf_info_t vf_info_unsharp;
extern const vf_info_t vf_info_swapuv;
extern const vf_info_t vf_info_down3dright;
extern const vf_info_t vf_info_hqdn3d;
extern const vf_info_t vf_info_ilpack;
extern const vf_info_t vf_info_dsize;
extern const vf_info_t vf_info_softpulldown;
extern const vf_info_t vf_info_pullup;
extern const vf_info_t vf_info_delogo;
extern const vf_info_t vf_info_phase;
extern const vf_info_t vf_info_divtc;
extern const vf_info_t vf_info_softskip;
extern const vf_info_t vf_info_screenshot;
extern const vf_info_t vf_info_sub;
extern const vf_info_t vf_info_yadif;
extern const vf_info_t vf_info_stereo3d;
extern const vf_info_t vf_info_dlopen;
extern const vf_info_t vf_info_lavfi;
extern const vf_info_t vf_info_vaapi;
// list of available filters:
static const vf_info_t *const filter_list[] = {
&vf_info_crop,
&vf_info_expand,
&vf_info_scale,
&vf_info_vo,
&vf_info_format,
&vf_info_noformat,
&vf_info_flip,
&vf_info_rotate,
&vf_info_mirror,
#if HAVE_LIBPOSTPROC
&vf_info_pp,
#endif
#if HAVE_VF_LAVFI
&vf_info_lavfi,
#endif
&vf_info_screenshot,
&vf_info_noise,
&vf_info_eq,
&vf_info_gradfun,
&vf_info_unsharp,
&vf_info_swapuv,
&vf_info_down3dright,
&vf_info_hqdn3d,
&vf_info_ilpack,
&vf_info_dsize,
&vf_info_softpulldown,
&vf_info_pullup,
&vf_info_delogo,
&vf_info_phase,
&vf_info_divtc,
&vf_info_sub,
&vf_info_yadif,
&vf_info_stereo3d,
#if HAVE_DLOPEN
&vf_info_dlopen,
#endif
#if HAVE_VAAPI_VPP
&vf_info_vaapi,
#endif
NULL
};
static bool get_desc(struct m_obj_desc *dst, int index)
{
if (index >= MP_ARRAY_SIZE(filter_list) - 1)
return false;
const vf_info_t *vf = filter_list[index];
*dst = (struct m_obj_desc) {
.name = vf->name,
.description = vf->description,
.priv_size = vf->priv_size,
.priv_defaults = vf->priv_defaults,
.options = vf->options,
.p = vf,
};
return true;
}
// For the vf option
const struct m_obj_list vf_obj_list = {
.get_desc = get_desc,
.description = "video filters",
.legacy_hacks = true, // some filters have custom option parsing
};
int vf_control(struct vf_instance *vf, int cmd, void *arg)
{
return vf->control(vf, cmd, arg);
}
static void vf_fix_img_params(struct mp_image *img, struct mp_image_params *p)
{
// Filters must absolutely set these correctly.
assert(img->w == p->w && img->h == p->h);
assert(img->imgfmt == p->imgfmt);
// Too many things don't set this correctly.
// If --colormatrix is used, decoder and filter chain disagree too.
// In general, it's probably more convenient to force these here,
// instead of requiring filters to set these correctly.
img->colorspace = p->colorspace;
img->levels = p->colorlevels;
img->chroma_location = p->chroma_location;
mp_image_set_display_size(img, p->d_w, p->d_h);
}
// Get a new image for filter output, with size and pixel format according to
// the last vf_config call.
