optional hardware accelerated scaling

- extend hardware config with input_width, input_height fields
- if specified and different from width/height perform hardware accelerated scaling before encoding
- update examples
- update docs
- update readme

Adds dependency on libavfilter.

Closes #24
Makes #5 more complex
Indirectly related to #25
Indirectly related to #6

CMakeLists.txt

@@ -5,7 +5,7 @@ project(
 )
 
 add_library(hve hve.c)
-target_link_libraries(hve avcodec avutil)
+target_link_libraries(hve avcodec avutil avfilter)
 install(TARGETS hve DESTINATION lib)
 install(FILES hve.h DESTINATION include)
 

README.md

@@ -1,6 +1,6 @@
 # HVE - Hardware Video Encoder C library
 
-This library wraps hardware video encoding in a simple interface.
+This library wraps hardware video encoding and scaling in a simple interface.
 There are no performance loses (at the cost of library flexibility).
 
 Currently it supports VAAPI and various codecs (H.264, HEVC, ...).\
@@ -20,7 +20,7 @@ Raw encoding (H264, HEVC, ...):
 - raw dumping (H264, HEVC, ...)
 - ...
 
-Complex pipelines (muxing, scaling, color conversions, filtering) are beyond the scope of this library.
+Complex pipelines (muxing, filtering) are beyond the scope of this library.
 
 ## Platforms 
 
@@ -34,7 +34,7 @@ Intel VAAPI compatible hardware encoders ([Quick Sync Video](https://ark.intel.c
 ## Dependencies
 
 Library depends on:
-- FFmpeg `avcodec` and `avutil` (at least 3.4 version)
+- FFmpeg `avcodec`, `avutil`, `avfilter` (at least 3.4 version)
 
 Works with system FFmpeg on Ubuntu 18.04 and doesn't on 16.04 (outdated FFmpeg and VAAPI ecosystem).
 
@@ -46,7 +46,7 @@ Tested on Ubuntu 18.04.
 # update package repositories
 sudo apt-get update 
 # get avcodec and avutil (and ffmpeg for testing)
-sudo apt-get install ffmpeg libavcodec-dev libavutil-dev
+sudo apt-get install ffmpeg libavcodec-dev libavutil-dev libavfilter-dev
 # get compilers and make and cmake
 sudo apt-get install build-essential
 # get cmake - we need to specify libcurl4 for Ubuntu 18.04 dependencies problem
@@ -108,7 +108,7 @@ You should see procedurally generated video (moving through greyscale).
 
 ## Using
 
-See examples directory for a more complete and commented examples with error handling.
+See examples directory for more complete and commented examples with error handling.
 
 There are just 4 functions and 3 user-visible data types:
 - `hve_init`
@@ -117,15 +117,15 @@ There are just 4 functions and 3 user-visible data types:
 - `hve_close`
 
 ```C
-	struct hve_config hardware_config = {WIDTH, HEIGHT, FRAMERATE, DEVICE, ENCODER,
-		PIXEL_FORMAT, PROFILE, BFRAMES, BITRATE, QP, GOP_SIZE, COMPRESSION_LEVEL};
+	struct hve_config hardware_config = {WIDTH, HEIGHT, INPUT_WIDTH, INPUT_HEIGHT, FRAMERATE,
+		DEVICE, ENCODER, PIXEL_FORMAT, PROFILE, BFRAMES, BITRATE, QP, GOP_SIZE, COMPRESSION_LEVEL};
 	struct hve *hardware_encoder=hve_init(&hardware_config);
 	struct hve_frame frame = { 0 };
 
 	//later assuming PIXEL_FORMAT is "nv12" (you may use something else)
 
 	//fill with your stride (width including padding if any)
-	frame.linesize[0] = frame.linesize[1] = WIDTH;
+	frame.linesize[0] = frame.linesize[1] = INPUT_WIDTH;
 
 	AVPacket *packet; //encoded data is returned in FFmpeg packet
 	int failed; //error indicator while encoding
@@ -170,13 +170,13 @@ You have several options.
 
 For static linking of HVE and dynamic linking of FFmpeg libraries (easiest):
 - copy `hve.h` and `hve.c` to your project and add them in your favourite IDE
-- add `avcodec` and `avutil` to linked libraries in IDE project configuration
+- add `avcodec`, `avutil`, `avfilter` to linked libraries in IDE project configuration
 
 For dynamic linking of HVE and FFmpeg libraries:
 - place `hve.h` where compiler can find it (e.g. `make install` for `/usr/local/include/hve.h`)
 - place `libhve.so` where linker can find it (e.g. `make install` for `/usr/local/lib/libhve.so`)
 - make sure `/usr/local/...` is considered for libraries
-- add `hve`, `avcodec` and `avutil` to linked libraries in IDE project configuration
+- add `hve`, `avcodec`, `avutil`, `avfilter` to linked libraries in IDE project configuration
 - make sure `libhve.so` is reachable to you program at runtime (e.g. set `LD_LIBRARIES_PATH`)
 
 ### CMake
@@ -208,7 +208,7 @@ add_library(hve SHARED hardware-video-encoder/hve.c)
 
 add_executable(your-project main.cpp)
 target_include_directories(your-project PRIVATE hardware-video-encoder)
-target_link_libraries(your-project hve avcodec avutil)
+target_link_libraries(your-project hve avcodec avutil avfilter)
 ```
 
