mirror of
https://github.com/leejet/stable-diffusion.cpp.git
synced 2026-06-09 15:56:39 +00:00
67dda3f897
* add GemmaTokenizer * add basic ltx2.3 support * change vocab file encoding * fix ci * fix ubuntu build * add temporal tiling support * add ltx audio support * update ggml submodule url * fix generate_video * add i2v support * minify bundled Gemma tokenizer vocab sources * pass video fps into temporal rope embeddings * fix av_ca_timestep_scale_multiplier * add LTX2Scheduler support * update docs * fix ci
114 lines
5.2 KiB
C++
114 lines
5.2 KiB
C++
#ifndef __MEDIA_IO_H__
|
|
#define __MEDIA_IO_H__
|
|
|
|
#include <cstdint>
|
|
#include <string>
|
|
#include <vector>
|
|
|
|
#include "stable-diffusion.h"
|
|
|
|
enum class EncodedImageFormat {
|
|
JPEG,
|
|
PNG,
|
|
WEBP,
|
|
UNKNOWN,
|
|
};
|
|
|
|
EncodedImageFormat encoded_image_format_from_path(const std::string& path);
|
|
|
|
std::vector<uint8_t> encode_image_to_vector(EncodedImageFormat format,
|
|
const uint8_t* image,
|
|
int width,
|
|
int height,
|
|
int channels,
|
|
const std::string& parameters = "",
|
|
int quality = 90);
|
|
|
|
bool write_image_to_file(const std::string& path,
|
|
const uint8_t* image,
|
|
int width,
|
|
int height,
|
|
int channels,
|
|
const std::string& parameters = "",
|
|
int quality = 90);
|
|
|
|
uint8_t* load_image_from_file(const char* image_path,
|
|
int& width,
|
|
int& height,
|
|
int expected_width = 0,
|
|
int expected_height = 0,
|
|
int expected_channel = 3);
|
|
|
|
bool load_sd_image_from_file(sd_image_t* image,
|
|
const char* image_path,
|
|
int expected_width = 0,
|
|
int expected_height = 0,
|
|
int expected_channel = 3);
|
|
|
|
uint8_t* load_image_from_memory(const char* image_bytes,
|
|
int len,
|
|
int& width,
|
|
int& height,
|
|
int expected_width = 0,
|
|
int expected_height = 0,
|
|
int expected_channel = 3);
|
|
|
|
int create_mjpg_avi_from_sd_images(const char* filename,
|
|
sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90,
|
|
const sd_audio_t* audio = nullptr);
|
|
std::vector<uint8_t> create_mjpg_avi_from_sd_images_to_vector(sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90,
|
|
const sd_audio_t* audio = nullptr);
|
|
|
|
#ifdef SD_USE_WEBP
|
|
int create_animated_webp_from_sd_images(const char* filename,
|
|
sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90);
|
|
std::vector<uint8_t> create_animated_webp_from_sd_images_to_vector(sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90);
|
|
#endif
|
|
|
|
#ifdef SD_USE_WEBM
|
|
int create_webm_from_sd_images(const char* filename,
|
|
sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90,
|
|
const sd_audio_t* audio = nullptr);
|
|
std::vector<uint8_t> create_webm_from_sd_images_to_vector(sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90,
|
|
const sd_audio_t* audio = nullptr);
|
|
#endif
|
|
|
|
int create_video_from_sd_images(const char* filename,
|
|
sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90,
|
|
const sd_audio_t* audio = nullptr);
|
|
std::vector<uint8_t> create_video_from_sd_images_to_vector(const std::string& output_format,
|
|
sd_image_t* images,
|
|
int num_images,
|
|
int fps,
|
|
int quality = 90,
|
|
const sd_audio_t* audio = nullptr);
|
|
|
|
bool write_wav_to_file(const std::string& path,
|
|
const float* interleaved_samples,
|
|
uint64_t sample_count,
|
|
uint32_t channels,
|
|
uint32_t sample_rate);
|
|
|
|
#endif // __MEDIA_IO_H__
|