struct mp_image *vf_alloc_out_image(struct vf_instance *vf)
{
assert(vf->fmt_out.configured);
struct mp_image_params *p = &vf->fmt_out.params;
struct mp_image *img = mp_image_pool_get(vf->out_pool, p->imgfmt, p->w, p->h);
vf_fix_img_params(img, p);
return img;
}
void vf_make_out_image_writeable(struct vf_instance *vf, struct mp_image *img)
{
struct mp_image_params *p = &vf->fmt_out.params;
assert(vf->fmt_out.configured);
assert(p->imgfmt == img->imgfmt);
assert(p->w == img->w && p->h == img->h);
mp_image_pool_make_writeable(vf->out_pool, img);
}
//============================================================================
static int vf_default_query_format(struct vf_instance *vf, unsigned int fmt)
{
return vf_next_query_format(vf, fmt);
}
static struct mp_image *vf_default_filter(struct vf_instance *vf,
struct mp_image *mpi)
{
assert(!vf->filter_ext);
return mpi;
}
static void print_fmt(int msglevel, struct vf_format *fmt)
{
if (fmt && fmt->configured) {
struct mp_image_params *p = &fmt->params;
mp_msg(MSGT_VFILTER, msglevel, "%dx%d", p->w, p->h);
if (p->w != p->d_w || p->h != p->d_h)
mp_msg(MSGT_VFILTER, msglevel, "->%dx%d", p->d_w, p->d_h);
mp_msg(MSGT_VFILTER, msglevel, " %s %#x", mp_imgfmt_to_name(p->imgfmt),
fmt->flags);
mp_msg(MSGT_VFILTER, msglevel, " %s/%s", mp_csp_names[p->colorspace],
mp_csp_levels_names[p->colorlevels]);
} else {
mp_msg(MSGT_VFILTER, msglevel, "???");
}
}
void vf_print_filter_chain(int msglevel, struct vf_instance *vf)
{
if (!mp_msg_test(MSGT_VFILTER, msglevel))
return;
for (vf_instance_t *f = vf; f; f = f->next) {
mp_msg(MSGT_VFILTER, msglevel, " [%s] ", f->info->name);
print_fmt(msglevel, &f->fmt_in);
if (f->next) {
mp_msg(MSGT_VFILTER, msglevel, " -> ");
print_fmt(msglevel, &f->fmt_out);
}
mp_msg(MSGT_VFILTER, msglevel, "\n");
}
}
static struct vf_instance *vf_open(struct MPOpts *opts, vf_instance_t *next,
const char *name, char **args)
{
struct m_obj_desc desc;
if (!m_obj_list_find(&desc, &vf_obj_list, bstr0(name))) {
mp_tmsg(MSGT_VFILTER, MSGL_ERR,
"Couldn't find video filter '%s'.\n", name);
return NULL;
}
vf_instance_t *vf = talloc_zero(NULL, struct vf_instance);
*vf = (vf_instance_t) {
.info = desc.p,
.opts = opts,
.next = next,
.config = vf_next_config,
.control = vf_next_control,
.query_format = vf_default_query_format,
.filter = vf_default_filter,
.out_pool = talloc_steal(vf, mp_image_pool_new(16)),
};
struct m_config *config = m_config_from_obj_desc(vf, &desc);
void *priv = NULL;
if (m_config_initialize_obj(config, &desc, &priv, &args) < 0)
goto error;
vf->priv = priv;
int retcode = vf->info->open(vf, (char *)args);
if (retcode < 1)
goto error;
return vf;
error:
talloc_free(vf);
return NULL;
}
vf_instance_t *vf_open_filter(struct MPOpts *opts, vf_instance_t *next,
const char *name, char **args)
{
if (args && strcmp(args[0], "_oldargs_")) {
int i, l = 0;
for (i = 0; args && args[2 * i]; i++)
l += 1 + strlen(args[2 * i]) + 1 + strlen(args[2 * i + 1]);
l += strlen(name);
{
char str[l + 1];
char *p = str;
p += sprintf(str, "%s", name);
for (i = 0; args && args[2 * i]; i++)
p += sprintf(p, " %s=%s", args[2 * i], args[2 * i + 1]);
mp_msg(MSGT_VFILTER, MSGL_INFO, "%s[%s]\n",
mp_gtext("Opening video filter: "), str);
}
} else if (strcmp(name, "vo")) {
if (args && strcmp(args[0], "_oldargs_") == 0)
mp_msg(MSGT_VFILTER, MSGL_INFO, "%s[%s=%s]\n",
mp_gtext("Opening video filter: "), name, args[1]);
else
mp_msg(MSGT_VFILTER, MSGL_INFO, "%s[%s]\n",
mp_gtext("Opening video filter: "), name);
}
return vf_open(opts, next, name, args);
}
/**
* \brief adds a filter before the last one (which should be the vo filter).
