v0.32.0

修正text image节点的padding问题

CNAME

@@ -0,0 +1 @@
+mixlabnodes.com

README.md

@@ -6,13 +6,14 @@
 
 ##### `最新`：
 
+- 增加 SiliconflowLLM，可以使用由Siliconflow提供的免费LLM
+
 - 增加 Edit Mask，方便在生成的时候手动绘制 mask [workflow](./workflow/edit-mask-workflow.json)
 
+<!-- - ChatGPT 节点支持 Local LLM（llama.cpp），Phi3、llama3 都可以直接一个节点运行了。模型下载后，放置到 `models/llamafile/` -->
 
-- ChatGPT 节点支持 Local LLM（llama.cpp），Phi3、llama3 都可以直接一个节点运行了。模型下载后，放置到 `models/llamafile/`
-
-- 右键菜单支持 text-to-text，方便对 prompt 词补全
-
+<!-- - 右键菜单支持 text-to-text，方便对 prompt 词补全 -->
+<!-- 
 强烈推荐：
 [Phi-3-mini-4k-instruct-function-calling-GGUF](https://huggingface.co/nold/Phi-3-mini-4k-instruct-function-calling-GGUF)
 
@@ -21,11 +22,13 @@
 - 右键菜单支持 image-to-text，使用多模态模型，多模态使用 [llava-phi-3-mini-gguf](https://huggingface.co/xtuner/llava-phi-3-mini-gguf/tree/main)，注意需要把llava-phi-3-mini-mmproj-f16.gguf也下载
 
 ![](./assets/prompt_ai_setup.png)
-![](./assets/prompt-ai.png)
+![](./assets/prompt-ai.png) -->
 
 
 #### `相关插件推荐`
 
+[comfyui-liveportrait](https://github.com/shadowcz007/comfyui-liveportrait)
+
 [Comfyui-ChatTTS](https://github.com/shadowcz007/Comfyui-ChatTTS)
 
 [comfyui-sound-lab](https://github.com/shadowcz007/comfyui-sound-lab)
@@ -167,6 +170,9 @@ pip install llama-cpp-python \
 
 > A new layer class node has been added, allowing you to separate the image into layers. After merging the images, you can input the controlnet for further processing.
 
+> The composite images node overlays a foreground image onto a background image at specified positions and scales, with optional blending modes and masking capabilities. position : 'overall',"center_center","left_bottom","center_bottom","right_bottom","left_top","center_top","right_top"
+
+
 ![layers](./assets/layers-workflow.svg)
 
 ![poster](./assets/poster-workflow.svg)
@@ -198,6 +204,12 @@ pip install llama-cpp-python \
 
 > Conveniently load images from a fixed address on the internet to ensure that default images in the workflow can be executed.
 
+#### TextImage
+
+> [下载字体](https://drxie.github.io/OSFCC/)放到 ```custom_nodes/comfyui-mixlab-nodes/assets/fonts```
+
+
+
 ### Style
 
 > Apply VisualStyle Prompting , Modified from [ComfyUI_VisualStylePrompting](https://github.com/ExponentialML/ComfyUI_VisualStylePrompting)

__init__.py

@@ -3,12 +3,14 @@ import os
 import subprocess
 import importlib.util
 import sys,json
-import urllib
+import execution
+import uuid
 import hashlib
 import datetime
 import folder_paths
 import logging
 import base64,io,re
+import random
 from PIL import Image
 from comfy.cli_args import args
 python = sys.executable
@@ -171,15 +173,14 @@ def create_for_https():
         os.mkdir(https_key_path)
     if not os.path.exists(crt):
         create_key(key,crt)
-
-    print('https_key OK: ', crt,key)
+    # print('https_key OK: ', crt,key)
     return (crt,key)
 
 
 
 # workflow  目录下的所有json
 def read_workflow_json_files_all(folder_path):
-    print('#read_workflow_json_files_all',folder_path)
+    # print('#read_workflow_json_files_all',folder_path)
     json_files = []
     for root, dirs, files in os.walk(folder_path):
         for file in files:
@@ -309,9 +310,10 @@ def get_my_workflow_for_app(filename="my_workflow_app.json",category="",is_all=F
         print('app_workflow_path: ',app_workflow_path)
         try:
             with open(app_workflow_path) as json_file:
+                json_data=json.load(json_file)
                 apps = [{
                     'filename':filename,
-                    'data':json.load(json_file)
+                    'data':json_data
                 }]
         except Exception as e:
             print("发生异常：", str(e))
@@ -521,9 +523,17 @@ async def new_start(self, address, port, verbose=True, call_on_start=None):
             logging.info("\n")
             logging.info("\n\nStarting server")
 
+            import socket
+
+            hostname = socket.gethostname()
+            ip_address = socket.gethostbyname(hostname)
+
+            # print(f"本机的IP地址是: {ip_address}")
+
+
             # print("\033[93mStarting server\n")
-            logging.info("\033[93mTo see the GUI go to: http://{}:{}".format(address, http_port))
-            logging.info("\033[93mTo see the GUI go to: https://{}:{}\033[0m".format(address, https_port))
+            logging.info("\033[93mTo see the GUI go to: http://{}:{} or http://{}:{}".format(ip_address, http_port,address,http_port))
+            logging.info("\033[93mTo see the GUI go to: https://{}:{} or https://{}:{}\033[0m".format(ip_address, https_port,address,https_port))
  
             # print("\033[93mTo see the GUI go to: http://{}:{}".format(address, http_port))
             # print("\033[93mTo see the GUI go to: https://{}:{}\033[0m".format(address, https_port))
@@ -608,13 +618,34 @@ async def mixlab_workflow_hander(request):
                     category=data['category']
                 if 'admin' in data:
                     admin=data['admin']
+
+                ds=get_my_workflow_for_app(filename,category,admin)
+                data=[]           
+                for json_data in ds:
+                    # 不传给前端
+                    if 'output' in json_data['data']:
+                        del json_data['data']['output']
+                    if 'workflow' in json_data['data']:
+                        del json_data['data']['workflow']
+                    data.append(json_data)
+
                 result={
-                    'data':get_my_workflow_for_app(filename,category,admin),
+                    'data':data,
                     'status':'success',
                 }
             elif data['task']=='list':
+                ds=get_workflows()
+                data=[]
+                for json_data in ds:
+                    # 不传给前端
+                    if 'output' in json_data['data']:
+                        del json_data['data']['output']
+                    if 'workflow' in json_data['data']:
+                        del json_data['data']['workflow']
+                    data.append(json_data)
+
                 result={
-                    'data':get_workflows(),
+                    'data':data,
                     'status':'success',
                 }
     except Exception as e:
@@ -699,23 +730,134 @@ async def rembg_hander(request):
 
     return web.json_response(result)
 
-
-@routes.post("/mixlab/prompt_result")
-async def post_prompt_result(request):
-    data = await request.json()
-    res=None
-    # print(data)
-    try:
-        action=data['action']
-        if action=='save':
-            result=data['data']
-            res=save_prompt_result(result['prompt_id'],result)
-        elif action=='all':
-            res=get_prompt_result()
-    except Exception as e:
-        print('/mixlab/prompt_result',False,e)
+# 保存运行结果？暂时去掉
+# @routes.post("/mixlab/prompt_result")
+# async def post_prompt_result(request):
+#     data = await request.json()
+#     res=None
+#     # print(data)
+#     try:
+#         action=data['action']
+#         if action=='save':
+#             result=data['data']
+#             res=save_prompt_result(result['prompt_id'],result)
+#         elif action=='all':
+#             res=get_prompt_result()
+#     except Exception as e:
+#         print('/mixlab/prompt_result',False,e)
     
-    return web.json_response({"result":res})
+#     return web.json_response({"result":res})
+
+# 种子设置
+def random_seed(seed, data):
+    max_seed = 4294967295
+
+    for id, value in data.items():
+        if 'seed' in value['inputs'] and not isinstance(value['inputs']['seed'], list) and seed[id] in ['increment', 'decrement', 'randomize']:
+            value['inputs']['seed'] = round(random.random() * max_seed)
+        
+        if 'noise_seed' in value['inputs'] and not isinstance(value['inputs']['noise_seed'], list) and seed[id] in ['increment', 'decrement', 'randomize']:
+            value['inputs']['noise_seed'] = round(random.random() * max_seed)
+        
+        if value.get('class_type') == "Seed_" and seed[id] in ['increment', 'decrement', 'randomize']:
+            value['inputs']['seed'] = round(random.random() * max_seed)
+        
+        print('new Seed', value)
+    
+    return data
+
+
+# 运行工作流，代替官方的prompt接口
+@routes.post("/mixlab/prompt")
+async def mixlab_post_prompt(request):
+    p_intance=PromptServer.instance
+    logging.info("got prompt")
+    resp_code = 200
+    out_string = ""
+    json_data =  await request.json()
+    # json_data = p_intance.trigger_on_prompt(json_data)
+    # filename,category, client_id ,input
+    # workflow 的 filename,category
+
+    # 输入的参数
+    input_data=json_data['input'] if "input" in json_data else []
+    # 种子
+    seed=json_data['seed'] if "seed" in json_data else {}
+
+    apps=get_my_workflow_for_app(json_data['filename'],json_data['category'],False)
+    
+    prompt=json_data['prompt'] if 'prompt' in json_data else None
+
+    if len(apps)==1:
+        # 取到prompt
+        prompt=apps[0]['data']['output']
+        # 更新input_data到prompt里
+        '''
+          {
+                "inputs": {
+                    "number": 512,
+                    "min_value": 512,
+                    "max_value": 2048,
+                    "step": 1
+                },
+                "class_type": "IntNumber",
+                "id": "22"
+            },
+        '''
+
+        for inp in input_data:
+            id=inp['id']
+            if prompt[id]['class_type']==inp['class_type']:
+                prompt[id]['inputs'].update(inp['inputs']) 
+
+
+    if prompt==None:
+        return web.json_response({"error": "no prompt", "node_errors": []}, status=400)
+    else:
+        # 种子更新 
+        '''
+            "seed": {
+                    "45": "randomize",
+                    "46": "randomize"
+                }
+            '''
+        json_data["prompt"]=random_seed(seed,prompt)
+
+    # print("#json_data",prompt)
+    # 需要把apps处理成 prompt
+    # 注意seed的处理
+    
+    if "number" in json_data:
+        number = float(json_data['number'])
+    else:
+        number = p_intance.number
+        if "front" in json_data:
+            if json_data['front']:
+                number = -number
+
+        p_intance.number += 1
+
+    if "prompt" in json_data:
+        prompt = json_data["prompt"]
+        valid = execution.validate_prompt(prompt)
+        extra_data = {}
+        if "extra_data" in json_data:
+            extra_data = json_data["extra_data"]
+
+        if "client_id" in json_data:
+            extra_data["client_id"] = json_data["client_id"]
+        if valid[0]:
+            prompt_id = str(uuid.uuid4())
+            outputs_to_execute = valid[2]
+            p_intance.prompt_queue.put((number, prompt_id, prompt, extra_data, outputs_to_execute))
+            response = {"prompt_id": prompt_id, "number": number, "node_errors": valid[3]}
+            return web.json_response(response)
+        else:
+            logging.warning("invalid prompt: {}".format(valid[1]))
+            return web.json_response({"error": valid[1], "node_errors": valid[3]}, status=400)
+    else:
+        return web.json_response({"error": "no prompt", "node_errors": []}, status=400)
+
 
 
 async def start_local_llm(data):
@@ -760,8 +902,7 @@ async def start_local_llm(data):
             if file.startswith(file_name):
                 clip_model_path=os.path.join(os.path.dirname(model),file)
                 chat_format='llava-1-5'
-    print('#clip_model_path',chat_format,clip_model_path)
-
+    # print('#clip_model_path',chat_format,clip_model_path,model)
 
     address="127.0.0.1"
     port=9090
@@ -779,38 +920,39 @@ async def start_local_llm(data):
     server_settings=ServerSettings(host=address,port=port)
 
     name, ext = os.path.splitext(os.path.basename(model))
-    print('#model',name)
-    app = create_app(
-                server_settings=server_settings,
-                model_settings=[
-                    ModelSettings(
-                    model=model,
-                    model_alias=name,
-                    n_gpu_layers=n_gpu_layers,
-                    n_ctx=4098,
-                    chat_format=chat_format,
-                    embedding=False,
-                    clip_model_path=clip_model_path
-                    )])
+    if name:
+        # print('#model',name)
+        app = create_app(
+                    server_settings=server_settings,
+                    model_settings=[
+                        ModelSettings(
+                        model=model,
+                        model_alias=name,
+                        n_gpu_layers=n_gpu_layers,
+                        n_ctx=4098,
+                        chat_format=chat_format,
+                        embedding=False,
+                        clip_model_path=clip_model_path
+                        )])
 
