fix:svdLoader error when the positive is empty

README.en.md

@@ -39,6 +39,10 @@ Usage：<br>
 
 ## Changelog
 
+**2024-02-29**
+
+- Fixed `easy svdLoader` error when the positive or negative is empty
+
 **2024-02-28**
 
 - Fixed the issue that 'easy preSampling' and other similar node, latent could not be generated based on the batch index after passing in

README.md

@@ -43,6 +43,10 @@ stage_c 与 stage_b 可以使用[checkpoints](https://huggingface.co/stabilityai
 
 ## 更新日志
 
+**2024-02-29**
+
+- 修复 `easy svdLoader` 报错
+
 **2024-02-28**
 
 - 修复`easy preSampling`等 latent传入后无法根据批次索引生成的问题

py/easyNodes.py

@@ -1254,13 +1254,13 @@ class svdLoader:
         negative = [[torch.zeros_like(pooled),
                      {"motion_bucket_id": motion_bucket_id, "fps": fps, "augmentation_level": augmentation_level,
                       "concat_latent_image": torch.zeros_like(t)}]]
-        if optional_positive is not None:
+        if optional_positive is not None and optional_positive != '':
             if clip_name == 'None':
                 raise Exception("You need choose a open_clip model when positive is not empty")
             clip = easyCache.load_clip(clip_name)
             positive_embeddings_final, = CLIPTextEncode().encode(clip, optional_positive)
             positive, = ConditioningConcat().concat(positive, positive_embeddings_final)
-        if optional_negative is not None:
+        if optional_negative is not None and optional_negative != '':
             if clip_name == 'None':
                 raise Exception("You need choose a open_clip model when negative is not empty")
             negative_embeddings_final, = CLIPTextEncode().encode(clip, optional_negative)

py/libs/loader.py

@@ -174,7 +174,7 @@ class easyLoader:
             cache_name = ckpt_name + "_" + config_name
         if cache_name in self.loaded_objects["ckpt"]:
             clip_vision = self.loaded_objects["clip_vision"][cache_name][0] if load_vision else None
-            clip = self.loaded_objects["clip"][cache_name][0]
+            clip = self.loaded_objects["clip"][cache_name][0] if not load_vision else None
             return self.loaded_objects["ckpt"][cache_name][0], clip, self.loaded_objects["bvae"][cache_name][0], clip_vision
 
         ckpt_path = folder_paths.get_full_path("checkpoints", ckpt_name)