updated sd.cpp

Header/stable-diffusion.h

@@ -31,37 +31,39 @@ extern "C" {
 enum rng_type_t {
     STD_DEFAULT_RNG,
     CUDA_RNG,
+    CPU_RNG,
     RNG_TYPE_COUNT
 };
 
 enum sample_method_t {
-    SAMPLE_METHOD_DEFAULT,
+    EULER_SAMPLE_METHOD,
-    EULER,
+    EULER_A_SAMPLE_METHOD,
-    HEUN,
+    HEUN_SAMPLE_METHOD,
-    DPM2,
+    DPM2_SAMPLE_METHOD,
-    DPMPP2S_A,
+    DPMPP2S_A_SAMPLE_METHOD,
-    DPMPP2M,
+    DPMPP2M_SAMPLE_METHOD,
-    DPMPP2Mv2,
+    DPMPP2Mv2_SAMPLE_METHOD,
-    IPNDM,
+    IPNDM_SAMPLE_METHOD,
-    IPNDM_V,
+    IPNDM_V_SAMPLE_METHOD,
-    LCM,
+    LCM_SAMPLE_METHOD,
-    DDIM_TRAILING,
+    DDIM_TRAILING_SAMPLE_METHOD,
-    TCD,
+    TCD_SAMPLE_METHOD,
-    EULER_A,
+			
     SAMPLE_METHOD_COUNT
 };
 
 enum scheduler_t {
-    DEFAULT,
+			
-    DISCRETE,
+    DISCRETE_SCHEDULER,
-    KARRAS,
+    KARRAS_SCHEDULER,
-    EXPONENTIAL,
+    EXPONENTIAL_SCHEDULER,
-    AYS,
+    AYS_SCHEDULER,
-    GITS,
+    GITS_SCHEDULER,
-    SGM_UNIFORM,
+    SGM_UNIFORM_SCHEDULER,
-    SIMPLE,
+    SIMPLE_SCHEDULER,
-    SMOOTHSTEP,
+    SMOOTHSTEP_SCHEDULER,
-    SCHEDULE_COUNT
+    LCM_SCHEDULER,
+    SCHEDULER_COUNT
 };
 
 enum prediction_t {
@@ -166,11 +168,13 @@ typedef struct {
     const char* lora_model_dir;
     const char* embedding_dir;
     const char* photo_maker_path;
+    const char* tensor_type_rules;
     bool vae_decode_only;
     bool free_params_immediately;
     int n_threads;
     enum sd_type_t wtype;
     enum rng_type_t rng_type;
+    enum rng_type_t sampler_rng_type;
     enum prediction_t prediction;
     enum lora_apply_mode_t lora_apply_mode;
     bool offload_params_to_cpu;
@@ -226,6 +230,13 @@ typedef struct {
     float style_strength;
 } sd_pm_params_t;  // photo maker
 
+typedef struct {
+    bool enabled;
+    float reuse_threshold;
+    float start_percent;
+    float end_percent;
+} sd_easycache_params_t;
+
 typedef struct {
     const char* prompt;
     const char* negative_prompt;
@@ -246,6 +257,7 @@ typedef struct {
     float control_strength;
     sd_pm_params_t pm_params;
     sd_tiling_params_t vae_tiling_params;
+    sd_easycache_params_t easycache;
 } sd_img_gen_params_t;
 
 typedef struct {
@@ -265,6 +277,7 @@ typedef struct {
     int64_t seed;
     int video_frames;
     float vace_strength;
+    sd_easycache_params_t easycache;
 } sd_vid_gen_params_t;
 
 typedef struct sd_ctx_t sd_ctx_t;
@@ -285,8 +298,8 @@ SD_API const char* sd_rng_type_name(enum rng_type_t rng_type);
 SD_API enum rng_type_t str_to_rng_type(const char* str);
 SD_API const char* sd_sample_method_name(enum sample_method_t sample_method);
 SD_API enum sample_method_t str_to_sample_method(const char* str);
-SD_API const char* sd_schedule_name(enum scheduler_t scheduler);
+SD_API const char* sd_scheduler_name(enum scheduler_t scheduler);
-SD_API enum scheduler_t str_to_schedule(const char* str);
+SD_API enum scheduler_t str_to_scheduler(const char* str);
 SD_API const char* sd_prediction_name(enum prediction_t prediction);
 SD_API enum prediction_t str_to_prediction(const char* str);
 SD_API const char* sd_preview_name(enum preview_t preview);
@@ -294,16 +307,21 @@ SD_API enum preview_t str_to_preview(const char* str);
 SD_API const char* sd_lora_apply_mode_name(enum lora_apply_mode_t mode);
 SD_API enum lora_apply_mode_t str_to_lora_apply_mode(const char* str);
 