-    def run_uvicorn():
-        uvicorn.run(
-                app,
-                host=os.getenv("HOST", server_settings.host),
-                port=int(os.getenv("PORT", server_settings.port)),
-                ssl_keyfile=server_settings.ssl_keyfile,
-                ssl_certfile=server_settings.ssl_certfile,
-            )
+        def run_uvicorn():
+            uvicorn.run(
+                    app,
+                    host=os.getenv("HOST", server_settings.host),
+                    port=int(os.getenv("PORT", server_settings.port)),
+                    ssl_keyfile=server_settings.ssl_keyfile,
+                    ssl_certfile=server_settings.ssl_certfile,
+                )
 
-    # 创建一个子线程
-    thread = threading.Thread(target=run_uvicorn)
+        # 创建一个子线程
+        thread = threading.Thread(target=run_uvicorn)
 
-    # 启动子线程
-    thread.start()
+        # 启动子线程
+        thread.start()
 
-    llama_port=port
-    llama_model=data['model']
-    llama_chat_format=chat_format
+        llama_port=port
+        llama_model=data['model']
+        llama_chat_format=chat_format
 
     return  {"port":llama_port,"model":llama_model,"chat_format":llama_chat_format}
 
@@ -939,6 +1081,10 @@ NODE_CLASS_MAPPINGS = {
 # 一个包含节点友好/可读的标题的字典
 NODE_DISPLAY_NAME_MAPPINGS = {
     "AppInfo":"App Info ♾️MixlabApp",
+    "ScreenShare":"Screen Share ♾️Mixlab",
+    "FloatingVideo":"Floating Video ♾️Mixlab",
+    "TextImage":"Text Image ♾️Mixlab",
+    
     "Color":"Color Input ♾️MixlabApp",
     "TextInput_":"Text Input ♾️MixlabApp",
     "FloatSlider":"Float Slider Input ♾️MixlabApp",
@@ -952,8 +1098,7 @@ NODE_DISPLAY_NAME_MAPPINGS = {
     "SplitLongMask":"Splitting a long image into sections",
     "VAELoaderConsistencyDecoder":"Consistency Decoder Loader",
     "VAEDecodeConsistencyDecoder":"Consistency Decoder Decode",
-    "ScreenShare":"Screen Share ♾️Mixlab",
-    "FloatingVideo":"FloatingVideo ♾️Mixlab",
+    
     
     "MergeLayers":"Merge Layers ♾️Mixlab",
     "SpeechSynthesis":"SpeechSynthesis ♾️Mixlab",
@@ -988,7 +1133,9 @@ NODE_DISPLAY_NAME_MAPPINGS = {
     "IncrementingListNode_":"Create Incrementing Number List ♾️Mixlab",
     "LoadImagesToBatch":"Load Images(base64) ♾️Mixlab",
     "PreviewMask_":"Preview Mask",
-    "AudioPlay":"Audio Play ♾️Mixlab"
+    "AudioPlay":"Preview Audio ♾️Mixlab",
+
+     "MultiplicationNode":"Math Operation ♾️Mixlab",
 }
 
 # web ui的节点功能
@@ -1000,11 +1147,12 @@ logging.info('\033[91m ### Mixlab Nodes: \033[93mLoaded')
 # print('\033[91m ### Mixlab Nodes: \033[93mLoaded')
 
 try:
-    from .nodes.ChatGPT import ChatGPTNode,ShowTextForGPT,CharacterInText,TextSplitByDelimiter
+    from .nodes.ChatGPT import ChatGPTNode,ShowTextForGPT,CharacterInText,TextSplitByDelimiter,SiliconflowFreeNode
     logging.info('ChatGPT.available True')
 
     NODE_CLASS_MAPPINGS_V = {
        "ChatGPTOpenAI":ChatGPTNode,
+       "SiliconflowLLM":SiliconflowFreeNode,
         "ShowTextForGPT":ShowTextForGPT,
         "CharacterInText":CharacterInText,
         "TextSplitByDelimiter":TextSplitByDelimiter,
@@ -1013,6 +1161,7 @@ try:
     # 一个包含节点友好/可读的标题的字典
     NODE_DISPLAY_NAME_MAPPINGS_V = {
         "ChatGPTOpenAI":"ChatGPT & Local LLM ♾️Mixlab",
+        "SiliconflowLLM":"LLM Siliconflow ♾️Mixlab",
         "ShowTextForGPT":"Show Text ♾️MixlabApp",
         "CharacterInText":"Character In Text",
         "TextSplitByDelimiter":"Text Split By Delimiter",
@@ -1092,7 +1241,7 @@ try:
         "VideoCombine_Adv":"Video Combine",
         "LoadAndCombinedAudio_":"Load And Combined Audio",
         "CombineAudioVideo":"Combine Audio Video",
-        "ScenesNode_":"Scenes Node", 
+        "ScenesNode_":"Select Scene", 
         "GenerateFramesByCount":"Generate Frames By Count"
     }
 

nodes/Audio.py

@@ -90,7 +90,7 @@ class AudioPlayNode:
             # {'waveform': tensor([], size=(1, 1, 0)), 'sample_rate': 44100}
             is_tensor=True
 
-        if is_tensor:
+        if is_tensor and (not 'audio_path' in audio):
             filename_prefix=""
             # 保存
             filename_prefix += self.prefix_append
@@ -108,7 +108,12 @@ class AudioPlayNode:
                 })
             
         else:
-            results=[audio]
+            results=[{
+                    "filename": audio['filename'],
+                    "subfolder":audio['subfolder'],
+                    "type": audio['type'],
+                    "audio_path":audio['audio_path']
+                }]
                 
 
         # print(audio)

nodes/ChatGPT.py

@@ -53,8 +53,8 @@ def azure_client(key,url):
 
 def openai_client(key,url):
     client = openai.OpenAI(
-    api_key=key,
-    base_url=url
+        api_key=key,
+        base_url=url
     )
     return client
 
@@ -163,7 +163,7 @@ def llama_cpp_client(file_name):
 
 
 def chat(client, model_name,messages ):
-
+        print('#chat',model_name,messages)
         try_count = 0
         while True:
             try_count += 1
@@ -215,15 +215,32 @@ class ChatGPTNode:
 
     @classmethod
     def INPUT_TYPES(cls):
-        model_list=llama_modes_list+[
+        model_list=llama_modes_list+[ 
                     "gpt-3.5-turbo",
-                    "gpt-3.5-turbo-0125",
-                    "gpt-35-turbo",
-                    "gpt-3.5-turbo-16k", 
-                    "gpt-3.5-turbo-16k-0613", 
-                    "gpt-4-0613",
-                    "gpt-4-1106-preview",
-                    "glm-4"
+            "gpt-3.5-turbo-16k",
+            "gpt-4o",
+            "gpt-4o-2024-05-13",
+            "gpt-4",
+            "gpt-4-0314",
+            "gpt-4-0613",
+            "gpt-3.5-turbo-0301",
+            "gpt-3.5-turbo-0613",
+            "gpt-3.5-turbo-16k-0613",
+            "qwen-turbo",
+            "qwen-plus",
+            "qwen-long",
+            "qwen-max",
+            "qwen-max-longcontext",
+            "glm-4",
+            "glm-3-turbo",
+            "moonshot-v1-8k",
+            "moonshot-v1-32k",
+            "moonshot-v1-128k",
+            "deepseek-chat",
+            "Qwen/Qwen2-7B-Instruct",
+            "THUDM/glm-4-9b-chat",
+            "01-ai/Yi-1.5-9B-Chat-16K",
+            "meta-llama/Meta-Llama-3.1-8B-Instruct"
                     ]
         return {
             "required": {
@@ -287,7 +304,7 @@ class ChatGPTNode:
                 client=llama_cpp_client(model)
             else :
                 client = openai_client(api_key,api_url)  # 使用 ChatGPT  的接口
-                print('using ChatGPT interface')
+                # print('using ChatGPT interface',api_key,api_url)
 
         # 把用户的提示添加到会话历史中
         # 调用API时传递整个会话历史
@@ -303,6 +320,7 @@ class ChatGPTNode:
         session_history=crop_list_tail(self.session_history,context_size)
 
         messages=[{"role": "system", "content": self.system_content}]+session_history+[{"role": "user", "content": prompt}]
+
         response_content = chat(client,model,messages)
         
         self.session_history=self.session_history+[{"role": "user", "content": prompt}]+[{'role':'assistant',"content":response_content}]
@@ -323,6 +341,91 @@ class ChatGPTNode:
         return (response_content,json.dumps(messages, indent=4),json.dumps(self.session_history, indent=4),)
 
