Merge pull request #65 from DarthAffe/sdcppUpdate

Updated sd.cpp

Header/stable-diffusion.h

@@ -48,12 +48,10 @@ enum sample_method_t {
     LCM_SAMPLE_METHOD,
     DDIM_TRAILING_SAMPLE_METHOD,
     TCD_SAMPLE_METHOD,
-			
     SAMPLE_METHOD_COUNT
 };
 
 enum scheduler_t {
-			
     DISCRETE_SCHEDULER,
     KARRAS_SCHEDULER,
     EXPONENTIAL_SCHEDULER,
@@ -73,6 +71,7 @@ enum prediction_t {
     EDM_V_PRED,
     SD3_FLOW_PRED,
     FLUX_FLOW_PRED,
+    FLUX2_FLOW_PRED,
     PREDICTION_COUNT
 };
 
@@ -158,8 +157,8 @@ typedef struct {
     const char* clip_g_path;
     const char* clip_vision_path;
     const char* t5xxl_path;
-    const char* qwen2vl_path;
+    const char* llm_path;
-    const char* qwen2vl_vision_path;
+    const char* llm_vision_path;
     const char* diffusion_model_path;
     const char* high_noise_diffusion_model_path;
     const char* vae_path;
@@ -284,11 +283,11 @@ typedef struct sd_ctx_t sd_ctx_t;
 
 typedef void (*sd_log_cb_t)(enum sd_log_level_t level, const char* text, void* data);
 typedef void (*sd_progress_cb_t)(int step, int steps, float time, void* data);
-typedef void (*sd_preview_cb_t)(int step, int frame_count, sd_image_t* frames, bool is_noisy);
+typedef void (*sd_preview_cb_t)(int step, int frame_count, sd_image_t* frames, bool is_noisy, void* data);
 
 SD_API void sd_set_log_callback(sd_log_cb_t sd_log_cb, void* data);
 SD_API void sd_set_progress_callback(sd_progress_cb_t cb, void* data);
-SD_API void sd_set_preview_callback(sd_preview_cb_t cb, enum preview_t mode, int interval, bool denoised, bool noisy);
+SD_API void sd_set_preview_callback(sd_preview_cb_t cb, enum preview_t mode, int interval, bool denoised, bool noisy, void* data);
 SD_API int32_t get_num_physical_cores();
 SD_API const char* sd_get_system_info();
 
@@ -314,7 +313,6 @@ SD_API char* sd_ctx_params_to_str(const sd_ctx_params_t* sd_ctx_params);
 
 SD_API sd_ctx_t* new_sd_ctx(const sd_ctx_params_t* sd_ctx_params);
 SD_API void free_sd_ctx(sd_ctx_t* sd_ctx);
-																				 
 
 SD_API void sd_sample_params_init(sd_sample_params_t* sample_params);
 SD_API char* sd_sample_params_to_str(const sd_sample_params_t* sample_params);
@@ -360,4 +358,4 @@ SD_API bool preprocess_canny(sd_image_t image,
 }
 #endif
 
-#endif  // __STABLE_DIFFUSION_H__
+#endif  // __STABLE_DIFFUSION_H__

StableDiffusion.NET/Enums/Prediction.cs

@@ -7,5 +7,6 @@ public enum Prediction
     V,
     EDM_V,
     SD3Flow,
-    FluxFlow
+    FluxFlow,
+    Flux2Flow
 }

StableDiffusion.NET/Models/Parameter/DiffusionModelParameter.cs

@@ -1,4 +1,5 @@
-using JetBrains.Annotations;
+using System;
+using JetBrains.Annotations;
 
 namespace StableDiffusion.NET;
 
@@ -139,9 +140,13 @@ public sealed class DiffusionModelParameter
     /// </summary>
     public string T5xxlPath { get; set; } = string.Empty;
 
-    public string Qwen2VLPath { get; set; } = string.Empty;
+    [Obsolete("Use LLMPath instead")]
+    public string Qwen2VLPath { get => LLMPath; set => LLMPath = value; }
+    public string LLMPath { get; set; } = string.Empty;
 
-    public string Qwen2VLVisionPath { get; set; } = string.Empty;
+    [Obsolete("Use LLMVisionPath instead")]
+    public string Qwen2VLVisionPath { get => LLMVisionPath; set => LLMVisionPath = value; }
+    public string LLMVisionPath { get; set; } = string.Empty;
 
     public string ClipVisionPath { get; set; } = string.Empty;
     public string HighNoiseDiffusionModelPath { get; set; } = string.Empty;

StableDiffusion.NET/Models/Parameter/Extensions/DiffusionModelBuilderExtension.cs

@@ -236,16 +236,20 @@ public static class DiffusionModelBuilderExtension
         return parameter;
     }
 
-    public static DiffusionModelParameter WithQwen2VLPath(this DiffusionModelParameter parameter, string qwen2VLPath)
+    [Obsolete("Use WithLLMPath instead")]
+    public static DiffusionModelParameter WithQwen2VLPath(this DiffusionModelParameter parameter, string qwen2VLPath) => parameter.WithLLMPath(qwen2VLPath);
+    public static DiffusionModelParameter WithLLMPath(this DiffusionModelParameter parameter, string llmPath)
     {
-        parameter.Qwen2VLPath = qwen2VLPath;
+        parameter.LLMPath = llmPath;
 
         return parameter;
     }
 
-    public static DiffusionModelParameter WithQwen2VLVisionPath(this DiffusionModelParameter parameter, string qwen2VLVisionPath)
+    [Obsolete("Use WithLLMVisionPath instead")]
+    public static DiffusionModelParameter WithQwen2VLVisionPath(this DiffusionModelParameter parameter, string qwen2VLVisionPath) => parameter.WithLLMVisionPath(qwen2VLVisionPath);
+    public static DiffusionModelParameter WithLLMVisionPath(this DiffusionModelParameter parameter, string llmVisionPath)
     {
-        parameter.Qwen2VLVisionPath = qwen2VLVisionPath;
+        parameter.LLMVisionPath = llmVisionPath;
 