+SD_API void sd_easycache_params_init(sd_easycache_params_t* easycache_params);
+
 SD_API void sd_ctx_params_init(sd_ctx_params_t* sd_ctx_params);
 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 enum sample_method_t sd_get_default_sample_method(const 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);
 
+SD_API enum sample_method_t sd_get_default_sample_method(const sd_ctx_t* sd_ctx);
+SD_API enum scheduler_t sd_get_default_scheduler(const sd_ctx_t* sd_ctx);
+
 SD_API void sd_img_gen_params_init(sd_img_gen_params_t* sd_img_gen_params);
 SD_API char* sd_img_gen_params_to_str(const sd_img_gen_params_t* sd_img_gen_params);
 SD_API sd_image_t* generate_image(sd_ctx_t* sd_ctx, const sd_img_gen_params_t* sd_img_gen_params);
@@ -342,4 +360,4 @@ SD_API bool preprocess_canny(sd_image_t image,
 }
 #endif
 
-#endif  // __STABLE_DIFFUSION_H__
+#endif  // __STABLE_DIFFUSION_H__

StableDiffusion.NET/Enums/RngType.cs

@@ -3,5 +3,6 @@
 public enum RngType
 {
     Standard,
-    Cuda
+    Cuda,
+    Cpu
 }

StableDiffusion.NET/Enums/Sampler.cs

@@ -2,8 +2,8 @@
 
 public enum Sampler
 {
-    Default,
     Euler,
+    Euler_A,
     Heun,
     DPM2,
     DPMPP2SA,
@@ -14,5 +14,5 @@ public enum Sampler
     LCM,
     DDIM_Trailing,
     TCD,
-    Euler_A,
+    Default
 }

StableDiffusion.NET/Enums/Schedule.cs

@@ -2,7 +2,6 @@
 
 public enum Scheduler
 {
-    Default,
     Discrete,
     Karras,
     Exponential,
@@ -10,5 +9,7 @@ public enum Scheduler
     GITS,
     SGM_Uniform,
     Simple,
-    Smoothstep
+    Smoothstep,
+    LCM,
+    Default
 }

StableDiffusion.NET/Models/Parameter/DiffusionModelParameter.cs

@@ -91,6 +91,8 @@ public sealed class DiffusionModelParameter
     /// </summary>
     public RngType RngType { get; set; } = RngType.Standard;
 
+    public RngType SamplerRngType { get; set; } = RngType.Standard;
+
     public Prediction Prediction { get; set; } = Prediction.Default;
 
     public LoraApplyMode LoraApplyMode { get; set; } = LoraApplyMode.Auto;
@@ -110,6 +112,8 @@ public sealed class DiffusionModelParameter
     /// </summary>
     public string StackedIdEmbeddingsDirectory { get; set; } = string.Empty;
 
+    public string TensorTypeRules { get; set; } = string.Empty;
+
     /// <summary>
     /// path to full model
     /// </summary>

StableDiffusion.NET/Models/Parameter/EasyCache.cs

@@ -0,0 +1,11 @@
+namespace StableDiffusion.NET;
+
+public sealed class EasyCache
+{
+    public bool IsEnabled { get; set; }
+    public float ReuseThreshold { get; set; }
+    public float StartPercent { get; set; }
+    public float EndPercent { get; set; }
+
+    internal EasyCache() { }
+}

StableDiffusion.NET/Models/Parameter/ImageGenerationParameter.cs

@@ -59,6 +59,8 @@ public sealed class ImageGenerationParameter
 
     public TilingParameter VaeTiling { get; } = new();
 
+    public EasyCache EasyCache { get; } = new();
+
     #endregion
 
     public static ImageGenerationParameter Create() => new();