* \param vf start of the filter chain.
* \param name name of the filter to add.
* \param args argument list for the filter.
* \return pointer to the filter instance that was created.
*/
vf_instance_t *vf_add_before_vo(vf_instance_t **vf, char *name, char **args)
{
struct MPOpts *opts = (*vf)->opts;
vf_instance_t *vo, *prev = NULL, *new;
// Find the last filter (should be vf_vo)
for (vo = *vf; vo->next; vo = vo->next)
prev = vo;
new = vf_open_filter(opts, vo, name, args);
if (prev)
prev->next = new;
else
*vf = new;
return new;
}
//============================================================================
unsigned int vf_match_csp(vf_instance_t **vfp, const unsigned int *list,
unsigned int preferred)
{
vf_instance_t *vf = *vfp;
struct MPOpts *opts = vf->opts;
const unsigned int *p;
unsigned int best = 0;
int ret;
if ((p = list))
while (*p) {
ret = vf->query_format(vf, *p);
mp_msg(MSGT_VFILTER, MSGL_V, "[%s] query(%s) -> %x\n",
vf->info->name, vo_format_name(*p), ret);
if (ret & VFCAP_CSP_SUPPORTED_BY_HW) {
best = *p;
break;
}
if (ret & VFCAP_CSP_SUPPORTED && !best)
best = *p;
++p;
}
if (best)
return best; // bingo, they have common csp!
// ok, then try with scale:
if (vf->info == &vf_info_scale)
return 0; // avoid infinite recursion!
vf = vf_open_filter(opts, vf, "scale", NULL);
if (!vf)
return 0; // failed to init "scale"
// try the preferred csp first:
if (preferred && vf->query_format(vf, preferred))
best = preferred;
else
// try the list again, now with "scaler" :
if ((p = list))
while (*p) {
ret = vf->query_format(vf, *p);
mp_msg(MSGT_VFILTER, MSGL_V, "[%s] query(%s) -> %x\n",
vf->info->name, vo_format_name(*p), ret);
if (ret & VFCAP_CSP_SUPPORTED_BY_HW) {
best = *p;
break;
}
if (ret & VFCAP_CSP_SUPPORTED && !best)
best = *p;
++p;
}
if (best)
*vfp = vf; // else uninit vf !FIXME!
return best;
}
// Used by filters to add a filtered frame to the output queue.
// Ownership of img is transferred from caller to the filter chain.
void vf_add_output_frame(struct vf_instance *vf, struct mp_image *img)
{
if (img) {
// vf_vo doesn't have output config
if (vf->fmt_out.configured)
vf_fix_img_params(img, &vf->fmt_out.params);
MP_TARRAY_APPEND(vf, vf->out_queued, vf->num_out_queued, img);
}
}
static struct mp_image *vf_dequeue_output_frame(struct vf_instance *vf)
{
struct mp_image *res = NULL;
if (vf->num_out_queued) {
res = vf->out_queued[0];
MP_TARRAY_REMOVE_AT(vf->out_queued, vf->num_out_queued, 0);
}
return res;
}
// Input a frame into the filter chain.
// Return >= 0 on success, < 0 on failure (even if output frames were produced)
int vf_filter_frame(struct vf_instance *vf, struct mp_image *img)
{
assert(vf->fmt_in.configured);
vf_fix_img_params(img, &vf->fmt_in.params);
if (vf->filter_ext) {
return vf->filter_ext(vf, img);
} else {
vf_add_output_frame(vf, vf->filter(vf, img));
return 0;
}
}
// Output the next queued image (if any) from the full filter chain.