         return parameter;
     }

StableDiffusion.NET/Native/Marshaller/DiffusionModelParameterMarshaller.cs

@@ -14,8 +14,8 @@ internal static unsafe class DiffusionModelParameterMarshaller
             clip_g_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.ClipGPath),
             clip_vision_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.ClipVisionPath),
             t5xxl_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.T5xxlPath),
-            qwen2vl_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.Qwen2VLPath),
+            llm_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.LLMPath),
-            qwen2vl_vision_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.Qwen2VLVisionPath),
+            llm_vision_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.LLMVisionPath),
             diffusion_model_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.DiffusionModelPath),
             high_noise_diffusion_model_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.HighNoiseDiffusionModelPath),
             vae_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.VaePath),
@@ -56,8 +56,8 @@ internal static unsafe class DiffusionModelParameterMarshaller
             ClipGPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.clip_g_path) ?? string.Empty,
             ClipVisionPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.clip_vision_path) ?? string.Empty,
             T5xxlPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.t5xxl_path) ?? string.Empty,
-            Qwen2VLPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.qwen2vl_path) ?? string.Empty,
+            LLMPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.llm_path) ?? string.Empty,
-            Qwen2VLVisionPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.qwen2vl_vision_path) ?? string.Empty,
+            LLMVisionPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.llm_vision_path) ?? string.Empty,
             DiffusionModelPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.diffusion_model_path) ?? string.Empty,
             HighNoiseDiffusionModelPath = AnsiStringMarshaller.ConvertToManaged(unmanaged.high_noise_diffusion_model_path) ?? string.Empty,
             VaePath = AnsiStringMarshaller.ConvertToManaged(unmanaged.vae_path) ?? string.Empty,
@@ -96,8 +96,8 @@ internal static unsafe class DiffusionModelParameterMarshaller
         AnsiStringMarshaller.Free(unmanaged.clip_l_path);
         AnsiStringMarshaller.Free(unmanaged.clip_g_path);
         AnsiStringMarshaller.Free(unmanaged.t5xxl_path);
-        AnsiStringMarshaller.Free(unmanaged.qwen2vl_path);
+        AnsiStringMarshaller.Free(unmanaged.llm_path);
-        AnsiStringMarshaller.Free(unmanaged.qwen2vl_vision_path);
+        AnsiStringMarshaller.Free(unmanaged.llm_vision_path);
         AnsiStringMarshaller.Free(unmanaged.diffusion_model_path);
         AnsiStringMarshaller.Free(unmanaged.vae_path);
         AnsiStringMarshaller.Free(unmanaged.taesd_path);

StableDiffusion.NET/Native/Native.cs

@@ -60,8 +60,8 @@ internal unsafe partial class Native
             public byte* clip_g_path;
             public byte* clip_vision_path;
             public byte* t5xxl_path;
-            public byte* qwen2vl_path;
+            public byte* llm_path;
-            public byte* qwen2vl_vision_path;
+            public byte* llm_vision_path;
             public byte* diffusion_model_path;
             public byte* high_noise_diffusion_model_path;
             public byte* vae_path;
@@ -207,7 +207,7 @@ internal unsafe partial class Native
 
     internal delegate void sd_log_cb_t(sd_log_level_t level, [MarshalAs(UnmanagedType.LPStr)] string text, void* data);
     internal delegate void sd_progress_cb_t(int step, int steps, float time, void* data);
-    internal delegate void sd_preview_cb_t(int step, int frame_count, sd_image_t* frames, bool is_noisy);
+    internal delegate void sd_preview_cb_t(int step, int frame_count, sd_image_t* frames, bool is_noisy, void* data);
 
     #endregion
 
@@ -220,7 +220,7 @@ internal unsafe partial class Native
     internal static partial void sd_set_progress_callback(sd_progress_cb_t cb, void* data);
 
     [LibraryImport(LIB_NAME, EntryPoint = "sd_set_preview_callback")]
-    internal static partial void sd_set_preview_callback(sd_preview_cb_t? cb, preview_t mode, int interval, [MarshalAs(UnmanagedType.I1)] bool denoised, [MarshalAs(UnmanagedType.I1)] bool noisy);
+    internal static partial void sd_set_preview_callback(sd_preview_cb_t? cb, preview_t mode, int interval, [MarshalAs(UnmanagedType.I1)] bool denoised, [MarshalAs(UnmanagedType.I1)] bool noisy, void* data);
 
     [LibraryImport(LIB_NAME, EntryPoint = "get_num_physical_cores")]
     internal static partial int32_t get_num_physical_cores();

StableDiffusion.NET/StableDiffusionCpp.cs

@@ -45,7 +45,7 @@ public static unsafe class StableDiffusionCpp
         else if (_previewCallback == null)
             _previewCallback = OnPreview;
 
-        Native.sd_set_preview_callback(_previewCallback, mode, interval, denoised, noisy);
+        Native.sd_set_preview_callback(_previewCallback, mode, interval, denoised, noisy, null);
     }
 
     public static void Convert(string modelPath, string vaePath, Quantization quantization, string outputPath, string tensorTypeRules = "")
@@ -104,7 +104,7 @@ public static unsafe class StableDiffusionCpp
         catch { /**/ }
     }
 
-    private static void OnPreview(int step, int frameCount, Native.Types.sd_image_t* frames, bool isNoisy)
+    private static void OnPreview(int step, int frameCount, Native.Types.sd_image_t* frames, bool isNoisy, void* data)
     {
         try
         {