StableDiffusion.NET/Models/Parameter/VideoGenerationParameter.cs

@@ -38,6 +38,8 @@ public sealed class VideoGenerationParameter
 
     public float VaceStrength { get; set; }
 
+    public EasyCache EasyCache { get; } = new();
+
     #endregion
 
     public static VideoGenerationParameter Create() => new();

StableDiffusion.NET/Native/Marshaller/DiffusionModelParameterMarshaller.cs

@@ -24,11 +24,13 @@ internal static unsafe class DiffusionModelParameterMarshaller
             lora_model_dir = AnsiStringMarshaller.ConvertToUnmanaged(managed.LoraModelDirectory),
             embedding_dir = AnsiStringMarshaller.ConvertToUnmanaged(managed.EmbeddingsDirectory),
             photo_maker_path = AnsiStringMarshaller.ConvertToUnmanaged(managed.StackedIdEmbeddingsDirectory),
+            tensor_type_rules = AnsiStringMarshaller.ConvertToUnmanaged(managed.TensorTypeRules),
             vae_decode_only = (sbyte)(managed.VaeDecodeOnly ? 1 : 0),
             free_params_immediately = (sbyte)(managed.FreeParamsImmediately ? 1 : 0),
             n_threads = managed.ThreadCount,
             wtype = managed.Quantization,
             rng_type = managed.RngType,
+            sampler_rng_type = managed.SamplerRngType,
             prediction = managed.Prediction,
             lora_apply_mode = managed.LoraApplyMode,
             offload_params_to_cpu = (sbyte)(managed.OffloadParamsToCPU ? 1 : 0),
@@ -64,11 +66,13 @@ internal static unsafe class DiffusionModelParameterMarshaller
             LoraModelDirectory = AnsiStringMarshaller.ConvertToManaged(unmanaged.lora_model_dir) ?? string.Empty,
             EmbeddingsDirectory = AnsiStringMarshaller.ConvertToManaged(unmanaged.embedding_dir) ?? string.Empty,
             StackedIdEmbeddingsDirectory = AnsiStringMarshaller.ConvertToManaged(unmanaged.photo_maker_path) ?? string.Empty,
+            TensorTypeRules = AnsiStringMarshaller.ConvertToManaged(unmanaged.tensor_type_rules) ?? string.Empty,
             VaeDecodeOnly = unmanaged.vae_decode_only == 1,
             FreeParamsImmediately = unmanaged.free_params_immediately == 1,
             ThreadCount = unmanaged.n_threads,
             Quantization = unmanaged.wtype,
             RngType = unmanaged.rng_type,
+            SamplerRngType = unmanaged.sampler_rng_type,
             Prediction = unmanaged.prediction,
             LoraApplyMode = unmanaged.lora_apply_mode,
             OffloadParamsToCPU = unmanaged.offload_params_to_cpu == 1,
@@ -101,5 +105,6 @@ internal static unsafe class DiffusionModelParameterMarshaller
         AnsiStringMarshaller.Free(unmanaged.lora_model_dir);
         AnsiStringMarshaller.Free(unmanaged.embedding_dir);
         AnsiStringMarshaller.Free(unmanaged.photo_maker_path);
+        AnsiStringMarshaller.Free(unmanaged.tensor_type_rules);
     }
 }

StableDiffusion.NET/Native/Marshaller/ImageGenerationParameterMarshaller.cs

@@ -46,6 +46,13 @@ internal static class ImageGenerationParameterMarshaller
                 TargetOverlap = unmanaged.vae_tiling_params.target_overlap,
                 RelSizeX = unmanaged.vae_tiling_params.rel_size_x,
                 RelSizeY = unmanaged.vae_tiling_params.rel_size_y
+            },
+            EasyCache =
+            {
+                IsEnabled = unmanaged.easycache.enabled == 1,
+                ReuseThreshold = unmanaged.easycache.reuse_threshold,
+                StartPercent = unmanaged.easycache.start_percent,
+                EndPercent = unmanaged.easycache.end_percent
             }
         };
 
@@ -126,6 +133,14 @@ internal static class ImageGenerationParameterMarshaller
                 rel_size_y = managed.VaeTiling.RelSizeY
             };
 
