@runanywhere/core 0.17.8 → 0.18.0

package/ios/Binaries/RACommons.xcframework/ios-arm64_x86_64-simulator/RACommons.framework/Headers/rac_diffusion_types.h

@@ -0,0 +1,454 @@
+/**
+ * @file rac_diffusion_types.h
+ * @brief RunAnywhere Commons - Diffusion Types and Data Structures
+ *
+ * This header defines data structures for image generation using diffusion models
+ * (Stable Diffusion). Supports text-to-image, image-to-image, and inpainting.
+ *
+ * This header defines data structures only. For the service interface,
+ * see rac_diffusion_service.h.
+ */
+
+#ifndef RAC_DIFFUSION_TYPES_H
+#define RAC_DIFFUSION_TYPES_H
+
+#include <RACommons/rac_types.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// =============================================================================
+// SCHEDULER TYPES
+// =============================================================================
+
+/**
+ * @brief Diffusion scheduler/sampler types
+ *
+ * Different scheduling algorithms for the denoising process.
+ * DPM++ 2M Karras is recommended for best quality/speed tradeoff.
+ */
+typedef enum rac_diffusion_scheduler {
+    RAC_DIFFUSION_SCHEDULER_DPM_PP_2M_KARRAS = 0, /**< DPM++ 2M Karras (recommended) */
+    RAC_DIFFUSION_SCHEDULER_DPM_PP_2M = 1,        /**< DPM++ 2M */
+    RAC_DIFFUSION_SCHEDULER_DPM_PP_2M_SDE = 2,    /**< DPM++ 2M SDE */
+    RAC_DIFFUSION_SCHEDULER_DDIM = 3,             /**< DDIM */
+    RAC_DIFFUSION_SCHEDULER_EULER = 4,            /**< Euler */
+    RAC_DIFFUSION_SCHEDULER_EULER_ANCESTRAL = 5,  /**< Euler Ancestral */
+    RAC_DIFFUSION_SCHEDULER_PNDM = 6,             /**< PNDM */
+    RAC_DIFFUSION_SCHEDULER_LMS = 7,              /**< LMS */
+} rac_diffusion_scheduler_t;
+
+/**
+ * @brief Model variant types
+ *
+ * Different Stable Diffusion model variants with different capabilities.
+ */
+typedef enum rac_diffusion_model_variant {
+    RAC_DIFFUSION_MODEL_SD_1_5 = 0,     /**< Stable Diffusion 1.5 (512x512 default) */
+    RAC_DIFFUSION_MODEL_SD_2_1 = 1,     /**< Stable Diffusion 2.1 (768x768 default) */
+    RAC_DIFFUSION_MODEL_SDXL = 2,       /**< SDXL (1024x1024 default, requires 8GB+ RAM) */
+    RAC_DIFFUSION_MODEL_SDXL_TURBO = 3, /**< SDXL Turbo (fast, fewer steps, no CFG) */
+    RAC_DIFFUSION_MODEL_SDXS = 4,       /**< SDXS - Ultra-fast 1-step model (no CFG) */
+    RAC_DIFFUSION_MODEL_LCM = 5,        /**< LCM - Latent Consistency Model (4 steps) */
+} rac_diffusion_model_variant_t;
+
+/**
+ * @brief Generation mode
+ */
+typedef enum rac_diffusion_mode {
+    RAC_DIFFUSION_MODE_TEXT_TO_IMAGE = 0, /**< Generate image from text prompt */
+    RAC_DIFFUSION_MODE_IMAGE_TO_IMAGE = 1, /**< Transform input image with prompt */
+    RAC_DIFFUSION_MODE_INPAINTING = 2,     /**< Edit specific regions with mask */
+} rac_diffusion_mode_t;
+
+// =============================================================================
+// TOKENIZER CONFIGURATION
+// =============================================================================
+
+/**
+ * @brief Tokenizer source presets
+ *
+ * Predefined HuggingFace repository sources for tokenizer files.
+ * Apple's compiled CoreML models don't include tokenizer files (vocab.json, merges.txt),
+ * so they must be downloaded separately from HuggingFace.
+ *
+ * Developers can use RAC_DIFFUSION_TOKENIZER_CUSTOM with a custom_base_url
+ * to specify their own tokenizer source.