 
+class SiliconflowFreeNode:
+    def __init__(self):
+        # self.__client = OpenAI()
+        self.session_history = []  # 用于存储会话历史的列表
+        # self.seed=0
+        self.system_content="You are ChatGPT, a large language model trained by OpenAI. Answer as concisely as possible."
+
+    @classmethod
+    def INPUT_TYPES(cls):
+        model_list= [ 
+            "Qwen/Qwen2-7B-Instruct",
+            "THUDM/glm-4-9b-chat",
+            "01-ai/Yi-1.5-9B-Chat-16K",
+            "meta-llama/Meta-Llama-3.1-8B-Instruct"
+            ]
+        return {
+            "required": {
+                "api_key":("KEY", {"default": "", "multiline": True,"dynamicPrompts": False}),
+                "prompt": ("STRING", {"multiline": True,"dynamicPrompts": False}),
+                "system_content": ("STRING", 
+                                   {
+                                       "default": "You are ChatGPT, a large language model trained by OpenAI. Answer as concisely as possible.", 
+                                       "multiline": True,"dynamicPrompts": False
+                                       }),
+                "model": ( model_list, 
+                    {"default": model_list[0]}),
+                "seed": ("INT", {"default": 0, "min": 0, "max": 0xffffffffffffffff, "step": 1}),
+                "context_size":("INT", {"default": 1, "min": 0, "max":30, "step": 1}),
+            },
+             "hidden": {
+                "unique_id": "UNIQUE_ID",
+                "extra_pnginfo": "EXTRA_PNGINFO",
+            },
+        }
+
+    RETURN_TYPES = ("STRING","STRING","STRING",)
+    RETURN_NAMES = ("text","messages","session_history",)
+    FUNCTION = "generate_contextual_text"
+    CATEGORY = "♾️Mixlab/GPT"
+    INPUT_IS_LIST = False
+    OUTPUT_IS_LIST = (False,False,False,)
+
+    
+    def generate_contextual_text(self,
+                                 api_key,
+                                 prompt, 
+                                 system_content,
+                                model, 
+                                seed,context_size,unique_id = None, extra_pnginfo=None):
+    
+        api_url="https://api.siliconflow.cn/v1"
+        
+        # 把系统信息和初始信息添加到会话历史中
+        if system_content:
+            self.system_content=system_content
+            # self.session_history=[]
+            # self.session_history.append({"role": "system", "content": system_content})
+        
+        # 
+        client = openai_client(api_key,api_url)  # 使用 ChatGPT  的接口
+        # print('using ChatGPT interface',api_key,api_url)
+
+        # 把用户的提示添加到会话历史中
+        # 调用API时传递整个会话历史
+
+        def crop_list_tail(lst, size):
+            if size >= len(lst):
+                return lst
+            elif size==0:
+                return []
+            else:
+                return lst[-size:]
+            
+        session_history=crop_list_tail(self.session_history,context_size)
+
+        messages=[{"role": "system", "content": self.system_content}]+session_history+[{"role": "user", "content": prompt}]
+
+        response_content = chat(client,model,messages)
+        
+        self.session_history=self.session_history+[{"role": "user", "content": prompt}]+[{'role':'assistant',"content":response_content}]
+
+        return (response_content,json.dumps(messages, indent=4),json.dumps(self.session_history, indent=4),)
+
+
+
 
 class ShowTextForGPT:
     @classmethod

nodes/ClipInterrogator.py

@@ -79,7 +79,7 @@ def get_clip_interrogator_path():
 
 cache_path=get_clip_interrogator_path()
 
-caption_model_path=os.path.join(cache_path, "Salesforce/blip-image-captioning-base")
+caption_model_path=os.path.join(cache_path, "Salesforce","blip-image-captioning-base")
 if not os.path.exists(caption_model_path):
     print(f"## clip_interrogator_model not found: {caption_model_path}, pls download from https://huggingface.co/Salesforce/blip-image-captioning-base")
     caption_model_path='Salesforce/blip-image-captioning-base'

nodes/ImageNode.py

@@ -15,8 +15,8 @@ import cv2
 import string
 import math,glob
 from .Watcher import FolderWatcher
-import hashlib
 
+from itertools import product
 
 
 # 将PIL图片转换为OpenCV格式
@@ -29,142 +29,105 @@ def opencv_to_pil(image):
     pil_image = Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
     return pil_image
 
+# 列出目录下面的所有文件
+def get_files_with_extension(directory, extensions):
+    file_list = []
+    # 确保extensions参数是一个list，即使只有一个元素
+    if not isinstance(extensions, (tuple, list)):
+        extensions = [extensions]
+    for root, dirs, files in os.walk(directory):
+        # print(f"Files at {root}: {files}")  # 确认files是一个字符串列表
+        for file in files:
+            # 检查文件是否以任何一个提供的扩展名结尾
+            if any(file.endswith(ext) for ext in extensions):
+                # 直接将文件名添加到列表中
+                file_list.append(file)
+    return file_list
+	
+def composite_images(foreground, background, mask, is_multiply_blend=False, position="overall", scale=0.25):
+    width, height = foreground.size
+    bg_image = background
+    bwidth, bheight = bg_image.size
 
-def composite_images(foreground, background, mask,is_multiply_blend=False,position="overall"):
-    width,height=foreground.size
- 
-    bg_image=background
+    scale=max(scale,1/bwidth)
+    scale=max(scale,1/bheight)
 
-    bwidth,bheight=bg_image.size
+    def determine_scale_option(width, height):
+        return 'height' if height > width else 'width'
 
-    # 按z-index排序
-    if position=="overall":
+    if position == "overall":
         layer = {
-                "x":0,
-                "y":0,
-                "width":bwidth,
-                "height":bheight,
-                "z_index":88,
-                "scale_option":'overall',
-                "image":foreground,
-                "mask":mask
+            "x": 0,
+            "y": 0,
+            "width": bwidth,
+            "height": bheight,
+            "z_index": 88,
+            "scale_option": 'overall',
+            "image": foreground,
+            "mask": mask
         }
+    else:
+        scale_option = determine_scale_option(width, height)
+        if scale_option == 'height':
+            scale = int(bheight * scale) / height
+        else:
+            scale = int(bwidth * scale) / width
 
-    elif position=='center_bottom':
-        
-        scale = int(bwidth*0.25) / width
+        new_width = int(width * scale)
         new_height = int(height * scale)
 
-        layer = {
-                "x":int(bwidth*0.75*0.5),
-                "y":bheight-new_height-24,
-                "width":int(bwidth*0.25),
-                "height":int(bheight*0.25),
-                "z_index":88,
-                "scale_option":'width',
-                "image":foreground,
-                "mask":mask
-        }
-
-    elif position=='right_bottom':
-        
-        scale = int(bwidth*0.25) / width
-        new_height = int(height * scale)
+        if position == 'center_bottom':
+            x_position = int((bwidth - new_width) * 0.5)
+            y_position = bheight - new_height - 24
+        elif position == 'right_bottom':
+            x_position = bwidth - new_width - 24
+            y_position = bheight - new_height - 24
+        elif position == 'center_top':
+            x_position = int((bwidth - new_width) * 0.5)
+            y_position = 24
+        elif position == 'right_top':
+            x_position = bwidth - new_width - 24
+            y_position = 24
+        elif position == 'left_top':
+            x_position = 24
+            y_position = 24
+        elif position == 'left_bottom':
+            x_position = 24
+            y_position = bheight - new_height - 24
+        elif position == 'center_center':
+            x_position = int((bwidth - new_width) * 0.5)
+            y_position = int((bheight - new_height) * 0.5)
 
         layer = {
-                "x":bwidth-int(bwidth*0.25)-24,
-                "y":bheight-new_height-24,
-                "width":int(bwidth*0.25),
-                "height":int(bheight*0.25),
-                "z_index":88,
-                "scale_option":'width',
-                "image":foreground,
-                "mask":mask
+            "x": x_position,
+            "y": y_position,
+            "width": new_width,
+            "height": new_height,
+            "z_index": 88,
+            "scale_option": scale_option,
+            "image": foreground,
+            "mask": mask
         }
 
+    layer_image = layer['image']
+    layer_mask = layer['mask']
 
-    elif position=='center_top':
-        
-        scale = int(bwidth*0.25) / width
-        new_height = int(height * scale)
+    bg_image = merge_images(bg_image,
+                            layer_image,
+                            layer_mask,
+                            layer['x'],
+                            layer['y'],
+                            layer['width'],
+                            layer['height'],
+                            layer['scale_option'],
+                            is_multiply_blend)
 
-        layer = {
-                "x":int( bwidth*0.75*0.5),
-                "y":24,
-                "width":int(bwidth*0.25),
-                "height":int(bheight*0.25),
-                "z_index":88,
-                "scale_option":'width',
-                "image":foreground,
-                "mask":mask
-        }
+    bg_image = bg_image.convert('RGB')
 
-    elif position=='right_top':
-        
-        scale = int(bwidth*0.25) / width
-        new_height = int(height * scale)
-
-        layer = {
-                "x":bwidth-int(bwidth*0.25)-24,
-                "y":24,
-                "width":int(bwidth*0.25),
-                "height":int(bheight*0.25),
-                "z_index":88,
-                "scale_option":'width',
-                "image":foreground,
-                "mask":mask
-        }
-    elif position=='left_top':
-        
-        scale = int(bwidth*0.25) / width
-        new_height = int(height * scale)
-
-        layer = {
-                "x":24,
-                "y":24,
-                "width":int(bwidth*0.25),
-                "height":int(bheight*0.25),
-                "z_index":88,
-                "scale_option":'width',
-                "image":foreground,
-                "mask":mask
-        }
-    elif position=='left_bottom':
-        
-        scale = int(bwidth*0.25) / width
-        new_height = int(height * scale)
-
-        layer = {
-                "x":24,
-                "y":bheight-new_height-24,
-                "width":int(bwidth*0.25),
-                "height":int(bheight*0.25),
-                "z_index":88,
-                "scale_option":'width',
-                "image":foreground,
-                "mask":mask
-        }
-                
-    # width, height = bg_image.size
-
-    layer_image=layer['image']
-    layer_mask=layer['mask']
-    
-    bg_image=merge_images(bg_image,
-                        layer_image,
-                        layer_mask,
-                        layer['x'],
-                        layer['y'],
-                        layer['width'],
-                        layer['height'],
-                        layer['scale_option'],
-                        is_multiply_blend )
-                   
-    bg_image=bg_image.convert('RGB')
-    
     return bg_image
 
 
+
 def count_files_in_directory(directory):
     file_count = 0
     for _, _, files in os.walk(directory):
@@ -201,7 +164,8 @@ class AnyType(str):
 any_type = AnyType("*")
 
 
-FONT_PATH= os.path.abspath(os.path.join(os.path.dirname(__file__),'../assets/王汉宗颜楷体繁.ttf'))
+FONT_PATH= os.path.abspath(os.path.join(os.path.dirname(__file__),"..","assets","fonts"))
+
 
 MAX_RESOLUTION=8192
 
@@ -803,85 +767,78 @@ def multiply_blend(image1, image2):
 
 # cv2.imwrite('result.jpg', result)
 
+# 使用gpt4o优化代码
+# 为了消除图像合并时出现的灰色描边，可以使用以下方法：
+# 调整透明度：确保透明像素不会引入不需要的颜色。
+# 预处理图像：在缩放图像之前，可以先将图像的边缘进行预处理，例如扩展边缘颜色，减少抗锯齿带来的过渡效果。
 
-def merge_images(bg_image, layer_image, mask, x, y, width, height, scale_option,is_multiply_blend=False):
+def merge_images(bg_image, layer_image, mask, x, y, width, height, scale_option, is_multiply_blend=False):
     # 打开底图
     bg_image = bg_image.convert("RGBA")
 
     # 打开图层
     layer_image = layer_image.convert("RGBA")
-    # layer_image = layer_image.resize((width, height))
-
+   
     # 根据缩放选项调整图像大小
     if scale_option == "height":
         # 按照高度比例缩放
         original_width, original_height = layer_image.size
         scale = height / original_height
         new_width = int(original_width * scale)
-        layer_image = layer_image.resize((new_width, height))
+        layer_image = layer_image.resize((new_width, height), Image.NEAREST)
     elif scale_option == "width":
         # 按照宽度比例缩放
         original_width, original_height = layer_image.size
         scale = width / original_width
         new_height = int(original_height * scale)
-        layer_image = layer_image.resize((width, new_height))
+        layer_image = layer_image.resize((width, new_height), Image.NEAREST)
     elif scale_option == "overall":
         # 整体缩放
-        layer_image = layer_image.resize((width, height))
-
+        layer_image = layer_image.resize((width, height), Image.NEAREST)
     elif scale_option == "longest":
         original_width, original_height = layer_image.size
         if original_width > original_height:
-            new_width=width
+            new_width = width
             scale = width / original_width
             new_height = int(original_height * scale)
-            x=0
-            y=int((height-new_height)*0.5)
+            x = 0
+            y = int((height - new_height) * 0.5)
         else:
-            new_height=height
+            new_height = height
             scale = height / original_height
             new_width = int(original_height * scale)
-            x=int((width-new_width)*0.5)
-            y=0
-    # elif side == "shortest":
-    #         if width < height:
-    #              
-    #         else:
-    #              
-    
+            x = int((width - new_width) * 0.5)
+            y = 0
 