+            Native.Types.sd_easycache_params_t easyCache = new()
+            {
+                enabled = (sbyte)(managed.EasyCache.IsEnabled ? 1 : 0),
+                reuse_threshold = managed.EasyCache.ReuseThreshold,
+                start_percent = managed.EasyCache.StartPercent,
+                end_percent = managed.EasyCache.EndPercent,
+            };
+
             _imgGenParams = new Native.Types.sd_img_gen_params_t
             {
                 prompt = AnsiStringMarshaller.ConvertToUnmanaged(managed.Prompt),
@@ -145,7 +160,8 @@ internal static class ImageGenerationParameterMarshaller
                 control_image = _controlNetImage,
                 control_strength = managed.ControlNet.Strength,
                 pm_params = photoMakerParams,
-                vae_tiling_params = tilingParams
+                vae_tiling_params = tilingParams,
+                easycache = easyCache
             };
         }
 

StableDiffusion.NET/Native/Marshaller/VideoGenerationParameterMarshaller.cs

@@ -27,7 +27,14 @@ internal static class VideoGenerationParameterMarshaller
             Strength = unmanaged.strength,
             Seed = unmanaged.seed,
             FrameCount = unmanaged.video_frames,
-            VaceStrength = unmanaged.vace_strength
+            VaceStrength = unmanaged.vace_strength,
+            EasyCache =
+            {
+                IsEnabled = unmanaged.easycache.enabled == 1,
+                ReuseThreshold = unmanaged.easycache.reuse_threshold,
+                StartPercent = unmanaged.easycache.start_percent,
+                EndPercent = unmanaged.easycache.end_percent
+            }
         };
 
         return parameter;
@@ -68,6 +75,14 @@ internal static class VideoGenerationParameterMarshaller
             _initImage = managed.InitImage?.ToSdImage() ?? new Native.Types.sd_image_t();
             _endImage = managed.EndImage?.ToSdImage() ?? new Native.Types.sd_image_t();
             _controlFrames = managed.ControlFrames == null ? null : managed.ControlFrames.ToSdImage();
+            
+            Native.Types.sd_easycache_params_t easyCache = new()
+            {
+                enabled = (sbyte)(managed.EasyCache.IsEnabled ? 1 : 0),
+                reuse_threshold = managed.EasyCache.ReuseThreshold,
+                start_percent = managed.EasyCache.StartPercent,
+                end_percent = managed.EasyCache.EndPercent,
+            };
 
             _vidGenParams = new Native.Types.sd_vid_gen_params_t
             {
@@ -87,6 +102,7 @@ internal static class VideoGenerationParameterMarshaller
                 seed = managed.Seed,
                 video_frames = managed.FrameCount,
                 vace_strength = managed.VaceStrength,
+                easycache = easyCache,
             };
         }
 

StableDiffusion.NET/Native/Native.cs

@@ -25,6 +25,7 @@ using sd_type_t = Quantization;
 using sd_vid_gen_params_t = VideoGenerationParameter;
 using lora_apply_mode_t = LoraApplyMode;
 using preview_t = Preview;
+using sd_easycache_params_t = Native.Types.sd_easycache_params_t;
 using size_t = nuint;
 using uint32_t = uint;
 using uint8_t = byte;
@@ -69,11 +70,13 @@ internal unsafe partial class Native
             public byte* lora_model_dir;
             public byte* embedding_dir;
             public byte* photo_maker_path;
+            public byte* tensor_type_rules;
             public sbyte vae_decode_only;
             public sbyte free_params_immediately;
             public int n_threads;
             public sd_type_t wtype;
             public rng_type_t rng_type;
+            public rng_type_t sampler_rng_type;
             public prediction_t prediction;
             public lora_apply_mode_t lora_apply_mode;
             public sbyte offload_params_to_cpu;
@@ -131,6 +134,7 @@ internal unsafe partial class Native
             public int shifted_timestep;
         }
 
+        [StructLayout(LayoutKind.Sequential)]
         internal struct sd_pm_params_t
         {
             public sd_image_t* id_images;
@@ -139,6 +143,15 @@ internal unsafe partial class Native
             public float style_strength;
         }
 