+ */
+typedef enum rac_diffusion_tokenizer_source {
+    /** Stable Diffusion 1.x tokenizer (CLIP ViT-L/14)
+     *  Source: runwayml/stable-diffusion-v1-5 */
+    RAC_DIFFUSION_TOKENIZER_SD_1_5 = 0,
+
+    /** Stable Diffusion 2.x tokenizer (OpenCLIP ViT-H/14)
+     *  Source: stabilityai/stable-diffusion-2-1 */
+    RAC_DIFFUSION_TOKENIZER_SD_2_X = 1,
+
+    /** Stable Diffusion XL tokenizer (dual tokenizers)
+     *  Source: stabilityai/stable-diffusion-xl-base-1.0 */
+    RAC_DIFFUSION_TOKENIZER_SDXL = 2,
+
+    /** Custom tokenizer from a developer-specified URL
+     *  Requires custom_base_url to be set in rac_diffusion_tokenizer_config_t */
+    RAC_DIFFUSION_TOKENIZER_CUSTOM = 99,
+} rac_diffusion_tokenizer_source_t;
+
+/**
+ * @brief Tokenizer configuration
+ *
+ * Configuration for downloading and using tokenizer files.
+ * The SDK will automatically download missing tokenizer files (vocab.json, merges.txt)
+ * from the specified source URL.
+ *
+ * Example for custom URL:
+ * @code
+ * rac_diffusion_tokenizer_config_t tokenizer_config = {
+ *     .source = RAC_DIFFUSION_TOKENIZER_CUSTOM,
+ *     .custom_base_url = "https://huggingface.co/my-org/my-model/resolve/main/tokenizer",
+ *     .auto_download = RAC_TRUE
+ * };
+ * @endcode
+ */
+typedef struct rac_diffusion_tokenizer_config {
+    /** Tokenizer source preset (SD15, SD21, SDXL, or CUSTOM) */
+    rac_diffusion_tokenizer_source_t source;
+
+    /** Custom base URL for tokenizer files (only used when source == CUSTOM)
+     *  Should be a URL directory containing vocab.json and merges.txt
+     *  Example: "https://huggingface.co/my-org/my-model/resolve/main/tokenizer"
+     *  The SDK will append "/vocab.json" and "/merges.txt" to download files */
+    const char* custom_base_url;
+
+    /** Automatically download missing tokenizer files (default: true) */
+    rac_bool_t auto_download;
+} rac_diffusion_tokenizer_config_t;
+
+/**
+ * @brief Default tokenizer configuration
+ */
+static const rac_diffusion_tokenizer_config_t RAC_DIFFUSION_TOKENIZER_CONFIG_DEFAULT = {
+    .source = RAC_DIFFUSION_TOKENIZER_SD_1_5,
+    .custom_base_url = RAC_NULL,
+    .auto_download = RAC_TRUE};
+
+// =============================================================================
+// CONFIGURATION - Component configuration
+// =============================================================================
+
+/**
+ * @brief Diffusion component configuration
+ *
+ * Configuration for initializing the diffusion component.
+ */
+typedef struct rac_diffusion_config {
+    /** Model ID (optional - uses default if NULL) */
+    const char* model_id;
+
+    /** Preferred framework (use RAC_FRAMEWORK_UNKNOWN for auto) */
+    int32_t preferred_framework;
+
+    /** Model variant (SD 1.5, SD 2.1, SDXL, etc.) */
+    rac_diffusion_model_variant_t model_variant;
+
+    /** Enable safety checker for NSFW content filtering (default: true) */
+    rac_bool_t enable_safety_checker;
+
+    /** Reduce memory footprint (may reduce quality, default: false) */
+    rac_bool_t reduce_memory;
+
+    /** Tokenizer configuration for downloading missing tokenizer files
+     *  Apple's compiled CoreML models don't include tokenizer files */
+    rac_diffusion_tokenizer_config_t tokenizer;
+} rac_diffusion_config_t;
+
+/**
+ * @brief Default diffusion configuration
+ */
+static const rac_diffusion_config_t RAC_DIFFUSION_CONFIG_DEFAULT = {
+    .model_id = RAC_NULL,
+    .preferred_framework = 99, // RAC_FRAMEWORK_UNKNOWN
+    .model_variant = RAC_DIFFUSION_MODEL_SD_1_5,
+    .enable_safety_checker = RAC_TRUE,
+    .reduce_memory = RAC_FALSE,
+    .tokenizer = {.source = RAC_DIFFUSION_TOKENIZER_SD_1_5,
+                  .custom_base_url = RAC_NULL,
+                  .auto_download = RAC_TRUE}};
+
+// =============================================================================
+// OPTIONS - Generation options
+// =============================================================================
+
+/**
+ * @brief Diffusion generation options
+ *
+ * Options for controlling image generation.
