[feature] support split server (#50)

* add load_transformer methods for split server

* add service utils

* [feature] support split servers

configs/deploy/hunyuan_i2v.json

+{
+    "infer_steps": 20,
+    "target_video_length": 33,
+    "i2v_resolution": "720p",
+    "attention_type": "flash_attn3",
+    "seed": 0,
+    "sub_servers": {
+        "prompt_enhancer": ["http://localhost:9001"],
+        "text_encoders": ["http://localhost:9002"],
+        "vae_model": ["http://localhost:9004"]
+    }
+}

configs/deploy/hunyuan_t2v.json

+{
+    "infer_steps": 20,
+    "target_video_length": 33,
+    "target_height": 720,
+    "target_width": 1280,
+    "attention_type": "flash_attn3",
+    "seed": 42,
+    "sub_servers": {
+        "prompt_enhancer": ["http://localhost:9001"],
+        "text_encoders": ["http://localhost:9002"],
+        "vae_model": ["http://localhost:9004"]
+    }
+}

configs/deploy/wan_i2v.json

+{
+    "infer_steps": 40,
+    "target_video_length": 81,
+    "target_height": 480,
+    "target_width": 832,
+    "attention_type": "flash_attn3",
+    "seed": 42,
+    "sample_guide_scale": 5,
+    "sample_shift": 5,
+    "enable_cfg": true,
+    "cpu_offload": false,
+    "sub_servers": {
+        "prompt_enhancer": ["http://localhost:9001"],
+        "text_encoders": ["http://localhost:9002"],
+        "image_encoder": ["http://localhost:9003"],
+        "vae_model": ["http://localhost:9004"]
+    }
+}

configs/deploy/wan_t2v.json

+{
+    "infer_steps": 50,
+    "target_video_length": 81,
+    "text_len": 512,
+    "target_height": 480,
+    "target_width": 832,
+    "attention_type": "flash_attn3",
+    "seed": 42,
+    "sample_guide_scale": 6,
+    "sample_shift": 8,
+    "enable_cfg": true,
+    "cpu_offload": false,
+    "sub_servers": {
+        "prompt_enhancer": ["http://localhost:9001"],
+        "text_encoders": ["http://localhost:9002"],
+        "image_encoder": ["http://localhost:9003"],
+        "vae_model": ["http://localhost:9004"]
+    }
+}

lightx2v/api_server.py

-import signal
-import sys
-import psutil
+import asyncio
 import argparse
 from fastapi import FastAPI
 from pydantic import BaseModel
@@ -17,32 +15,7 @@ import torch
 from lightx2v.utils.profiler import ProfilingContext
 from lightx2v.utils.set_config import set_config
 from lightx2v.infer import init_runner
-
-
-# =========================
-# Signal & Process Control
-# =========================
-
-
-def kill_all_related_processes():
-    """Kill the current process and all its child processes"""
-    current_process = psutil.Process()
-    children = current_process.children(recursive=True)
-    for child in children:
-        try:
-            child.kill()
-        except Exception as e:
-            logger.info(f"Failed to kill child process {child.pid}: {e}")
-    try:
-        current_process.kill()
-    except Exception as e:
-        logger.info(f"Failed to kill main process: {e}")
-
-
-def signal_handler(sig, frame):
-    logger.info("\nReceived Ctrl+C, shutting down all related processes...")
-    kill_all_related_processes()
-    sys.exit(0)
+from lightx2v.utils.service_utils import TaskStatusMessage, BaseServiceStatus, ProcessManager
 
 
 # =========================
@@ -70,67 +43,8 @@ class Message(BaseModel):
         return getattr(self, key, default)
 
 
-class TaskStatusMessage(BaseModel):
-    task_id: str
-
-
-class ServiceStatus:
-    _lock = threading.Lock()
-    _current_task = None
-    _result_store = {}
-
-    @classmethod
-    def start_task(cls, message: Message):
-        with cls._lock:
-            if cls._current_task is not None:
-                raise RuntimeError("Service busy")
-            if message.task_id_must_unique and message.task_id in cls._result_store:
-                raise RuntimeError(f"Task ID {message.task_id} already exists")
-            cls._current_task = {"message": message, "start_time": datetime.now()}
-            return message.task_id
-
-    @classmethod
-    def complete_task(cls, message: Message):
-        with cls._lock:
-            cls._result_store[message.task_id] = {"success": True, "message": message, "start_time": cls._current_task["start_time"], "completion_time": datetime.now()}
-            cls._current_task = None
-
-    @classmethod
-    def record_failed_task(cls, message: Message, error: Optional[str] = None):
-        """Record a failed task with an error message."""
-        with cls._lock:
-            cls._result_store[message.task_id] = {"success": False, "message": message, "start_time": cls._current_task["start_time"], "error": error}
-            cls._current_task = None
-
-    @classmethod
-    def clean_stopped_task(cls):
-        with cls._lock:
-            if cls._current_task:
-                message = cls._current_task["message"]
-                error = "Task stopped by user"
-                cls._result_store[message.task_id] = {"success": False, "message": message, "start_time": cls._current_task["start_time"], "error": error}
-                cls._current_task = None
-
-    @classmethod
-    def get_status_task_id(cls, task_id: str):
-        with cls._lock:
-            if cls._current_task and cls._current_task["message"].task_id == task_id:
-                return {"task_status": "processing"}
-            if task_id in cls._result_store:
-                return {"task_status": "completed", **cls._result_store[task_id]}
-            return {"task_status": "not_found"}
-
-    @classmethod
-    def get_status_service(cls):
-        with cls._lock:
-            if cls._current_task:
-                return {"service_status": "busy", "task_id": cls._current_task["message"].task_id}
-            return {"service_status": "idle"}
-
-    @classmethod
-    def get_all_tasks(cls):
-        with cls._lock:
-            return cls._result_store
+class ApiServerServiceStatus(BaseServiceStatus):
+    pass
 