+        [StructLayout(LayoutKind.Sequential)]
+        internal struct sd_easycache_params_t
+        {
+            public sbyte enabled;
+            public float reuse_threshold;
+            public float start_percent;
+            public float end_percent;
+        }
+
         [StructLayout(LayoutKind.Sequential)]
         internal struct sd_img_gen_params_t
         {
@@ -161,6 +174,7 @@ internal unsafe partial class Native
             public float control_strength;
             public sd_pm_params_t pm_params;
             public sd_tiling_params_t vae_tiling_params;
+            public sd_easycache_params_t easycache;
         }
 
         [StructLayout(LayoutKind.Sequential)]
@@ -182,6 +196,7 @@ internal unsafe partial class Native
             public int64_t seed;
             public int video_frames;
             public float vace_strength;
+            public sd_easycache_params_t easycache;
         }
 
         internal struct sd_ctx_t;
@@ -237,12 +252,12 @@ internal unsafe partial class Native
     [LibraryImport(LIB_NAME, EntryPoint = "str_to_sample_method")]
     internal static partial sample_method_t str_to_sample_method([MarshalAs(UnmanagedType.LPStr)] string str);
 
-    [LibraryImport(LIB_NAME, EntryPoint = "sd_schedule_name")]
+    [LibraryImport(LIB_NAME, EntryPoint = "sd_scheduler_name")]
     [return: MarshalAs(UnmanagedType.LPStr)]
-    internal static partial string sd_schedule_name(scheduler_t schedule);
+    internal static partial string sd_scheduler_name(scheduler_t schedule);
 
-    [LibraryImport(LIB_NAME, EntryPoint = "str_to_schedule")]
+    [LibraryImport(LIB_NAME, EntryPoint = "str_to_scheduler")]
-    internal static partial scheduler_t str_to_schedule([MarshalAs(UnmanagedType.LPStr)] string str);
+    internal static partial scheduler_t str_to_scheduler([MarshalAs(UnmanagedType.LPStr)] string str);
 
     [LibraryImport(LIB_NAME, EntryPoint = "sd_prediction_name")]
     [return: MarshalAs(UnmanagedType.LPStr)]
@@ -265,6 +280,9 @@ internal unsafe partial class Native
     [LibraryImport(LIB_NAME, EntryPoint = "str_to_lora_apply_mode")]
     internal static partial lora_apply_mode_t str_to_lora_apply_mode([MarshalAs(UnmanagedType.LPStr)] string str);
 
+    [LibraryImport(LIB_NAME, EntryPoint = "sd_easycache_params_init")]
+    internal static partial void sd_easycache_params_init(ref sd_easycache_params_t easycache_params);
+
     //
 
     [LibraryImport(LIB_NAME, EntryPoint = "sd_ctx_params_init")]
@@ -282,9 +300,6 @@ internal unsafe partial class Native
     [LibraryImport(LIB_NAME, EntryPoint = "free_sd_ctx")]
     internal static partial void free_sd_ctx(sd_ctx_t* sd_ctx);
 
-    [LibraryImport(LIB_NAME, EntryPoint = "sd_get_default_sample_method")]
-    internal static partial sample_method_t sd_get_default_sample_method(sd_ctx_t* sd_ctx);
-
     //
 
     [LibraryImport(LIB_NAME, EntryPoint = "sd_sample_params_init")]
@@ -301,6 +316,12 @@ internal unsafe partial class Native
     [return: MarshalAs(UnmanagedType.LPStr)]
     internal static partial string sd_img_gen_params_to_str([MarshalUsing(typeof(ImageGenerationParameterMarshaller))] in sd_img_gen_params_t sd_img_gen_params);
 
+    [LibraryImport(LIB_NAME, EntryPoint = "sd_get_default_sample_method")]
+    internal static partial sample_method_t sd_get_default_sample_method(sd_ctx_t* sd_ctx);
+
+    [LibraryImport(LIB_NAME, EntryPoint = "sd_get_default_scheduler")]
+    internal static partial scheduler_t sd_get_default_scheduler(sd_ctx_t* sd_ctx);
+
     [LibraryImport(LIB_NAME, EntryPoint = "generate_image")]
     internal static partial sd_image_t* generate_image(sd_ctx_t* sd_ctx, [MarshalUsing(typeof(ImageGenerationParameterMarshaller))] in sd_img_gen_params_t sd_img_gen_params);