+ */
+typedef struct rac_diffusion_options {
+    /** Text prompt describing the desired image */
+    const char* prompt;
+
+    /** Negative prompt - things to avoid in the image (can be NULL) */
+    const char* negative_prompt;
+
+    /** Output image width in pixels (default: 512 for SD 1.5, 1024 for SDXL) */
+    int32_t width;
+
+    /** Output image height in pixels (default: 512 for SD 1.5, 1024 for SDXL) */
+    int32_t height;
+
+    /** Number of denoising steps (default: 28, range: 10-50) */
+    int32_t steps;
+
+    /** Classifier-free guidance scale (default: 7.5, range: 1.0-20.0) */
+    float guidance_scale;
+
+    /** Random seed for reproducibility (-1 for random, default: -1) */
+    int64_t seed;
+
+    /** Scheduler/sampler algorithm (default: DPM++ 2M Karras) */
+    rac_diffusion_scheduler_t scheduler;
+
+    // --- Image-to-image / Inpainting options ---
+
+    /** Generation mode (text-to-image, img2img, inpainting) */
+    rac_diffusion_mode_t mode;
+
+    /** Input image RGBA data for img2img/inpainting (can be NULL) */
+    const uint8_t* input_image_data;
+
+    /** Input image data size in bytes */
+    size_t input_image_size;
+
+    /** Input image width (required if input_image_data is set) */
+    int32_t input_image_width;
+
+    /** Input image height (required if input_image_data is set) */
+    int32_t input_image_height;
+
+    /** Mask image data for inpainting - grayscale (can be NULL) */
+    const uint8_t* mask_data;
+
+    /** Mask data size in bytes */
+    size_t mask_size;
+
+    /** Denoising strength for img2img (0.0-1.0, default: 0.75) */
+    float denoise_strength;
+
+    // --- Progress reporting options ---
+
+    /** Report intermediate images during generation (default: false) */
+    rac_bool_t report_intermediate_images;
+
+    /** Report progress every N steps (default: 1) */
+    int32_t progress_stride;
+} rac_diffusion_options_t;
+
+/**
+ * @brief Default diffusion generation options
+ */
+static const rac_diffusion_options_t RAC_DIFFUSION_OPTIONS_DEFAULT = {
+    .prompt = RAC_NULL,
+    .negative_prompt = RAC_NULL,
+    .width = 512,
+    .height = 512,
+    .steps = 28,
+    .guidance_scale = 7.5f,
+    .seed = -1,
+    .scheduler = RAC_DIFFUSION_SCHEDULER_DPM_PP_2M_KARRAS,
+    .mode = RAC_DIFFUSION_MODE_TEXT_TO_IMAGE,
+    .input_image_data = RAC_NULL,
+    .input_image_size = 0,
+    .input_image_width = 0,
+    .input_image_height = 0,
+    .mask_data = RAC_NULL,
+    .mask_size = 0,
+    .denoise_strength = 0.75f,
+    .report_intermediate_images = RAC_FALSE,
+    .progress_stride = 1};
+
+// =============================================================================
+// PROGRESS - Generation progress
+// =============================================================================
+
+/**
+ * @brief Diffusion generation progress
+ *
+ * Reports progress during image generation.
+ */
+typedef struct rac_diffusion_progress {
+    /** Progress percentage (0.0 - 1.0) */
+    float progress;
+
+    /** Current step number (1-based) */
+    int32_t current_step;
+
+    /** Total number of steps */
+    int32_t total_steps;
+
+    /** Current stage description (e.g., "Encoding", "Denoising", "Decoding") */
+    const char* stage;
+
+    /** Intermediate image RGBA data (can be NULL if not requested) */
+    const uint8_t* intermediate_image_data;
+
+    /** Intermediate image data size */
+    size_t intermediate_image_size;
+
+    /** Intermediate image width */
+    int32_t intermediate_image_width;
+
+    /** Intermediate image height */
+    int32_t intermediate_image_height;
+} rac_diffusion_progress_t;
+
+// =============================================================================
+// RESULT - Generation result
+// =============================================================================
+
+/**
+ * @brief Diffusion generation result
+ *
+ * Contains the generated image and metadata.
