port ppm vhook to swscaler

patch by Victor Paesa % wzrlpy A arsystel P com %
Original thread:
Date: Sep 2, 2006 11:41 PM
Subject: Re: [Ffmpeg-devel] [PATCH] enable libswscale

Originally committed as revision 6159 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Víctor Paesa 2006-09-03 19:25:06 +00:00 committed by Guillaume Poirier
parent efa4b887d5
commit a4dd5fb73a
1 changed files with 49 additions and 16 deletions

View File

@ -25,6 +25,9 @@
#include <ctype.h>
#include "framehook.h"
#include "avformat.h"
#include "swscale.h"
static int sws_flags = SWS_BICUBIC;
/** Bi-directional pipe structure.
*/
@ -188,6 +191,12 @@ typedef struct
char *buf1;
int size2;
char *buf2;
// This vhook first converts frame to RGB ...
struct SwsContext *toRGB_convert_ctx;
// ... then processes it via a PPM command pipe ...
// ... and finally converts back frame from RGB to initial format
struct SwsContext *fromRGB_convert_ctx;
}
ContextInfo;
@ -246,8 +255,24 @@ void Process(void *ctx, AVPicture *picture, enum PixelFormat pix_fmt, int width,
if ( !err )
{
avpicture_fill(&picture1, ci->buf1, PIX_FMT_RGB24, width, height);
if (img_convert(&picture1, PIX_FMT_RGB24, picture, pix_fmt, width, height) < 0)
err = 1;
// if we already got a SWS context, let's realloc if is not re-useable
ci->toRGB_convert_ctx = sws_getCachedContext(ci->toRGB_convert_ctx,
width, height, pix_fmt,
width, height, PIX_FMT_RGB24,
sws_flags, NULL, NULL, NULL);
if (ci->toRGB_convert_ctx == NULL) {
av_log(NULL, AV_LOG_ERROR,
"Cannot initialize the toRGB conversion context\n");
exit(1);
}
// img_convert parameters are 2 first destination, then 4 source
// sws_scale parameters are context, 4 first source, then 2 destination
sws_scale(ci->toRGB_convert_ctx,
picture->data, picture->linesize, 0, height,
picture1.data, picture1.linesize);
pict = &picture1;
}
}
@ -294,22 +319,28 @@ void Process(void *ctx, AVPicture *picture, enum PixelFormat pix_fmt, int width,
/* Convert the returned PPM back to the input format */
if ( !err )
{
/* Actually, this is wrong, since the out_width/out_height returned from the
* filter won't necessarily be the same as width and height - img_resample
* won't scale rgb24, so the only way out of this is to convert to something
* that img_resample does like [which may or may not be pix_fmt], rescale
* and finally convert to pix_fmt... slow, but would provide the most flexibility.
*
* Currently, we take the upper left width/height pixels from the filtered image,
* smaller images are going to be corrupted or cause a crash.
*
* Finally, what should we do in case of this call failing? Up to now, failures
* are gracefully ignored and the original image is returned - in this case, a
* failure may corrupt the input.
/* The out_width/out_height returned from the PPM
* filter won't necessarily be the same as width and height
* but it will be scaled anyway to width/height.
*/
if (img_convert(picture, pix_fmt, &picture2, PIX_FMT_RGB24, width, height) < 0)
{
av_log(NULL, AV_LOG_DEBUG,
"PPM vhook: Input dimensions: %d x %d Output dimensions: %d x %d\n",
width, height, out_width, out_height);
ci->fromRGB_convert_ctx = sws_getCachedContext(ci->fromRGB_convert_ctx,
out_width, out_height, PIX_FMT_RGB24,
width, height, pix_fmt,
sws_flags, NULL, NULL, NULL);
if (ci->fromRGB_convert_ctx == NULL) {
av_log(NULL, AV_LOG_ERROR,
"Cannot initialize the fromRGB conversion context\n");
exit(1);
}
// img_convert parameters are 2 first destination, then 4 source
// sws_scale parameters are context, 4 first source, then 2 destination
sws_scale(ci->fromRGB_convert_ctx,
picture2.data, picture2.linesize, 0, out_height,
picture->data, picture->linesize);
}
}
@ -326,6 +357,8 @@ void Release(void *ctx)
rwpipe_close( ci->rw );
av_free( ci->buf1 );
av_free( ci->buf2 );
sws_freeContext(ci->toRGB_convert_ctx);
sws_freeContext(ci->fromRGB_convert_ctx);
av_free(ctx);
}
}