 For example see [realsense-ir-to-vaapi-h264](https://github.com/bmegli/realsense-ir-to-vaapi-h264)
@@ -219,14 +219,14 @@ Assuming your `main.c`/`main.cpp` and `hve.h`, `hve.c` are all in the same direc
 
 C
 ```bash
-gcc main.c hve.c -lavcodec -lavutil -o your-program
+gcc main.c hve.c -lavcodec -lavutil -lavfilter -o your-program
 ```
 
 C++
 ```bash
 gcc -c hve.c
 g++ -c main.cpp
-g++ hve.o main.o -lavcodec -lavutil -o your program
+g++ hve.o main.o -lavcodec -lavutil -lavfilter -o your program
 ```
 
 ## License
@@ -240,7 +240,7 @@ This is similiar to LGPL but more permissive:
 Like in LGPL, if you modify this library, you have to make your changes available.
 Making a github fork of the library with your changes satisfies those requirements perfectly.
 
-Since you are linking to FFmpeg libraries. Consider also `avcodec` and `avutil` licensing.
+Since you are linking to FFmpeg libraries consider also `avcodec`, `avutil` and `avfilter` licensing.
 
 ## Additional information
 

examples/hve_encode_raw_h264.c

@@ -16,6 +16,8 @@
 
 const int WIDTH=1280;
 const int HEIGHT=720;
+const int INPUT_WIDTH=1280; //optional hardware scaling if different from width
+const int INPUT_HEIGHT=720; //optional hardware scaling if different from height
 const int FRAMERATE=30;
 int SECONDS=10;
 const char *DEVICE=NULL; //NULL for default or device e.g. "/dev/dri/renderD128"
@@ -40,8 +42,8 @@ int main(int argc, char* argv[])
 		return -1;
 
 	//prepare library data
-	struct hve_config hardware_config = {WIDTH, HEIGHT, FRAMERATE, DEVICE, ENCODER,
-			PIXEL_FORMAT, PROFILE, BFRAMES, BITRATE, QP, GOP_SIZE, COMPRESSION_LEVEL};
+	struct hve_config hardware_config = {WIDTH, HEIGHT, INPUT_WIDTH, INPUT_HEIGHT, FRAMERATE,
+	DEVICE, ENCODER, PIXEL_FORMAT, PROFILE, BFRAMES, BITRATE, QP, GOP_SIZE, COMPRESSION_LEVEL};
 	struct hve *hardware_encoder;
 
 	//prepare file for raw H.264 output
@@ -76,20 +78,20 @@ int encoding_loop(struct hve *hardware_encoder, FILE *output_file)
 	//we are working with NV12 because we specified nv12 pixel format
 	//when calling hve_init, in principle we could use other format
 	//if hardware supported it (e.g. RGB0 is supported on my Intel)
-	uint8_t Y[WIDTH*HEIGHT]; //dummy NV12 luminance data
-	uint8_t color[WIDTH*HEIGHT/2]; //dummy NV12 color data
+	uint8_t Y[INPUT_WIDTH*INPUT_HEIGHT]; //dummy NV12 luminance data
+	uint8_t color[INPUT_WIDTH*INPUT_HEIGHT/2]; //dummy NV12 color data
 
 	//fill with your stride (width including padding if any)
-	frame.linesize[0] = frame.linesize[1] = WIDTH;
+	frame.linesize[0] = frame.linesize[1] = INPUT_WIDTH;
 