+ */
+typedef struct rac_diffusion_result {
+    /** Generated image RGBA data (owned, must be freed with rac_diffusion_result_free) */
+    uint8_t* image_data;
+
+    /** Image data size in bytes */
+    size_t image_size;
+
+    /** Image width in pixels */
+    int32_t width;
+
+    /** Image height in pixels */
+    int32_t height;
+
+    /** Seed used for generation (useful for reproducibility) */
+    int64_t seed_used;
+
+    /** Total generation time in milliseconds */
+    int64_t generation_time_ms;
+
+    /** Whether the image was flagged by safety checker */
+    rac_bool_t safety_flagged;
+
+    /** Error code if generation failed (RAC_SUCCESS on success) */
+    rac_result_t error_code;
+
+    /** Error message if generation failed (can be NULL) */
+    char* error_message;
+} rac_diffusion_result_t;
+
+// =============================================================================
+// INFO - Service information
+// =============================================================================
+
+/**
+ * @brief Diffusion service information
+ *
+ * Information about the loaded diffusion service.
+ */
+typedef struct rac_diffusion_info {
+    /** Whether the service is ready for generation */
+    rac_bool_t is_ready;
+
+    /** Current model identifier (can be NULL) */
+    const char* current_model;
+
+    /** Model variant */
+    rac_diffusion_model_variant_t model_variant;
+
+    /** Whether text-to-image is supported */
+    rac_bool_t supports_text_to_image;
+
+    /** Whether image-to-image is supported */
+    rac_bool_t supports_image_to_image;
+
+    /** Whether inpainting is supported */
+    rac_bool_t supports_inpainting;
+
+    /** Whether safety checker is enabled */
+    rac_bool_t safety_checker_enabled;
+
+    /** Maximum supported width */
+    int32_t max_width;
+
+    /** Maximum supported height */
+    int32_t max_height;
+} rac_diffusion_info_t;
+
+// =============================================================================
+// CALLBACKS
+// =============================================================================
+
+/**
+ * @brief Diffusion progress callback
+ *
+ * Called during generation to report progress.
+ *
+ * @param progress Progress information
+ * @param user_data User-provided context
+ * @return RAC_TRUE to continue, RAC_FALSE to cancel generation
+ */
+typedef rac_bool_t (*rac_diffusion_progress_callback_fn)(const rac_diffusion_progress_t* progress,
+                                                         void* user_data);
+
+/**
+ * @brief Diffusion completion callback
+ *
+ * Called when generation completes successfully.
+ *
+ * @param result Generation result
+ * @param user_data User-provided context
+ */
+typedef void (*rac_diffusion_complete_callback_fn)(const rac_diffusion_result_t* result,
+                                                   void* user_data);
+
+/**
+ * @brief Diffusion error callback
+ *
+ * Called when generation fails.