struct mp_image *vf_chain_output_queued_frame(struct vf_instance *vf)
{
while (1) {
struct vf_instance *last = NULL;
for (struct vf_instance * cur = vf; cur; cur = cur->next) {
if (cur->num_out_queued)
last = cur;
}
if (!last)
return NULL;
struct mp_image *img = vf_dequeue_output_frame(last);
if (!last->next)
return img;
vf_filter_frame(last->next, img);
}
}
static void vf_forget_frames(struct vf_instance *vf)
{
for (int n = 0; n < vf->num_out_queued; n++)
talloc_free(vf->out_queued[n]);
vf->num_out_queued = 0;
}
void vf_chain_seek_reset(struct vf_instance *vf)
{
vf->control(vf, VFCTRL_SEEK_RESET, NULL);
for (struct vf_instance *cur = vf; cur; cur = cur->next)
vf_forget_frames(cur);
}
int vf_reconfig_wrapper(struct vf_instance *vf, const struct mp_image_params *p,
int flags)
{
vf_forget_frames(vf);
mp_image_pool_clear(vf->out_pool);
vf->fmt_in = (struct vf_format) {
.params = *p,
.flags = flags,
};
vf->fmt_out = (struct vf_format){0};
int r;
if (vf->reconfig) {
struct mp_image_params params = *p;
r = vf->reconfig(vf, &params, flags);
} else {
r = vf->config(vf, p->w, p->h, p->d_w, p->d_h, flags, p->imgfmt);
r = r ? 0 : -1;
}
if (r >= 0) {
vf->fmt_in.configured = 1;
if (vf->next)
vf->fmt_out = vf->next->fmt_in;
}
return r;
}
int vf_next_reconfig(struct vf_instance *vf, struct mp_image_params *p,
int outflags)
{
struct MPOpts *opts = vf->opts;
int flags = vf->next->query_format(vf->next, p->imgfmt);
if (!flags) {
// hmm. colorspace mismatch!!!
// let's insert the 'scale' filter, it does the job for us:
vf_instance_t *vf2;
if (vf->next->info == &vf_info_scale)
return -1; // scale->scale
vf2 = vf_open_filter(opts, vf->next, "scale", NULL);
if (!vf2)
return -1; // shouldn't happen!
vf->next = vf2;
flags = vf->next->query_format(vf->next, p->imgfmt);
if (!flags) {
mp_tmsg(MSGT_VFILTER, MSGL_ERR, "Cannot find matching colorspace, "
"even by inserting 'scale' :(\n");
return -1; // FAIL
}
}
return vf_reconfig_wrapper(vf->next, p, outflags);
}
int vf_next_config(struct vf_instance *vf,
int width, int height, int d_width, int d_height,
unsigned int voflags, unsigned int outfmt)
{
struct mp_image_params p = {
.imgfmt = outfmt,
.w = width,
.h = height,
.d_w = d_width,
.d_h = d_height,
.colorspace = vf->fmt_in.params.colorspace,
.colorlevels = vf->fmt_in.params.colorlevels,
.chroma_location = vf->fmt_in.params.chroma_location,
.outputlevels = vf->fmt_in.params.outputlevels,
};
// Fix csp in case of pixel format change
mp_image_params_guess_csp(&p);
int r = vf_reconfig_wrapper(vf->next, &p, voflags);
return r < 0 ? 0 : 1;
}
int vf_next_control(struct vf_instance *vf, int request, void *data)
{
return vf->next->control(vf->next, request, data);
}
int vf_next_query_format(struct vf_instance *vf, unsigned int fmt)
{
return vf->next->query_format(vf->next, fmt);
}
//============================================================================
vf_instance_t *append_filters(vf_instance_t *last,
struct m_obj_settings *vf_settings)
{
struct MPOpts *opts = last->opts;
vf_instance_t *vf;
int i;
if (vf_settings) {
// We want to add them in the 'right order'
for (i = 0; vf_settings[i].name; i++)
/* NOP */;
for (i--; i >= 0; i--) {
//printf("Open filter %s\n",vf_settings[i].name);
vf = vf_open_filter(opts, last, vf_settings[i].name,
vf_settings[i].attribs);
if (vf) {
if (vf_settings[i].label)
vf->label = talloc_strdup(vf, vf_settings[i].label);
last = vf;
}
}
}
return last;
}
vf_instance_t *vf_find_by_label(vf_instance_t *chain, const char *label)
{
while (chain) {
if (chain->label && label && strcmp(chain->label, label) == 0)
return chain;
chain = chain->next;
}
return NULL;
}
//============================================================================
void vf_uninit_filter(vf_instance_t *vf)
{
if (vf->uninit)
vf->uninit(vf);
vf_forget_frames(vf);
talloc_free(vf);
}
void vf_uninit_filter_chain(vf_instance_t *vf)
{
while (vf) {
vf_instance_t *next = vf->next;
vf_uninit_filter(vf);
vf = next;
}
}
// When changing the size of an image that had old_w/old_h with
// DAR *d_width/*d_height to the new size new_w/new_h, adjust
// *d_width/*d_height such that the new image has the same pixel aspect ratio.