     # 调整mask的大小
     nw, nh = layer_image.size
-    mask = mask.resize((nw, nh))
+    mask = mask.resize((nw, nh), Image.NEAREST)
 
-    # # 分离出a通道
-    # r, g, b, alpha = layer_image.split()
-    # alpha = ImageOps.invert(alpha)
-    # # 创建一个新的RGB图像
-    # new_rgb_image = Image.new("RGB", layer_image.size)
-    # # 将透明通道粘贴到新的RGB图像上
-    # new_rgb_image.paste(layer_image, (0, 0), mask=alpha)
-   
-    # new_rgb_image.paste(layer_image, (x, y), mask=mask)
-    # mask=new_rgb_image.convert('L')
-    # mask = ImageOps.invert(mask)
+    # 预处理图像边缘以减少灰色描边
+    layer_image = layer_image.filter(ImageFilter.SMOOTH)
 
     if is_multiply_blend:
-        bg_image_white=Image.new("RGB", bg_image.size,(255, 255, 255))
+        bg_image_white = Image.new("RGB", bg_image.size, (255, 255, 255))
 
         bg_image_white.paste(layer_image, (x, y), mask=mask)
-        bg_image=multiply_blend(bg_image_white,bg_image)
-        bg_image=bg_image.convert("RGBA")
+        bg_image = multiply_blend(bg_image_white, bg_image)
+        bg_image = bg_image.convert("RGBA")
     else:
-        transparent_img = Image.new("RGBA",layer_image.size, (255, 255, 255, 0))
-        transparent_img.paste(layer_image,(0, 0), mask)
-        # transparent_img.save('test.png')
-        bg_image.paste(transparent_img, (x, y), transparent_img)
+        transparent_img = Image.new("RGBA", layer_image.size, (255, 255, 255, 0))
+        # 调整透明度处理
+        for i in range(transparent_img.size[0]):
+            for j in range(transparent_img.size[1]):
+                r, g, b, a = transparent_img.getpixel((i, j))
+                if a > 0:
+                    transparent_img.putpixel((i, j), (r, g, b, 255))
 
+        transparent_img.paste(layer_image, (0, 0), mask)
+        bg_image.paste(transparent_img, (x, y), transparent_img)
 
     # 输出合成后的图片
     return bg_image
 
+#MixCopilot
 
 def resize_2(img):
     # 检查图像的高度是否是2的倍数，如果不是，则调整高度
@@ -955,53 +912,13 @@ def resize_image(layer_image, scale_option, width, height,color="white"):
     return layer_image
 
 
-
-
-
-
-# def generate_text_image(text_list, font_path, font_size, text_color, vertical=True, spacing=0):
-#     # Load Chinese font
-#     font = ImageFont.truetype(font_path, font_size)
-
-#     # Calculate image size based on the number of characters and orientation
-#     if vertical:
-#         width = font_size + 100
-#         height = font_size * len(text_list) + (len(text_list) - 1) * spacing + 100
-#     else:
-#         width = font_size * len(text_list) + (len(text_list) - 1) * spacing + 100
-#         height = font_size + 100
-
-#     # Create a blank image
-#     image = Image.new('RGBA', (width, height), (255, 255, 255,0))
-#     draw = ImageDraw.Draw(image)
-
-#     # Draw text
-#     if vertical:
-#         for i, char in enumerate(text_list):
-#             char_position = (50, 50 + i * font_size)
-#             draw.text(char_position, char, font=font, fill=text_color)
-#     else:
-#         for i, char in enumerate(text_list):
-#             char_position = (50 + i * (font_size + spacing), 50)
-#             draw.text(char_position, char, font=font, fill=text_color)
-
-#     # Save the image
-#     # image.save(output_image_path)
-
-#     # 分离alpha通道
-#     alpha_channel = image.split()[3]
-
-#     # 创建一个只有alpha通道的新图像
-#     alpha_image = Image.new('L', image.size)
-#     alpha_image.putdata(alpha_channel.getdata())
-
-#     image=image.convert('RGB')
-
-#     return (image,alpha_image)
-def generate_text_image(text, font_path, font_size, text_color, vertical=True, stroke=False, stroke_color=(0, 0, 0), stroke_width=1, spacing=0):
+def generate_text_image(text, font_path, font_size, text_color, vertical=True, stroke=False, stroke_color=(0, 0, 0), stroke_width=1, spacing=0, line_spacing=0,padding=4):
     # Split text into lines based on line breaks
     lines = text.split("\n")
 
+    # Load font
+    font = ImageFont.truetype(font_path, font_size)
+
     # 1. Determine layout direction
     if vertical:
         layout = "vertical"
@@ -1010,49 +927,54 @@ def generate_text_image(text, font_path, font_size, text_color, vertical=True, s
 
     # 2. Calculate absolute coordinates for each character
     char_coordinates = []
-    if layout == "vertical":
-        x = 0
-        y = 0
-        for i in range(len(lines)):
-            line = lines[i]
-            for char in line:
-                char_coordinates.append((x, y))
-                y += font_size + spacing
-            x += font_size + spacing
-            y = 0
-    else:
-        x = 0
-        y = 0
-        for line in lines:
-            for char in line:
-                char_coordinates.append((x, y))
-                x += font_size + spacing
-            y += font_size + spacing
-            x = 0
+    x, y = padding, padding
+    max_width, max_height = 0, 0
 
-    # 3. Calculate image width and height
     if layout == "vertical":
-        width = (len(lines) * (font_size + spacing)) - spacing
-        height = ((len(max(lines, key=len)) + 1) * (font_size + spacing)) + spacing
+        for line in lines:
+            max_char_width = max(font.getsize(char)[0] for char in line)
+            for char in line:
+                char_width, char_height = font.getsize(char)
+                char_coordinates.append((x, y))
+                y += char_height + spacing
+                max_height = max(max_height, y + padding)
+            x += max_char_width + line_spacing
+            y = padding
+        max_width = x
+        total_line_width = sum(font.getsize(line)[1] for line in lines)
+        total_spacing = line_spacing * (len(lines) - 1)
+        # 确保左边和右边的padding都被计入max_width
+        max_width = total_line_width + total_spacing + padding * 2
     else:
-        width = (len(max(lines, key=len)) * (font_size + spacing)) - spacing
-        height = ((len(lines) - 1) * (font_size + spacing)) + font_size
+        for line in lines:
+            line_width, line_height = font.getsize(line)
+            for char in line:
+                char_width, char_height = font.getsize(char)
+                char_coordinates.append((x, y))
+                x += char_width + spacing
+                max_width = max(max_width, x + padding)
+            y += line_height + line_spacing
+            x = padding
+        # max_height = y
+        total_line_heights = sum(font.getsize(line)[1] for line in lines)
+        total_spacing = line_spacing * (len(lines) - 1)
+        # 确保顶部和底部的padding都被计入max_height
+        max_height = total_line_heights + total_spacing + padding * 2
+
+    # 3. Create image with calculated width and height
+    image = Image.new('RGBA', (max_width, max_height), (255, 255, 255, 0))
+    draw = ImageDraw.Draw(image)
 
     # 4. Draw each character on the image
-    image = Image.new('RGBA', (width, height), (255, 255, 255, 0))
-    draw = ImageDraw.Draw(image)
-    font = ImageFont.truetype(font_path, font_size)
-
     index = 0
-    for i, line in enumerate(lines):
-        for j, char in enumerate(line):
+    for line in lines:
+        for char in line:
             x, y = char_coordinates[index]
-            
             if stroke:
-                draw.text((x-stroke_width, y), char, font=font, fill=stroke_color)
-                draw.text((x+stroke_width, y), char, font=font, fill=stroke_color)
-                draw.text((x, y-stroke_width), char, font=font, fill=stroke_color)
-                draw.text((x, y+stroke_width), char, font=font, fill=stroke_color)
+                draw.text((x-stroke_width, y), char, font=font, fill=text_color)
+                draw.text((x+stroke_width, y), char, font=font, fill=text_color)
+                draw.text((x, y-stroke_width), char, font=font, fill=text_color)
+                draw.text((x, y+stroke_width), char, font=font, fill=text_color)
             
             draw.text((x, y), char, font=font, fill=text_color)
             index += 1
@@ -1377,6 +1299,9 @@ class LoadImages_:
             image=pil2tensor(image)
             ims.append(image)
 
+        if len(ims)==0:
+            image1 = Image.new('RGB', (512, 512), color='black')
+            return (pil2tensor(image1),)
         image1 = ims[0]
         for image2 in ims[1:]:
             if image1.shape[1:] != image2.shape[1:]:
@@ -1579,7 +1504,7 @@ class ImageCropByAlpha:
 
 
 
-
+# get_files_with_extension(FONT_PATH,'.ttf')
 
 class TextImage:
     @classmethod
@@ -1587,7 +1512,7 @@ class TextImage:
         return {"required": { 
             
                     "text": ("STRING",{"multiline": True,"default": "龍馬精神迎新歲","dynamicPrompts": False}),
-                    "font_path": ("STRING",{"multiline": False,"default": FONT_PATH,"dynamicPrompts": False}),
+                    "font": (get_files_with_extension(FONT_PATH,['.ttf','.otf']),),#后缀为 ttf
                     "font_size": ("INT",{
                                 "default":100, 
                                 "min": 100, #Minimum value
@@ -1602,6 +1527,20 @@ class TextImage:
                                 "step": 1, #Slider's step
                                 "display": "number" # Cosmetic only: display as "number" or "slider"
                                 }), 
+                    "line_spacing": ("INT",{
+                                "default":12, 
+                                "min": -200, #Minimum value
+                                "max": 200, #Maximum value
+                                "step": 1, #Slider's step
+                                "display": "number" # Cosmetic only: display as "number" or "slider"
+                                }), 
+                    "padding": ("INT",{
+                                "default":8, 
+                                "min": 0, #Minimum value
+                                "max": 200, #Maximum value
+                                "step": 1, #Slider's step
+                                "display": "number" # Cosmetic only: display as "number" or "slider"
+                                }), 
                     "text_color":("STRING",{"multiline": False,"default": "#000000","dynamicPrompts": False}),
                     "vertical":("BOOLEAN", {"default": True},),
                     "stroke":("BOOLEAN", {"default": False},),
@@ -1609,7 +1548,7 @@ class TextImage:
                 }
     
     RETURN_TYPES = ("IMAGE","MASK",)
-    # RETURN_NAMES = ("WIDTH","HEIGHT","X","Y",)
+    RETURN_NAMES = ("image","mask",)
 
     FUNCTION = "run"
 
@@ -1618,11 +1557,14 @@ class TextImage:
     INPUT_IS_LIST = False
     OUTPUT_IS_LIST = (False,False,)
 
-    def run(self,text,font_path,font_size,spacing,text_color,vertical,stroke):
+    def run(self,text,font,font_size,spacing,line_spacing,padding,text_color,vertical,stroke):
         