+ *
+ * @param error_code Error code
+ * @param error_message Error message
+ * @param user_data User-provided context
+ */
+typedef void (*rac_diffusion_error_callback_fn)(rac_result_t error_code, const char* error_message,
+                                                void* user_data);
+
+// =============================================================================
+// CAPABILITY FLAGS
+// =============================================================================
+
+/** Supports text-to-image generation */
+#define RAC_DIFFUSION_CAP_TEXT_TO_IMAGE (1 << 0)
+
+/** Supports image-to-image transformation */
+#define RAC_DIFFUSION_CAP_IMAGE_TO_IMAGE (1 << 1)
+
+/** Supports inpainting with mask */
+#define RAC_DIFFUSION_CAP_INPAINTING (1 << 2)
+
+/** Supports intermediate image reporting */
+#define RAC_DIFFUSION_CAP_INTERMEDIATE_IMAGES (1 << 3)
+
+/** Has safety checker */
+#define RAC_DIFFUSION_CAP_SAFETY_CHECKER (1 << 4)
+
+// =============================================================================
+// MEMORY MANAGEMENT
+// =============================================================================
+
+/**
+ * @brief Free diffusion result resources
+ *
+ * @param result Result to free (can be NULL)
+ */
+RAC_API void rac_diffusion_result_free(rac_diffusion_result_t* result);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* RAC_DIFFUSION_TYPES_H */

package/ios/Binaries/RACommons.xcframework/ios-arm64_x86_64-simulator/RACommons.framework/Headers/rac_endpoints.h

@@ -62,24 +62,10 @@ const char* rac_endpoint_device_registration(rac_environment_t env);
 const char* rac_endpoint_telemetry(rac_environment_t env);
 
 /**
- * @brief Build model assignments endpoint with query params
- * @param device_type Device type (e.g., "iphone", "android_phone")
- * @param platform Platform (e.g., "ios", "android")
- * @param out_buffer Buffer to write endpoint path
- * @param buffer_size Size of buffer
- * @return Length of written string, or -1 on error
- */
-int rac_endpoint_model_assignments(const char* device_type, const char* platform, char* out_buffer,
-                                   size_t buffer_size);
-
-/**
- * @brief Build model download endpoint
- * @param model_id The model ID
- * @param out_buffer Buffer to write endpoint path
- * @param buffer_size Size of buffer
- * @return Length of written string, or -1 on error
+ * @brief Get model assignments endpoint
+ * @return Endpoint path string
  */
-int rac_endpoint_model_download(const char* model_id, char* out_buffer, size_t buffer_size);
+const char* rac_endpoint_model_assignments(void);
 
 // =============================================================================
 // Full URL Building

package/ios/Binaries/RACommons.xcframework/ios-arm64_x86_64-simulator/RACommons.framework/Headers/rac_image_utils.h

@@ -0,0 +1,215 @@
+/**
+ * @file rac_image_utils.h
+ * @brief RunAnywhere Commons - Image Utilities
+ *
+ * Image loading and processing utilities for VLM backends.
+ * Supports loading from file paths, decoding base64, and resizing.
+ */
+
+#ifndef RAC_IMAGE_UTILS_H
+#define RAC_IMAGE_UTILS_H
+
+#include <RACommons/rac_error.h>
+#include <RACommons/rac_types.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// =============================================================================
+// IMAGE DATA STRUCTURES
+// =============================================================================
+
+/**
+ * @brief Loaded image data
+ *
+ * Contains RGB pixel data after loading an image.
+ * Must be freed with rac_image_free().
+ */
+typedef struct rac_image_data {
+    /** Raw RGB pixel data (RGBRGBRGB...) */
+    uint8_t* pixels;
+
+    /** Image width in pixels */
+    int32_t width;
+
+    /** Image height in pixels */
+    int32_t height;
+
+    /** Number of channels (3 for RGB) */
+    int32_t channels;
+
+    /** Total size in bytes (width * height * channels) */
+    size_t size;
+} rac_image_data_t;
+
+/**
+ * @brief Normalized float image data
+ *
+ * Contains normalized float32 pixel data (values in [-1, 1] or [0, 1]).
+ * Used by vision encoders.
+ */
+typedef struct rac_image_float {
+    /** Normalized float pixel data */
+    float* pixels;
+
+    /** Image width in pixels */
+    int32_t width;
+
+    /** Image height in pixels */
+    int32_t height;
+
+    /** Number of channels (3 for RGB) */
+    int32_t channels;
+
+    /** Total number of floats (width * height * channels) */
+    size_t count;
+} rac_image_float_t;
+
+// =============================================================================
+// IMAGE LOADING
+// =============================================================================
+
+/**
+ * @brief Load an image from a file path
+ *
+ * Supports JPEG, PNG, BMP, GIF, and other common formats via stb_image.
+ * Output is always RGB (3 channels).
+ *
+ * @param file_path Path to the image file
+ * @param out_image Output: Loaded image data (must be freed with rac_image_free)
+ * @return RAC_SUCCESS or error code
+ */
+RAC_API rac_result_t rac_image_load_file(const char* file_path, rac_image_data_t* out_image);
+
+/**
+ * @brief Decode a base64-encoded image
+ *
+ * Decodes base64 data and loads the image.
+ * Supports the same formats as rac_image_load_file.