void vf_rescale_dsize(int *d_width, int *d_height, int old_w, int old_h,
int new_w, int new_h)
{
*d_width = *d_width * new_w / old_w;
*d_height = *d_height * new_h / old_h;
}
// Set *d_width/*d_height to display aspect ratio with the givem source size
void vf_set_dar(int *d_w, int *d_h, int w, int h, double dar)
{
*d_w = w;
*d_h = h;
if (dar > 0.01) {
*d_w = h * dar;
// we don't like horizontal downscale
if (*d_w < w) {
*d_w = w;
*d_h = w / dar;
}
}
}
void vf_detc_init_pts_buf(struct vf_detc_pts_buf *p)
{
p->inpts_prev = MP_NOPTS_VALUE;
p->outpts_prev = MP_NOPTS_VALUE;
p->lastdelta = 0;
}
static double vf_detc_adjust_pts_internal(struct vf_detc_pts_buf *p,
double pts, bool reset_pattern,
bool skip_frame, double delta,
double boundfactor_minus,
double increasefactor,
double boundfactor_plus)
{
double newpts;
if (pts == MP_NOPTS_VALUE)
return pts;
if (delta <= 0) {
if (p->inpts_prev == MP_NOPTS_VALUE)
delta = 0;
else if (pts == p->inpts_prev)
delta = p->lastdelta;
else
delta = pts - p->inpts_prev;
}
p->inpts_prev = pts;
p->lastdelta = delta;
if (skip_frame)
return MP_NOPTS_VALUE;
/* detect bogus deltas and then passthru pts (possibly caused by seeking,
* or bad input) */
if (p->outpts_prev == MP_NOPTS_VALUE || reset_pattern || delta <= 0.0 ||
delta >= 0.5)
newpts = pts;
else {
// turn 5 frames into 4
newpts = p->outpts_prev + delta * increasefactor;
// bound to input pts in a sensible way; these numbers come because we
// map frames the following way when ivtc'ing:
// 0/30 -> 0/24 diff=0
// 1/30 -> 1/24 diff=1/120
// 2/30 -> -
// 3/30 -> 2/24 diff=-1/60
// 4/30 -> 3/24 diff=-1/120
if (newpts < pts - delta * boundfactor_minus)
newpts = pts - delta * boundfactor_minus;
if (newpts > pts + delta * boundfactor_plus)
newpts = pts + delta * boundfactor_plus;
if (newpts < p->outpts_prev)
newpts = p->outpts_prev; // damage control
}
p->outpts_prev = newpts;
return newpts;
}
double vf_detc_adjust_pts(struct vf_detc_pts_buf *p, double pts,
bool reset_pattern, bool skip_frame)
{
// standard telecine (see above)
return vf_detc_adjust_pts_internal(p, pts, reset_pattern, skip_frame,
0, 0.5, 1.25, 0.25);
}
double vf_softpulldown_adjust_pts(struct vf_detc_pts_buf *p, double pts,
bool reset_pattern, bool skip_frame,
int last_frame_duration)
{
// for the softpulldown filter we get:
// 0/60 -> 0/30
// 2/60 -> 1/30
// 5/60 -> 2/30
// 7/60 -> 3/30, 4/30
return vf_detc_adjust_pts_internal(p, pts, reset_pattern, skip_frame,
0, 1.0 / last_frame_duration,
2.0 / last_frame_duration,
1.0 / last_frame_duration);
}