-        # text_list=list(text)
+        font_path=os.path.join(FONT_PATH,font)
+
+        if text=="":
+            text=" "
         # stroke=False, stroke_color=(0, 0, 0), stroke_width=1, spacing=0
-        img,mask=generate_text_image(text,font_path,font_size,text_color,vertical,stroke,(0, 0, 0),1,spacing)
+        img,mask=generate_text_image(text,font_path,font_size,text_color,vertical,stroke,(0, 0, 0),1,spacing,line_spacing,padding)
         
         img=pil2tensor(img)
         mask=pil2tensor(mask)
@@ -1855,10 +1797,16 @@ class CompositeImages:
                     "mask":("MASK",),
                     "background": ("IMAGE",),
                     },
-             "optional":{
-                           
+             "optional":{ 
                   "is_multiply_blend":  ("BOOLEAN", {"default": False}),
-                  "position":  (['overall',"center_bottom","center_top","right_bottom","left_bottom","right_top","left_top"],),    
+                  "position":  (['overall',"center_center","left_bottom","center_bottom","right_bottom","left_top","center_top","right_top"],),
+                   "scale": ("FLOAT",{
+                                "default":0.35, 
+                                "min": 0.01, #Minimum value
+                                "max": 1, #Maximum value
+                                "step": 0.01, #Slider's step
+                                "display": "number" # Cosmetic only: display as "number" or "slider"
+                            }), 
                     }
                 }
     
@@ -1871,15 +1819,33 @@ class CompositeImages:
 
     # OUTPUT_IS_LIST = (True,)
 
-    def run(self, foreground,mask,background,is_multiply_blend,position):
-        foreground= tensor2pil(foreground)
-        mask= tensor2pil(mask)
-        background= tensor2pil(background)
-        res=composite_images(foreground,background,mask,is_multiply_blend,position)
+    # def run(self, foreground,mask,background,is_multiply_blend,position,scale):
+    #     foreground= tensor2pil(foreground)
+    #     mask= tensor2pil(mask)
+    #     background= tensor2pil(background)
+    #     res=composite_images(foreground,background,mask,is_multiply_blend,position,scale)
         
-        return (pil2tensor(res),)
+    #     return (pil2tensor(res),)
 
+    def run(self, foreground,mask,background, is_multiply_blend, position, scale):
+        results = []
+        
+        f1=[]
+        for fg, mask in zip(foreground, mask ):
+            f1.append([fg,mask])
 
+        
+        for f, bg in product(f1, background):
+            [fg,mask]=f
+            fg_pil = tensor2pil(fg)
+            mask_pil = tensor2pil(mask)
+            bg_pil = tensor2pil(bg)
+            res = composite_images(fg_pil, bg_pil, mask_pil, is_multiply_blend, position, scale)
+            results.append(pil2tensor(res))
+        
+        output_image = torch.cat(results, dim=0)
+
+        return (output_image,)
 
 
 class EmptyLayer:
@@ -3256,4 +3222,4 @@ class ImageListToBatch_:
 
         out = torch.cat(out, dim=0)
 
-        return (out,)
+        return (out,)

nodes/Lama.py

@@ -85,8 +85,6 @@ class LaMaInpainting:
             "image": ("IMAGE",),
             "mask": ("MASK",),
                              },
-
-            
                 }
     
     RETURN_TYPES = ("IMAGE",)

nodes/ScreenShareNode.py

@@ -90,7 +90,7 @@ class ScreenShareNode:
                 } }
     
     RETURN_TYPES = ('IMAGE','STRING','FLOAT',"INT")
-    RETURN_NAMES = ("IMAGE","PROMPT","FLOAT","INT")
+    RETURN_NAMES = ("current frame (image)","prompt","denoise (float)","seed (int)")
     FUNCTION = "run"
 
     CATEGORY = "♾️Mixlab/Screen"
@@ -109,7 +109,7 @@ class FloatingVideo:
     @classmethod
     def INPUT_TYPES(s):
         return { "required":{
-            "images": ("IMAGE",)
+            "image": ("IMAGE",)
         }, }
     
     # RETURN_TYPES = ('IMAGE','MASK')
@@ -124,16 +124,16 @@ class FloatingVideo:
     # OUTPUT_IS_LIST = (False,False,)
   
     # 运行的函数
-    def run(self,images):
+    def run(self,image):
        
         results = list()
     
-        for image in images:
-            image=tensor2pil(image)
+        for im in image:
+            im=tensor2pil(im)
             # image_base64 = base64.b64encode(image.tobytes())
 
             buffered = BytesIO()
-            image.save(buffered, format="JPEG")
+            im.save(buffered, format="JPEG")
             image_base64 = base64.b64encode(buffered.getvalue()).decode("utf-8")
 
             results.append(image_base64)

nodes/Utils.py

@@ -133,7 +133,7 @@ def get_font_files(directory):
 
     return font_files
 
-r_directory = os.path.join(os.path.dirname(__file__), '../assets/')
+r_directory = os.path.join(os.path.dirname(__file__), '..','assets','/')
 
 font_files = get_font_files(r_directory)
 # print(font_files)
@@ -566,7 +566,7 @@ class AppInfo:
                              },
 
                 "optional":{
-                    "IMAGE": ("IMAGE",),
+                    "image": ("IMAGE",),
                     "description":("STRING",{"multiline": True,"default": "","dynamicPrompts": False}),
                     "version":("INT", {
                         "default": 1, 
@@ -594,12 +594,12 @@ class AppInfo:
     INPUT_IS_LIST = True
     # OUTPUT_IS_LIST = (True,)
 
-    def run(self,name,input_ids,output_ids,IMAGE,description,version,share_prefix,link,category,auto_save):
+    def run(self,name,input_ids,output_ids,image,description,version,share_prefix,link,category,auto_save):
         name=name[0]
         
         im=None
-        if IMAGE:
-            im=IMAGE[0][0]
+        if image:
+            im=image[0][0]
             #TODO batch 的方式需要处理
             im=create_temp_file(im)
         # image [img,] img[batch,w,h,a] 列表里面是batch，

nodes/Video.py

@@ -179,6 +179,9 @@ def split_video(video_path, video_segment_frames, transition_frames, output_dir)
         
         # 打印当前片段的起始帧和结束帧
         print(f"Segment {i+1}: Start Frame {start_frame}, End Frame {end_frame}")
+
+        if end_frame<start_frame:
+            break
         
         # 保存当前片段为一个视频文件
         segment_video_path = f"{output_dir}/segment_{i+1}.avi"
@@ -187,6 +190,7 @@ def split_video(video_path, video_segment_frames, transition_frames, output_dir)
         segment_video = cv2.VideoWriter(segment_video_path, fourcc, fps, (int(video_capture.get(cv2.CAP_PROP_FRAME_WIDTH)),
                                                               int(video_capture.get(cv2.CAP_PROP_FRAME_HEIGHT))))
         
+
         for frame_num in range(start_frame, end_frame):
             ret, frame = video_capture.read()
             if ret:
@@ -526,30 +530,39 @@ class LoadAndCombinedAudio_:
 
     CATEGORY = "♾️Mixlab/Audio"
 
-    RETURN_TYPES = ("STRING",)
-    RETURN_NAMES = ("audio_file_path",)
+    RETURN_TYPES = ("STRING","AUDIO",)
+    RETURN_NAMES = ("audio_file_path","audio",)
     FUNCTION = "run"
 
     def run(self,audios, start_time, duration):
         output_dir = folder_paths.get_output_directory()
         counter=get_new_counter(output_dir,'audio_')
 
-        audio_file = f"audio_{counter:05}.wav"
+        audio_file_name = f"audio_{counter:05}.wav"
 
-        audio_file=save_audio_base64s_to_file(audios['base64'],output_dir,audio_file)
+        audio_file=save_audio_base64s_to_file(audios['base64'],output_dir,audio_file_name)
         # duration == -1 则不裁切
         if duration > -1:
             crop_audio(audio_file, start_time, duration)
 
-        return (audio_file,)
+        waveform, sample_rate = torchaudio.load(audio_file)
+        audio = {
+            "filename": audio_file_name,
+            "subfolder": "",
+            "type": "output",
+            "audio_path":audio_file,
+            "waveform": waveform.unsqueeze(0), 
+            "sample_rate": sample_rate}
+
+        return (audio_file,audio ,)
 
 class CombineAudioVideo:
     @classmethod
     def INPUT_TYPES(s):
        
         return {"required": {
-                     "video_file_path": ("STRING",  {"forceInput": True}),
-                     "audio_file_path": ("STRING",  {"forceInput": True}), 
+                     "video": ("SCENE_VIDEO",),
+                     "audio": ("AUDIO", ), 
                      },
                 }
 
@@ -557,23 +570,46 @@ class CombineAudioVideo:
 
     OUTPUT_NODE = True
     FUNCTION = "run" 
-    RETURN_TYPES = ()
-    RETURN_NAMES = ()
+    RETURN_TYPES = ("SCENE_VIDEO",)
+    RETURN_NAMES = ("SCENE_VIDEO",)
 
-    def run(self,video_file_path, audio_file_path):
+    def run(self,video, audio):
 
         output_dir = folder_paths.get_output_directory()
 
-        counter=get_new_counter(output_dir,'video_final_')
-        
+        # 判断是否是 Tensor 类型
+        is_tensor = not isinstance(audio, dict)
+        # print('#判断是否是 Tensor 类型',is_tensor,audio)
+        if not is_tensor and 'waveform' in audio and 'sample_rate' in audio:
+            # {'waveform': tensor([], size=(1, 1, 0)), 'sample_rate': 44100}
+            is_tensor=True
+
+        if "audio_path" in audio:
+            is_tensor=False
+            audio_file_path=audio["audio_path"]
+
+        if is_tensor:
+            filename_prefix="audio_tmp"
+            full_output_folder, filename, counter, subfolder, filename_prefix = folder_paths.get_save_image_path(
+                filename_prefix, 
+                folder_paths.get_temp_directory())
+            
+            filename_with_batch_num = filename.replace("%batch_num%", str(1))
+            file = f"{filename_with_batch_num}_{counter:05}_.wav"
+            
+            audio_file_path=os.path.join(full_output_folder, file)
+
+            torchaudio.save(audio_file_path, audio['waveform'].squeeze(0), audio["sample_rate"])
+            
         # 获取文件名和扩展名
-        base, ext = os.path.splitext(video_file_path)
+        base, ext = os.path.splitext(video)
+        counter=get_new_counter(output_dir,'video_final_')
 
         v_file = f"video_final_{counter:05}{ext}"
         
         v_file_path=os.path.join(output_dir, v_file)
 
-        combine_audio_video(audio_file_path,video_file_path,v_file_path)
+        combine_audio_video(audio_file_path,video,v_file_path)
 
         previews = [
             {
@@ -583,7 +619,8 @@ class CombineAudioVideo:
                 "format": get_mime_type(v_file),
             }
         ]
-        return {"ui": {"gifs": previews}}
+
+        return {"ui": {"gifs": previews},"result":(v_file_path,)}
 
 # The code is based on ComfyUI-VideoHelperSuite modification.
 class VideoCombine_Adv:
@@ -615,8 +652,8 @@ class VideoCombine_Adv:
             },
         }
 
-    RETURN_TYPES = ("STRING",)
-    RETURN_NAMES = ("video_file_path",)
+    RETURN_TYPES = ("SCENE_VIDEO",)
+    RETURN_NAMES = ("scenes_video",)
     OUTPUT_NODE = True
     CATEGORY = "♾️Mixlab/Video"
     FUNCTION = "run"
@@ -890,12 +927,12 @@ class scenesNode_:
         
         return {"required": {
                     "scenes_video": ('SCENE_VIDEO',),
-                     "index": ("INT", {"default": 0, "min": 0, "step": 1}),
+                    "index": ("INT", {"default": 0, "min": 0, "step": 1}),
                      