+ *
+ * @param base64_data Base64-encoded image data
+ * @param data_size Length of the base64 string
+ * @param out_image Output: Loaded image data (must be freed with rac_image_free)
+ * @return RAC_SUCCESS or error code
+ */
+RAC_API rac_result_t rac_image_decode_base64(const char* base64_data, size_t data_size,
+                                             rac_image_data_t* out_image);
+
+/**
+ * @brief Decode image from raw bytes
+ *
+ * Decodes an image from raw bytes (e.g., from network response).
+ *
+ * @param data Raw image data (JPEG, PNG, etc.)
+ * @param data_size Size of the data in bytes
+ * @param out_image Output: Loaded image data (must be freed with rac_image_free)
+ * @return RAC_SUCCESS or error code
+ */
+RAC_API rac_result_t rac_image_decode_bytes(const uint8_t* data, size_t data_size,
+                                            rac_image_data_t* out_image);
+
+// =============================================================================
+// IMAGE PROCESSING
+// =============================================================================
+
+/**
+ * @brief Resize an image
+ *
+ * Resizes the image to the specified dimensions using bilinear interpolation.
+ *
+ * @param image Input image
+ * @param new_width Target width
+ * @param new_height Target height
+ * @param out_image Output: Resized image (must be freed with rac_image_free)
+ * @return RAC_SUCCESS or error code
+ */
+RAC_API rac_result_t rac_image_resize(const rac_image_data_t* image, int32_t new_width,
+                                      int32_t new_height, rac_image_data_t* out_image);
+
+/**
+ * @brief Resize an image maintaining aspect ratio
+ *
+ * Resizes the image so that the longest dimension equals max_size.
+ * Aspect ratio is preserved.
+ *
+ * @param image Input image
+ * @param max_size Maximum dimension (width or height)
+ * @param out_image Output: Resized image (must be freed with rac_image_free)
+ * @return RAC_SUCCESS or error code
+ */
+RAC_API rac_result_t rac_image_resize_max(const rac_image_data_t* image, int32_t max_size,
+                                          rac_image_data_t* out_image);
+
+/**
+ * @brief Normalize image to float values
+ *
+ * Converts uint8 pixels to float32 with optional mean/std normalization.
+ * Commonly used for vision encoders (CLIP, SigLIP, etc.).
+ *
+ * Formula: pixel_normalized = (pixel / 255.0 - mean) / std
+ *
+ * @param image Input image
+ * @param mean Per-channel mean values (array of 3 floats, or NULL for [0,0,0])
+ * @param std Per-channel std values (array of 3 floats, or NULL for [1,1,1])
+ * @param out_float Output: Normalized float image (must be freed with rac_image_float_free)
+ * @return RAC_SUCCESS or error code
+ */
+RAC_API rac_result_t rac_image_normalize(const rac_image_data_t* image, const float* mean,
+                                         const float* std, rac_image_float_t* out_float);
+
+/**
+ * @brief Convert RGB to CHW format
+ *
+ * Converts from HWC (Height, Width, Channels) to CHW format.
+ * Many neural networks expect CHW format.
+ *
+ * @param image Input float image in HWC format
+ * @param out_chw Output: Float image in CHW format (must be freed with rac_image_float_free)
+ * @return RAC_SUCCESS or error code
+ */
+RAC_API rac_result_t rac_image_to_chw(const rac_image_float_t* image, rac_image_float_t* out_chw);
+
+// =============================================================================
+// MEMORY MANAGEMENT
+// =============================================================================
+
+/**
+ * @brief Free image data
+ *
+ * Frees the pixel data allocated by image loading functions.
+ *
+ * @param image Image to free (can be NULL)
+ */
+RAC_API void rac_image_free(rac_image_data_t* image);
+
+/**
+ * @brief Free float image data
+ *
+ * Frees the pixel data allocated by normalization functions.
+ *
+ * @param image Float image to free (can be NULL)
+ */
+RAC_API void rac_image_float_free(rac_image_float_t* image);
+
+// =============================================================================
+// UTILITY FUNCTIONS
+// =============================================================================
+
+/**
+ * @brief Calculate resized dimensions maintaining aspect ratio
+ *
+ * @param width Original width
+ * @param height Original height
+ * @param max_size Maximum dimension
+ * @param out_width Output: New width
+ * @param out_height Output: New height
+ */
+RAC_API void rac_image_calc_resize(int32_t width, int32_t height, int32_t max_size,
+                                   int32_t* out_width, int32_t* out_height);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* RAC_IMAGE_UTILS_H */