NOTE: The current pre-release version of the library only supports encoding of images in a multitude of modes. Decoding support is planned for the 1.0.0 version, but as of yet no work has been done in that direction.
SSTV encoder/decoder C library suitable for both desktop and embedded applications.
Output for PD180 mode of this test image.
The following modes are currently supported:
FAX480MARTIN_M1MARTIN_M2MARTIN_M3MARTIN_M4PD50PD90PD120PD160PD180PD240PD290ROBOT_BW8_BROBOT_BW8_GROBOT_BW8_RROBOT_BW12_BROBOT_BW12_GROBOT_BW12_RROBOT_BW24_BROBOT_BW24_GROBOT_BW24_RROBOT_BW36_BROBOT_BW36_GROBOT_BW36_RROBOT_C12ROBOT_C24ROBOT_C36ROBOT_C72SCOTTIE_DXSCOTTIE_S1SCOTTIE_S2SCOTTIE_S3SCOTTIE_S4
Compiling the library and encoding tool:
cmake .
make
This will generate the following files:
lib/libsstv.so- dynamic linking versioninclude/libsstv.h- C header file for librarybin/sstv-encode- encoding tool
If you want to skip building the encoding tool (and thus its dependencies) then you can do so by turning off the BUILD_TOOLS flag:
cmake . -DBUILD_TOOLS=OFF
make
Installation can be performed in the following manner:
cmake . -DCMAKE_INSTALL_PREFIX=<install_prefix>
make
make install
The library has no dependecies.
Building the encoding tool requires ImageMagick++ and libsndfile.
To install these packages in Ubuntu:
apt install libmagick++-dev libsndfile1-dev
To install these packages in ArchLinux:
pacman -S imagemagick libsndfile
The conversion tool takes four parameters: the desired mode, input image filename, output WAV filename and sampling rate. The last parameter can be omitted and defaults to 48000Hz.
Example usage:
./sstv-encode pd90 ../test/test-image.bmp test.wav 44800
The above call produces test.wav in the current directory.
Each library function returns an error code that it's advisable you check. A simple check for SSTV_OK is enough, unless you want to handle particular errors differently.
Library initialization is done with a call to sstv_init(alloc_func, dealloc_func), where the two parameters are function pointers to the allocation and deallocation functions you want libsstv to use internally (e.g. malloc and free).
If the library is not initialized it can still be used, but only a precompiled maximum number of encoders (DEFAULT_ENCODER_CONTEXT_COUNT, default value 4) can be used simultaneously. This number can be adjusted by passing it to cmake at compile time:
cmake . -DDEFAULT_ENCODER_CONTEXT_COUNT=8
If you only encode one image at a time it is safe to skip initialization.
Moreover, if you do not call sstv_init() you will not be able to use further APIs that would require memory allocation to be performed within (see the section on Images).
Before encoding and decoding, an image object must be created. The image buffer can either be allocated or it can be provided.
The quickest way is to create a valid image and allocate its buffer is to call sstv_create_image_from_mode(). This function requires that a call to sstv_init() has been made (see Initializetion above).
sstv_image_t image;
if (sstv_create_image_from_mode(&image, SSTV_MODE_FAX480) != SSTV_OK) {
... error handling ...
}
... write to the image buffer ...
Otherwise, the user would start by retrieving the image properties of a specific mode:
uint32_t width, height;
sstv_image_format_t format;
if (sstv_get_mode_image_props(SSTV_MODE_FAX480, &width, &height, &format) != SSTV_OK) {
... error handling ...
}
and either rely on sstv_create_image_from_props() (which also relies on sstv_init()) to allocate an image buffer:
sstv_image_t image;
if (sstv_create_image_from_props(&image, width, height, format) != SSTV_OK) {
... error handling ...
}
... write to the image buffer ...
or use an existing buffer to initialize an image structure with sstv_pack_image():
uint8_t *buffer = ...;
sstv_image_t image;
if (sstv_pack_image(&image, width, height, format, buffer) != SSTV_OK) {
... error handling ...
}
Only images initialized with sstv_create_image_from_mode() and sstv_create_image_from_props() must be destroyed with sstv_delete_image().
Pixel format conversions can be performed in-place on an image by calling sstv_convert_image(), with the limitation that the target format must fit within the original memory (e.g. grayscale to RGB conversions will fail with SSTV_UNSUPPORTED_CONVERSION).
Signals (sstv_signal_t) are objects that hold a chunk of the raw audio. They can only be created on a preallocated buffer:
int16_t signal_buffer[SIGNAL_BUFFER_CAPACITY];
sstv_signal_t signal;
if (sstv_pack_signal(&signal, SSTV_SAMPLE_INT16, SIGNAL_BUFFER_CAPACITY, signal_buffer) != SSTV_OK) {
... error handling ...
}
You can use sstv_create_encoder() and sstv_delete_encoder() to request and return an encoder context to the library.
sstv_image_t image;
... set up image ...
const size_t SAMPLE_RATE = 48000
void *ctx;
if (sstv_create_encoder(&ctx, image, SSTV_MODE_FAX480, SAMPLE_RATE) != SSTV_OK) {
... error handling ...
}
... encode the data ...
if (sstv_delete_encoder(ctx) != SSTV_OK) {
... error handling ...
}
The actual encoding is performed with multiple calls to sstv_encode(), until the whole output has been produced:
void *ctx;
... encoder context creation ...
sstv_signal_t signal;
... signal packing ...
sstv_error_t rc;
do {
rc = sstv_encode(ctx, &signal);
if (rc != SSTV_ENCODE_SUCCESSFUL && rc != SSTV_ENCODE_END) {
... error handling ...
}
} while (rc != SSTV_ENCODE_END);
... encoder context disposal ...
Note that sstv_encode() does not return SSTV_OK on success but SSTV_ENCODE_SUCCESSFUL.
The library does not allocate further memory than that allocated for the images or that provided by the user via images or signals.
Copyright (c) 2018-2023 Vasile Vilvoiu (YO7JBP) [email protected]
libsstv is free software; you can redistribute it and/or modify it under the terms of the MIT license. See LICENSE for details.
Taywee/args library by Taylor C. Richberger and Pavel Belikov, released under the MIT license.