                      },}
 
     RETURN_TYPES = ('IMAGE','INT',)
-    RETURN_NAMES = ("frames","count",)
+    RETURN_NAMES = ("video frames (batch)","count",)
     # OUTPUT_IS_LIST = (False,)
 
     FUNCTION = "run"
@@ -903,7 +940,7 @@ class scenesNode_:
     INPUT_IS_LIST = True
 
     def load_video_cv_fallback(self, video, frame_load_cap, skip_first_frames):
-        print('#video',video)
+        # print('#video',video)
         try:
             video_cap = cv2.VideoCapture(video)
             if not video_cap.isOpened():
@@ -958,7 +995,7 @@ class scenesNode_:
         index=index[0]
         if len(scenes_video) > index:
             vp=scenes_video[index]
-
-            return self.load_video_cv_fallback(vp,0,0)
+        else:
+            vp=scenes_video[-1]
         
-        return ([], 0,)
+        return self.load_video_cv_fallback(vp,0,0)

pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "comfyui-mixlab-nodes"
 description = "3D, ScreenShareNode & FloatingVideoNode, SpeechRecognition & SpeechSynthesis, GPT, LoadImagesFromLocal, Layers, Other Nodes, ..."
-version = "0.28.3"
+version = "0.32.0"
 license = "MIT"
 dependencies = ["numpy", "pyOpenSSL", "watchdog", "opencv-python-headless", "matplotlib", "openai", "simple-lama-inpainting", "clip-interrogator==0.6.0", "transformers>=4.36.0", "lark-parser", "imageio-ffmpeg", "rembg[gpu]", "omegaconf==2.3.0", "Pillow>=9.5.0", "einops==0.7.0", "trimesh>=4.0.5", "huggingface-hub", "scikit-image"]
 

requirements.txt

@@ -17,4 +17,4 @@ trimesh>=4.0.5
 huggingface-hub
 scikit-image
 torchaudio
-soundfile
+soundfile>=0.12.1

web/index.html

@@ -641,8 +641,13 @@
             imageElement.appendChild(im);
 
             let base64s = imageElement.querySelectorAll('.base64')
-            //更新输入
-            window._appData.data[nodeId].inputs.images.base64 = Array.from(base64s, (b) => b.src)
+            //更新输入 //兼容 ，如果data没有则不更新data
+            if (window._appData.data) window._appData.data[nodeId].inputs.images.base64 = Array.from(base64s, (b) => b.src)
+            //更新输入参数
+            updateInputData(nodeId, (inputs) => {
+                inputs.images.base64 = Array.from(base64s, (b) => b.src)
+                return inputs
+            })
 
             // 删除
             im.addEventListener('click', e => {
@@ -650,7 +655,11 @@
                 im.remove();
                 let base64s = imageElement.querySelectorAll('.base64')
                 //更新输入
-                window._appData.data[nodeId].inputs.images.base64 = Array.from(base64s, (b) => b.src)
+                if (window._appData.data) window._appData.data[nodeId].inputs.images.base64 = Array.from(base64s, (b) => b.src)
+                updateInputData(nodeId, (inputs) => {
+                    inputs.images.base64 = Array.from(base64s, (b) => b.src)
+                    return inputs
+                })
             })
         }
 
@@ -974,7 +983,7 @@
                     // // 获取读取的文件内容，即 Blob 对象
                     let hashId = await calculateImageHash(fileBlob)
 
-                    if (hashId == window._appData.data[data.id].hashId) {
+                    if (window._appData.data && hashId == window._appData.data[data.id].hashId) {
                         document.body.querySelector('.app').style.display = 'flex'
                         document.body.querySelector('#author').style.display = 'block'
                         return;
@@ -986,8 +995,15 @@
                     let { url, name } = await uploadMask(fileBlob, imgurl);
                     // 在这里可以对 Blob 对象进行进一步处理
                     // imageElement.src = url;
-                    window._appData.data[data.id].inputs.image = name;
-                    window._appData.data[data.id].hashId = hashId;
+                    if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.image = name;
+                    if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].hashId = hashId;
+
+                    //更新输入参数
+                    updateInputData(data.id, (inputs) => {
+                        inputs.image = name
+                        return inputs
+                    })
+
 
                     // console.log("上传的文件：", url, data.id, name);
                     //更新图片
@@ -1116,10 +1132,31 @@
 
         function updateSeed(id, val) {
             // console.log(val)
-            if (window._appData.data[id].inputs.seed && !Array.isArray(window._appData.data[id].inputs.seed)) window._appData.data[id].inputs.seed = Math.round(val);
-            if (window._appData.data[id].inputs.noise_seed && !Array.isArray(window._appData.data[id].inputs.noise_seed)) window._appData.data[id].inputs.noise_seed = Math.round(val);
+            if (window._appData.data
+                && window._appData.data[id].inputs.seed
+                && !Array.isArray(window._appData.data[id].inputs.seed)) window._appData.data[id].inputs.seed = Math.round(val);
+            if (window._appData.data
+                && window._appData.data[id].inputs.noise_seed
+                && !Array.isArray(window._appData.data[id].inputs.noise_seed)) window._appData.data[id].inputs.noise_seed = Math.round(val);
+
+            //更新输入参数
+            updateInputData(id, (inputs) => {
+                inputs.seed = Math.round(val);
+                inputs.noise_seed = Math.round(val);
+                return inputs
+            })
+
         }
 
+        //更新输入参数
+        function updateInputData(nodeId, callback) {
+            for (let index = 0; index < window._appData.input.length; index++) {
+                const inp = window._appData.input[index];
+                if (inp.id === nodeId) {
+                    window._appData.input[index].inputs = callback(window._appData.input[index].inputs);
+                }
+            }
+        }
 
         function queuePrompt(appInfo, promptWorkflow, seed, client_id) {
             // appinfo升级后 兼容，补丁
@@ -1131,21 +1168,6 @@
             // 随机seed
             promptWorkflow = randomSeed(seed, promptWorkflow);
 
-            // //动态提示,改为输入的时候，手动触发
-            // for (const id in promptWorkflow) {
-            //     let node = promptWorkflow[id]
-            //     if (["TextInput_", "CLIPTextEncode", "PromptSimplification", "ChinesePrompt_Mix"].includes(
-            //         node.class_type
-            //     )) {
-            //         if (node.class_type == "PromptSimplification") {
-            //             promptWorkflow[id].inputs.prompt = dynamicPrompts(node.inputs.prompt);
-            //         } else {
-            //             promptWorkflow[id].inputs.text = dynamicPrompts(node.inputs.text);
-            //         }
-            //         console.log('#动态提示', promptWorkflow[id].inputs)
-            //     }
-            // }
-
             let url = get_url()
             const data = JSON.stringify({ prompt: promptWorkflow, client_id });
             fetch(`${url}/prompt`, {
@@ -1170,6 +1192,24 @@
         }
 
 
+        // 新的运行工作流的接口
+        function queuePromptNew(filename, category, seed, input, client_id) {
+            let url = get_url()
+            // var filename = "Text-to-Image_1.json", category = "";
+
+            // 随机seed
+            //  promptWorkflow = randomSeed(seed, promptWorkflow);
+
+            const data = JSON.stringify({ filename, category, seed, input, client_id });
+            fetch(`${url}/mixlab/prompt`, {
+                method: 'POST',
+                headers: {
+                    'Content-Type': 'application/json',
+                },
+                body: data,
+            })
+        }
+
         function success(isSuccess, btn, text) {
             isSuccess ? btn.innerText = 'success' : text;
             setTimeout(() => {
@@ -1488,7 +1528,7 @@
                             // // 获取读取的文件内容，即 Blob 对象
                             let hashId = await calculateImageHash(fileBlob)
 
-                            if (hashId == window._appData.data[data.id].hashId) return
+                            if (window._appData.data && hashId == window._appData.data[data.id].hashId) return
 
                             let base64 = await blobToBase64(fileBlob)
 
@@ -1498,10 +1538,16 @@
                                 let { url, name } = await uploadImage(fileBlob);
                                 // 在这里可以对 Blob 对象进行进一步处理
                                 imageElement.src = url;
-                                window._appData.data[data.id].inputs.image = name;
-                                window._appData.data[data.id].hashId = hashId;
+                                if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.image = name;
+                                if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].hashId = hashId;
                                 console.log("上传的文件：", url, data.id, name);
 
+                                //更新输入参数
+                                updateInputData(data.id, (inputs) => {
+                                    inputs.image = name;
+                                    return inputs
+                                })
+
                                 //更换option里的default image
                                 window._appData.input = Array.from(window._appData.input, inp => {
                                     if (inp.id === data.id) {
@@ -1792,11 +1838,26 @@
                             data.options.images,
                             data.inputs.imageIndex,
                             (base64, text) => {
-                                window._appData.data[data.id].inputs.image_base64 = base64;
-                                window._appData.data[data.id].inputs.text = text;
+                                if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.image_base64 = base64;
+                                if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.text = text;
+
+                                //更新输入参数
+                                updateInputData(data.id, (inputs) => {
+                                    inputs.image_base64 = base64;
+                                    inputs.text = text;
+                                    return inputs
+                                })
+
                             })
                         uploadContainer.appendChild(imgDiv);
-                        window._appData.data[data.id].inputs.image_base64 = mainImage.querySelector('.images_prompt_main').src;
+                        if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.image_base64 = mainImage.querySelector('.images_prompt_main').src;
+
+                        //更新输入参数
+                        updateInputData(data.id, (inputs) => {
+                            inputs.image_base64 = mainImage.querySelector('.images_prompt_main').src;
+                            return inputs
+                        })
+
                     } else if (data.class_type === 'LoadImagesToBatch') {
                         // 多张base64 图片
                         let base64 = data.inputs.images.base64
@@ -1836,7 +1897,7 @@
                             // console.log( file.type.split('/')[1])
                             let hashId = await calculateImageHash(fileBlob)
 
-                            if (hashId == window._appData.data[data.id].hashId) return
+                            if (window._appData.data && hashId == window._appData.data[data.id].hashId) return
 
                             if (data.class_type === 'LoadImagesToBatch') {
                                 // 上传 ，转为base64
@@ -1849,8 +1910,16 @@
                                 let base64 = await parseImageToBase64(url);
 
                                 if (data.class_type === 'ImagesPrompt_') {
-                                    uploadContainer.querySelector('.images_prompt_main').src = base64
-                                    window._appData.data[data.id].inputs.image_base64 = base64;
+                                    uploadContainer.querySelector('.images_prompt_main').src = base64;
+
+                                    if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.image_base64 = base64;
+
+                                    //更新输入参数
+                                    updateInputData(data.id, (inputs) => {
+                                        inputs.image_base64 = base64;
+                                        return inputs
+                                    })
+
                                 } else {
                                     if (isVideoUpload) {
                                         imageElement.srcObject = null;
@@ -1859,7 +1928,14 @@
                                     imageElement.src = url;
 
                                     if (isVideoUpload) {
-                                        window._appData.data[data.id].inputs.video = name;
+                                        if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.video = name;
+
+                                        //更新输入参数
+                                        updateInputData(data.id, (inputs) => {
+                                            inputs.video = name;
+                                            return inputs
+                                        })
+
                                     } else {
                                         //更换option里的default image
                                         window._appData.input = Array.from(window._appData.input, inp => {
@@ -1869,13 +1945,19 @@
                                             return inp
                                         })
 