 	//encoded data is returned in FFmpeg packet
 	AVPacket *packet;
 
 	for(f=0;f<frames;++f)
 	{
 		//prepare dummy image data, normally you would take it from camera or other source
-		memset(Y, f % 255, WIDTH*HEIGHT); //NV12 luminance (ride through greyscale)
-		memset(color, 128, WIDTH*HEIGHT/2); //NV12 UV (no color really)
+		memset(Y, f % 255, INPUT_WIDTH*INPUT_HEIGHT); //NV12 luminance (ride through greyscale)
+		memset(color, 128, INPUT_WIDTH*INPUT_HEIGHT/2); //NV12 UV (no color really)
 
 		//fill hve_frame with pointers to your data in NV12 pixel format
 		frame.data[0]=Y;
@@ -149,7 +151,7 @@ int hint_user_on_failure(char *argv[])
 void hint_user_on_success()
 {
 	printf("finished successfully\n");
-	printf("output written to \"outout.h264\" file\n");
+	printf("output written to \"output.h264\" file\n");
 	printf("test with:\n\n");
 	printf("ffplay output.h264\n");
 }

examples/hve_encode_raw_hevc10.c

@@ -16,6 +16,8 @@
 
 const int WIDTH=1280;
 const int HEIGHT=720;
+const int INPUT_WIDTH=1280; //optional scaling if different from width
+const int INPUT_HEIGHT=720; //optional scaling if different from height
 const int FRAMERATE=30;
 int SECONDS=10;
 const char *DEVICE=NULL; //NULL for default or device e.g. "/dev/dri/renderD128"
@@ -40,8 +42,8 @@ int main(int argc, char* argv[])
 		return -1;
 
 	//prepare library data
-	struct hve_config hardware_config = {WIDTH, HEIGHT, FRAMERATE, DEVICE, ENCODER,
-			PIXEL_FORMAT, PROFILE, BFRAMES, BITRATE, QP, GOP_SIZE, COMPRESSION_LEVEL};
+	struct hve_config hardware_config = {WIDTH, HEIGHT, INPUT_WIDTH, INPUT_HEIGHT, FRAMERATE,
+	DEVICE, ENCODER, PIXEL_FORMAT, PROFILE, BFRAMES, BITRATE, QP, GOP_SIZE, COMPRESSION_LEVEL};
 	struct hve *hardware_encoder;
 
 	//prepare file for raw HEVC output
@@ -76,21 +78,21 @@ int encoding_loop(struct hve *hardware_encoder, FILE *output_file)
 	//we are working with P010LE because we specified p010le pixel format
 	//when calling hve_init, in principle we could use other format
 	//if hardware supported it (e.g. RGB0 is supported on my Intel)
-	uint16_t Y[WIDTH*HEIGHT]; //dummy p010le luminance data (or p016le)
-	uint16_t color[WIDTH*HEIGHT/2]; //dummy p010le color data (or p016le)
+	uint16_t Y[INPUT_WIDTH*INPUT_HEIGHT]; //dummy p010le luminance data (or p016le)
+	uint16_t color[INPUT_WIDTH*INPUT_HEIGHT/2]; //dummy p010le color data (or p016le)
 
 	//fill with your stride (width including padding if any)
-	frame.linesize[0] = frame.linesize[1] = WIDTH*2;
+	frame.linesize[0] = frame.linesize[1] = INPUT_WIDTH*2;
 
 	//encoded data is returned in FFmpeg packet
 	AVPacket *packet;
 
 	for(f=0;f<frames;++f)
 	{
 		//prepare dummy image data, normally you would take it from camera or other source
-		for(int i=0;i<WIDTH*HEIGHT;++i)
+		for(int i=0;i<INPUT_WIDTH*INPUT_HEIGHT;++i)
 			Y[i] = UINT16_MAX * f / frames; //linear interpolation between 0 and UINT16_MAX
-		for(int i=0;i<WIDTH*HEIGHT/2;++i)
+		for(int i=0;i<INPUT_WIDTH*INPUT_HEIGHT/2;++i)
 			color[i] = UINT16_MAX / 2; //dummy middle value for U/V, equals 128 << 8, equals 32768
 		//fill hve_frame with pointers to your data in P010LE pixel format
 		//note that we have actually prepared P016LE data but it is binary compatible with P010LE
@@ -152,7 +154,7 @@ int hint_user_on_failure(char *argv[])
 void hint_user_on_success()
 {
 	printf("finished successfully\n");
-	printf("output written to \"out.hevc\" file\n");
+	printf("output written to \"output.hevc\" file\n");
 	printf("test with:\n\n");
 	printf("ffplay output.hevc\n");
 }