-                                        window._appData.data[data.id].inputs.image = name;
+                                        if (window._appData.data && window._appData.data[data.id]) window._appData.data[data.id].inputs.image = name;
+                                        //更新输入参数
+                                        updateInputData(data.id, (inputs) => {
+                                            inputs.image = name;
+                                            return inputs
+                                        })
+
                                     }
 
                                 }
 
 
-                                window._appData.data[data.id].hashId = hashId;
+                                if (window._appData.data) window._appData.data[data.id].hashId = hashId;
 
                                 console.log("上传的文件：", url, data.id, name);
                             }
@@ -1915,7 +1997,12 @@
                     let silde = createNumSlide(label,
                         data.inputs.weight,
                         (v) => {
-                            window._appData.data[data.id].inputs.weight = parseFloat(v);
+                            if (window._appData.data) window._appData.data[data.id].inputs.weight = parseFloat(v);
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.weight = parseFloat(v);
+                                return inputs
+                            })
                         },
                         options.min,
                         options.max,
@@ -1939,7 +2026,14 @@
                         data.inputs.number,
                         (v) => {
                             // console.log(data.id,window._appData.data[data.id])
-                            window._appData.data[data.id].inputs.number = data.class_type === 'IntNumber' ? parseInt(v) : parseFloat(v);
+                            if (window._appData.data) window._appData.data[data.id].inputs.number = data.class_type === 'IntNumber' ? parseInt(v) : parseFloat(v);
+
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.number = data.class_type === 'IntNumber' ? parseInt(v) : parseFloat(v);
+                                return inputs
+                            })
+
                         },
                         options.min,
                         options.max,
@@ -1979,9 +2073,21 @@
                         textInput.style.height = height;
 
                         if (data.class_type == "PromptSimplification") {
-                            window._appData.data[data.id].inputs.prompt = textInput.value;
+                            if (window._appData.data) window._appData.data[data.id].inputs.prompt = textInput.value;
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.prompt = textInput.value;
+                                return inputs
+                            })
                         } else {
-                            window._appData.data[data.id].inputs.text = textInput.value;
+                            if (window._appData.data) window._appData.data[data.id].inputs.text = textInput.value;
+
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.text = textInput.value;
+                                return inputs
+                            })
+
                         }
 
                     } catch (error) {
@@ -2005,9 +2111,19 @@
                         textInput.setAttribute('title', prompt)
                         dynamicPromptsBtn.setAttribute('title', prompt)
                         if (data.class_type == "PromptSimplification") {
-                            window._appData.data[data.id].inputs.prompt = prompt;
+                            if (window._appData.data) window._appData.data[data.id].inputs.prompt = prompt;
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.prompt = prompt;
+                                return inputs
+                            })
                         } else {
-                            window._appData.data[data.id].inputs.text = prompt;
+                            if (window._appData.data) window._appData.data[data.id].inputs.text = prompt;
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.text = prompt;
+                                return inputs
+                            })
                         }
                     })
 
@@ -2022,9 +2138,19 @@
                         autoResize(textInput);
 
                         if (data.class_type == "PromptSimplification") {
-                            window._appData.data[data.id].inputs.prompt = textInput.value;
+                            if (window._appData.data) window._appData.data[data.id].inputs.prompt = textInput.value;
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.prompt = textInput.value;
+                                return inputs
+                            })
                         } else {
-                            window._appData.data[data.id].inputs.text = textInput.value;
+                            if (window._appData.data) window._appData.data[data.id].inputs.text = textInput.value;
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.text = textInput.value;
+                                return inputs
+                            })
                         }
                         localStorage.setItem(`t_${data.id}`, JSON.stringify({
                             value: textInput.value,
@@ -2071,10 +2197,24 @@
                         if (v) {
                             value = v;
                             if (data.class_type === 'CheckpointLoaderSimple') {
-                                window._appData.data[data.id].inputs.ckpt_name = value;
+
+                                if (window._appData.data) window._appData.data[data.id].inputs.ckpt_name = value;
+                                //更新输入参数
+                                updateInputData(data.id, (inputs) => {
+                                    inputs.ckpt_name = value;
+                                    return inputs
+                                })
+
                             }
                             if (data.class_type === 'LoraLoader') {
-                                window._appData.data[data.id].inputs.lora_name = value;
+
+                                if (window._appData.data) window._appData.data[data.id].inputs.lora_name = value;
+                                //更新输入参数
+                                updateInputData(data.id, (inputs) => {
+                                    inputs.lora_name = value;
+                                    return inputs
+                                })
+
                             }
                         }
                     } catch (error) {
@@ -2094,10 +2234,20 @@
                         e.preventDefault();
                         // console.log(selectDom.value)
                         if (data.class_type === 'CheckpointLoaderSimple') {
-                            window._appData.data[data.id].inputs.ckpt_name = selectDom.value;
+                            if (window._appData.data) window._appData.data[data.id].inputs.ckpt_name = selectDom.value;
+                            //更新输入参数
+                            updateInputData(nodeId, (inputs) => {
+                                inputs.ckpt_name = selectDom.value;
+                                return inputs
+                            })
                         }
                         if (data.class_type === 'LoraLoader') {
-                            window._appData.data[data.id].inputs.lora_name = selectDom.value;
+                            if (window._appData.data) window._appData.data[data.id].inputs.lora_name = selectDom.value;
+                            //更新输入参数
+                            updateInputData(nodeId, (inputs) => {
+                                inputs.lora_name = selectDom.value;
+                                return inputs
+                            })
                         }
 
                         localStorage.setItem(`_model_${data.id}_${data.class_type}`, selectDom.value)
@@ -2161,7 +2311,15 @@
 
                             audioE.src = base64;
 
-                            window._appData.data[data.id].inputs.audios.base64 = [base64];
+                            if (window._appData.data) window._appData.data[data.id].inputs.audios.base64 = [base64];
+
+                            //更新输入参数
+                            updateInputData(data.id, (inputs) => {
+                                inputs.audios.base64 = [base64];
+                                return inputs
+                            })
+
+
                         })
 
                         inp.click()
@@ -2225,11 +2383,26 @@
                         // console.log(color)
                         // window._appData.data[data.id].inputs.color.hex = color.toHEXA().toString();
                         let [r, g, b, a] = color.toRGBA();
-                        window._appData.data[nodeId].inputs.color = {
-                            ...window._appData.data[nodeId].inputs.color,
+                        r = parseInt(r);
+                        g = parseInt(g);
+                        b = parseInt(b);
+
+                        if (window._appData.data) window._appData.data[nodeId].inputs.color = {
+                            // ...window._appData.data[nodeId].inputs.color,
                             r, g, b, a,
                             hex: color.toHEXA().toString()
-                        }
+                        };
+
+                        //更新输入参数 //todo color里的参数
+                        updateInputData(nodeId, (inputs) => {
+                            inputs.color = {
+                                // ...window._appData.data[nodeId].inputs.color,
+                                r, g, b, a,
+                                hex: color.toHEXA().toString()
+                            };
+                            return inputs
+                        })
+
                     } catch (error) { }
                 })
                 .on('cancel', instance => {
@@ -2308,7 +2481,14 @@
 
                 let defaultValue = (targetId ? localStorage.getItem(`_slide_${targetId}`) : '') || keywords[0];
 
-                window._appData.data[targetId].inputs.prompt_keyword = defaultValue;
+                if (window._appData.data) window._appData.data[targetId].inputs.prompt_keyword = defaultValue;
+
+                //更新输入参数 //todo color里的参数
+                updateInputData(targetId, (inputs) => {
+                    inputs.prompt_keyword = defaultValue;
+                    return inputs
+                })
+
 
                 let selectTag = createSelect(Array.from(keywords, (k, i) => {
                     return {
@@ -2325,7 +2505,13 @@
                 // selectTag.setAttribute('data-content',labelText);
                 selectTag.addEventListener('change', e => {
                     e.preventDefault();
-                    window._appData.data[targetId].inputs.prompt_keyword = selectTag.value;
+                    if (window._appData.data) window._appData.data[targetId].inputs.prompt_keyword = selectTag.value;
+
+                    //更新输入参数 //todo color里的参数
+                    updateInputData(targetId, (inputs) => {
+                        inputs.prompt_keyword = selectTag.value;
+                        return inputs
+                    })
 
                     targetId ? localStorage.setItem(`_slide_${targetId}`, selectTag.value) : ''
                 })
@@ -2501,6 +2687,7 @@
         async function createUI(data, share = true) {
             // appData.input, appData.output, appData.seed, share, appData.link
             if (!data) return
+            // console.log('#createUI', data)
             const { input: inputData, output: outputData, data: workflow, seed, seedTitle, link, name } = data;
 
             let mainDiv = document.createElement('div');
@@ -2584,6 +2771,7 @@
             // seeds.textContent = 'Status';
             seeds.className = 'seeds';
 
+            console.log('#createUI', seed, data.data)
             try {
                 if (Object.keys(seed).length > 0) {
                     seeds.innerHTML = `<summary>SEED</summary>
@@ -2591,7 +2779,8 @@
                     const content = seeds.querySelector('.content')
                     for (const id in seed) {
                         const s = seed[id];
-                        if (!Array.isArray(workflow[id].inputs.seed)) {
+                        console.log('#createUI', !!(workflow && !Array.isArray(workflow[id].inputs.seed) || !workflow))
+                        if (!!(workflow && !Array.isArray(workflow[id].inputs.seed) || !workflow)) {
 
                             let seedInput = document.createElement('div');
                             content.appendChild(seedInput)
@@ -2670,7 +2859,7 @@
             leftDiv.appendChild(input1);
 
             submitDivBtn.appendChild(submitButton);
-            if (typeof (data.data) == 'object') mainDiv.appendChild(submitDiv);
+            mainDiv.appendChild(submitDiv);
 
             rightDiv.appendChild(output);
 
@@ -3041,13 +3230,22 @@
             ui.submitButton.update(
                 () => {
                     // 在提交按钮点击时执行的逻辑
-                    queuePrompt({
-                        name: window._appData.name,
-                        id: window._appData.id,
-                        icon: window._appData.icon,
-                        category: window._appData.category,
-                        filename: window._appData.filename
-                    }, window._appData.data, window._appData.seed, api.clientId);
+                    // queuePrompt({
+                    //     name: window._appData.name,
+                    //     id: window._appData.id,
+                    //     icon: window._appData.icon,
+                    //     category: window._appData.category,
+                    //     filename: window._appData.filename
+                    // }, window._appData.data, window._appData.seed, api.clientId);
+
+                    queuePromptNew(
+                        window._appData.filename,
+                        window._appData.category,
+                        window._appData.seed,
+                        window._appData.input.filter(inp => inp),
+                        api.clientId
+                    )
+
                 }, () => {
                     // 取消
                     if (api.runningCancel) {
@@ -3460,7 +3658,7 @@
 
                 createAppList(window._apps);
 
-                if (window._apps.length > 0) await createHistoryList(category || '');
+                // if (window._apps.length > 0) await createHistoryList(category || '');
 
                 createApp(window._appData);
             }

web/javascript/app_mixlab.js

@@ -418,7 +418,7 @@ async function save (json, download = false, showInfo = true) {
 
 function getInputsAndOutputs () {
   const inputs =
-      `LoadImage LoadImagesToBatch ImagesPrompt_ LoadAndCombinedAudio_ VHS_LoadVideo CLIPTextEncode PromptSlide TextInput_ Color FloatSlider IntNumber CheckpointLoaderSimple LoraLoader`.split(
+      `LoadImage LoadImagesToBatch ImagesPrompt_ LoadAndCombinedAudio_ LoadVideoAndSegment_ VHS_LoadVideo CLIPTextEncode PromptSlide TextInput_ Color FloatSlider IntNumber CheckpointLoaderSimple LoraLoader`.split(
         ' '
       ),
     outputs =
@@ -466,19 +466,18 @@ app.registerExtension({
         const { input, output } = getInputsAndOutputs()
         input_ids.value = input.join('\n')
         output_ids.value = output.join('\n')
-
         const widget = {
           type: 'div',
           name: 'AppInfoRun',
           draw (ctx, node, widget_width, y, widget_height) {
             Object.assign(
               this.div.style,
-              get_position_style(
+              {...get_position_style(
                 ctx,
                 widget_width,
                 node.size[1] - widget_height,
                 node.size[1]
-              )
+              ),zIndex:1}
             )
           }
         }
@@ -526,7 +525,7 @@ app.registerExtension({
 
         //td bg
         const tdBG = document.createElement('button')
-        tdBG.innerText = 'TDBG'
+        tdBG.innerText = 'Canvas Mode'
         tdBG.style = style
         tdBG.style.marginLeft = '12px'
 
@@ -709,6 +708,7 @@ app.registerExtension({
         this.serialize_widgets = true //需要保存参数
 
         window._mixlab_app_json = null
+ 
       }
 
       const onExecuted = nodeType.prototype.onExecuted
@@ -725,7 +725,7 @@ app.registerExtension({
 
           const div = this.widgets.filter(w => w.div)[0].div
           Array.from(div.querySelectorAll('button'), b =>
-            b.innerText != 'TDBG' ? (b.style.background = 'yellow') : ''
+            b.innerText != 'Canvas Mode' ? (b.style.background = 'yellow') : ''
           )
         } catch (error) {}
       }

web/javascript/audio_mixlab.js

@@ -444,13 +444,23 @@ const createInputAudioForBatch = (base64, widget) => {
   // Create a delete button
   let deleteButton = document.createElement('button')
   deleteButton.textContent = 'Delete'
-  deleteButton.style = 'margin-left: 10px;'
+
+  deleteButton.style = `cursor: pointer;
+        font-weight: 300;
+        margin: 2px; 
+        margin-left: 10px;
+        color: var(--descrip-text);
+        background-color: var(--comfy-input-bg);
+        border-radius: 8px;
+        border-color: var(--border-color);
+        border-style: solid;height: 30px;min-width: 122px;
+       `
 
   // Create a container for the audio and delete button
   let container = document.createElement('div')
   container.appendChild(audio)
   container.appendChild(deleteButton)
-  container.style= `display: flex`
+  container.style = `display: flex;margin-top: 12px;`
 
   // Add event listener for the delete button
   deleteButton.addEventListener('click', e => {
@@ -545,6 +555,7 @@ app.registerExtension({
           e.preventDefault()
           let inputAudio = document.createElement('input')
           inputAudio.type = 'file'
+          inputAudio.accept = "audio/*"
           inputAudio.style.display = 'none'
           inputAudio.addEventListener('change', async e => {
             e.preventDefault()
@@ -571,7 +582,7 @@ app.registerExtension({
         // document.addEventListener('wheel', handleMouseWheel)
 
         const onRemoved = this.onRemoved
-        this.onRemoved = () => { 
+        this.onRemoved = () => {
           widget.div.remove()
           try {
             // document.removeEventListener('wheel', handleMouseWheel)

web/javascript/checkVersion_mixlab.js

@@ -3,7 +3,7 @@ import { app } from '../../../scripts/app.js'
 const repoOwner = 'shadowcz007' // 替换为仓库的所有者
 const repoName = 'comfyui-mixlab-nodes' // 替换为仓库的名称
 
-const version = 'v0.30.0'
+const version = 'v0.32.0'
 
 fetch(`https://api.github.com/repos/${repoOwner}/${repoName}/releases/latest`)
   .then(response => response.json())

web/javascript/gpt_mixlab.js

@@ -199,6 +199,120 @@ app.registerExtension({
         url[id] || 'https://api.openai.com/v1'
     }
   }
+});
+
+
+app.registerExtension({
+  name: 'Mixlab.GPT.SiliconflowLLM',
+  async getCustomWidgets (app) {
+    return {
+      KEY (node, inputName, inputData, app) {
+        // console.log('##inputData', inputData)
+        const widget = {
+          type: inputData[0], // the type, CHEESE
+          name: inputName, // the name, slice
+          size: [128, 32], // a default size
+          draw (ctx, node, width, y) {},
+          computeSize (...args) {
+            return [128, 32] // a method to compute the current size of the widget
+          },
+          async serializeValue (nodeId, widgetIndex) {
+            let data = getLocalData('_mixlab_api_key')
+            return data[node.id] || 'by Mixlab'
+          }
+        }
+        //  widget.something = something;          // maybe adds stuff to it
+        node.addCustomWidget(widget) // adds it to the node
+        return widget // and returns it.
+      },
+    }
+  },
+
+  async beforeRegisterNodeDef (nodeType, nodeData, app) {
+    if (nodeType.comfyClass == 'SiliconflowLLM') {
+      const orig_nodeCreated = nodeType.prototype.onNodeCreated
+      nodeType.prototype.onNodeCreated = function () {
+        orig_nodeCreated?.apply(this, arguments)
+
+        const api_key = this.widgets.filter(w => w.name == 'api_key')[0]
+       
+        const widget = {
+          type: 'div',
+          name: 'chatgptdiv',
+          draw (ctx, node, widget_width, y, widget_height) {
+            Object.assign(
+              this.div.style,
+              get_position_style(ctx, widget_width, api_key.y, node.size[1])
+            )
+          }
+        }
+
+        widget.div = $el('div', {})
+
+        document.body.appendChild(widget.div)
+
+        const inputDiv = (key, placeholder) => {
+          let div = document.createElement('div')
+          const ip = document.createElement('input')
+          ip.type = placeholder === 'Key' ? 'password' : 'text'
+          ip.className = `${'comfy-multiline-input'} ${placeholder}`
+          div.style = `display: flex;
+          align-items: center; 
+          margin: 6px 8px;
+          margin-top: 0;`
+          ip.placeholder = placeholder
+          ip.value = placeholder
+
+          ip.style = `margin-left: 24px;
+          outline: none;
+          border: none;
+          padding: 4px;width: 100%;`
+          const label = document.createElement('label')
+          label.style = 'font-size: 10px;min-width:32px'
+          label.innerText = placeholder
+          div.appendChild(label)
+          div.appendChild(ip)
+
+          ip.addEventListener('change', () => {
+            let data = getLocalData(key)
+            data[this.id] = ip.value.trim()
+            localStorage.setItem(key, JSON.stringify(data))
+            console.log(this.id, key)
+          })
+          return div
+        }
+
+        let inputKey = inputDiv('_mixlab_api_key', 'Key') 
+
+        widget.div.appendChild(inputKey) 
+
+        this.addCustomWidget(widget)
+
+        const onRemoved = this.onRemoved
+        this.onRemoved = () => { 
+          inputKey.remove()
+          widget.div.remove()
+          return onRemoved?.()
+        }
+
+        this.serialize_widgets = true //需要保存参数
+      }
+    }
+  },
+  async loadedGraphNode (node, app) {
+    // Fires every time a node is constructed
+    // You can modify widgets/add handlers/etc here
+
+    if (node.type === 'SiliconflowLLM') {
+      let widget = node.widgets.filter(w => w.div)[0]
+
+      let apiKey = getLocalData('_mixlab_api_key');
+
+      let id = node.id
+ 
+      widget.div.querySelector('.Key').value = apiKey[id] || 'by Mixlab' 
+    }
+  }
 })
 
 app.registerExtension({
@@ -209,13 +323,16 @@ app.registerExtension({
         text = text.filter(t => t && t?.trim())
 
         if (this.widgets) {
+          // console.log('#ShowTextForGPT',this.widgets)
           // const pos = this.widgets.findIndex(w => w.name === 'text')
           for (let i = 0; i < this.widgets.length; i++) {
-            if (this.widgets[i].name == 'show_text') this.widgets[i].onRemove?.()
+            if (this.widgets[i].name == 'show_text')
+              this.widgets[i].onRemove?.()
+            console.log('#ShowTextForGPT', this.widgets[i])
           }
-          this.widgets.length = 1
+          this.widgets.length = 2
         }
-        // console.log('ShowTextForGPT',text)
+
         for (let list of text) {
           if (list) {
             // console.log('#####', list)
@@ -228,6 +345,8 @@ app.registerExtension({
             w.inputEl.readOnly = true
             w.inputEl.style.opacity = 0.6
 
+            // w.inputEl.style.display='none'
+
             try {
               if (typeof list != 'string') {
                 let data = JSON.parse(list)
@@ -280,5 +399,24 @@ app.registerExtension({
 
       this.serialize_widgets = true //需要保存参数
     }
+  },
+  async loadedGraphNode (node, app) {
+    if (node.type === 'ShowTextForGPT') {
+      let widget = node.widgets.filter(w => w.name == 'show_text')[0]
+
+      // if (widget.value) {
+      //   let [url, prompt] = widget.value
+
+      //   this[`wavesurfer_${node.id}`] = updateWaveWidgetValue(
+      //     node.widgets,
+      //     node.id,
+      //     url,
+      //     prompt,
+      //     this[`wavesurfer_${node.id}`]
+      //   )
+      // }
+
+      console.log('#loadedGraphNode', node)
+    }
   }
 })

web/javascript/main_mixlab.js

@@ -1267,7 +1267,7 @@ app.registerExtension({
         })
 
         widget.PictureInPicture = $el('button', {
-          innerText: 'PictureInPicture',
+          innerText: 'Picture In Picture',
           style: {
             display: 'pictureInPictureEnabled' in document ? 'block' : 'none',
             cursor: 'pointer',

web/javascript/ui_mixlab.js

@@ -100,7 +100,7 @@ async function start_llama (model = 'Phi-3-mini-4k-instruct-Q5_K_S.gguf') {
     })
 
     const data = await response.json()
-    if (data.llama_cpp_error) {
+    if (data.llama_cpp_error||!data.port) {
       return
     }
 
@@ -1439,10 +1439,14 @@ app.registerExtension({
 
     LGraphCanvas.prototype.fixTheNode = function (node) {
       let new_node = LiteGraph.createNode(node.comfyClass)
-      new_node.pos = [node.pos[0], node.pos[1]]
-      app.canvas.graph.add(new_node, false)
-      copyNodeValues(node, new_node)
-      app.canvas.graph.remove(node)
+      console.log(node)
+      if(new_node){
+        new_node.pos = [node.pos[0], node.pos[1]]
+        app.canvas.graph.add(new_node, false)
+        copyNodeValues(node, new_node)
+        app.canvas.graph.remove(node)
+      }
+     
     }
 
     smart_init()

web/javascript/video_mixlab.js

@@ -468,6 +468,8 @@ app.registerExtension({
         const prefix = 'vhs_gif_preview_'
         const r = onExecuted ? onExecuted.apply(this, message) : undefined
 
+        if(!this.widgets) this.widgets=[]
+
         if (this.widgets) {
           const pos = this.widgets.findIndex(w => w.name === `${prefix}_0`)
           if (pos !== -1) {
@@ -488,6 +490,7 @@ app.registerExtension({
                   params.format || 'image/gif'
                 )
               )
+              console.log(w)
               w.parent = this
             })
           }