 
 def local_video_generate(message: Message):
@@ -138,17 +52,22 @@ def local_video_generate(message: Message):
         global runner
         runner.set_inputs(message)
         logger.info(f"message: {message}")
-        runner.run_pipeline()
-        ServiceStatus.complete_task(message)
+        loop = asyncio.new_event_loop()
+        asyncio.set_event_loop(loop)
+        try:
+            loop.run_until_complete(runner.run_pipeline())
+        finally:
+            loop.close()
+        ApiServerServiceStatus.complete_task(message)
     except Exception as e:
         logger.error(f"task_id {message.task_id} failed: {str(e)}")
-        ServiceStatus.record_failed_task(message, error=str(e))
+        ApiServerServiceStatus.record_failed_task(message, error=str(e))
 
 
 @app.post("/v1/local/video/generate")
 async def v1_local_video_generate(message: Message):
     try:
-        task_id = ServiceStatus.start_task(message)
+        task_id = ApiServerServiceStatus.start_task(message)
         # Use background threads to perform long-running tasks
         global thread
         thread = threading.Thread(target=local_video_generate, args=(message,), daemon=True)
@@ -160,17 +79,17 @@ async def v1_local_video_generate(message: Message):
 
 @app.get("/v1/local/video/generate/service_status")
 async def get_service_status():
-    return ServiceStatus.get_status_service()
+    return ApiServerServiceStatus.get_status_service()
 
 
 @app.get("/v1/local/video/generate/get_all_tasks")
 async def get_all_tasks():
-    return ServiceStatus.get_all_tasks()
+    return ApiServerServiceStatus.get_all_tasks()
 
 
 @app.post("/v1/local/video/generate/task_status")
 async def get_task_status(message: TaskStatusMessage):
-    return ServiceStatus.get_status_task_id(message.task_id)
+    return ApiServerServiceStatus.get_status_task_id(message.task_id)
 
 
 def _async_raise(tid, exctype):
@@ -193,7 +112,7 @@ async def stop_running_task():
 
             # Clean up the thread reference
             thread = None
-            ServiceStatus.clean_stopped_task()
+            ApiServerServiceStatus.clean_stopped_task()
             gc.collect()
             torch.cuda.empty_cache()
             return {"stop_status": "success", "reason": "Task stopped successfully."}
@@ -208,13 +127,13 @@ async def stop_running_task():
 # =========================
 
 if __name__ == "__main__":
-    signal.signal(signal.SIGINT, signal_handler)
+    ProcessManager.register_signal_handler()
     parser = argparse.ArgumentParser()
     parser.add_argument("--model_cls", type=str, required=True, choices=["wan2.1", "hunyuan", "wan2.1_causvid", "wan2.1_skyreels_v2_df"], default="hunyuan")
     parser.add_argument("--task", type=str, choices=["t2v", "i2v"], default="t2v")
     parser.add_argument("--model_path", type=str, required=True)
     parser.add_argument("--config_json", type=str, required=True)
-    parser.add_argument("--prompt_enhancer", default=None)
+    parser.add_argument("--split", action="store_true")
 
     parser.add_argument("--port", type=int, default=8000)
     args = parser.parse_args()
@@ -222,6 +141,7 @@ if __name__ == "__main__":
 
     with ProfilingContext("Init Server Cost"):
         config = set_config(args)
+        config["mode"] = "split_server" if args.split else "server"
         logger.info(f"config:\n{json.dumps(config, ensure_ascii=False, indent=4)}")
         runner = init_runner(config)
 

lightx2v/common/apis/image_encoder.py

+import argparse
+from fastapi import FastAPI
+from pydantic import BaseModel
+from loguru import logger
+import uvicorn
+import json
+import os
+import torch
+import torchvision.transforms.functional as TF
+
+from lightx2v.models.input_encoders.hf.xlm_roberta.model import CLIPModel
+
+from lightx2v.utils.profiler import ProfilingContext
+from lightx2v.utils.set_config import set_config
+from lightx2v.utils.service_utils import TaskStatusMessage, BaseServiceStatus, ProcessManager, TensorTransporter, ImageTransporter
+
+tensor_transporter = TensorTransporter()
+image_transporter = ImageTransporter()
+
+# =========================
+# FastAPI Related Code
+# =========================
+
+runner = None
+
+app = FastAPI()
+
+
+class Message(BaseModel):
+    task_id: str
+    task_id_must_unique: bool = False
+
+    img: bytes
+
+    def get(self, key, default=None):
+        return getattr(self, key, default)
+
+
+class ImageEncoderServiceStatus(BaseServiceStatus):
+    pass
+
+
+class ImageEncoderRunner:
+    def __init__(self, config):
+        self.config = config
+        self.image_encoder = self.get_image_encoder_model()
+
+    def get_image_encoder_model(self):
+        if "wan2.1" in self.config.model_cls:
+            image_encoder = CLIPModel(
+                dtype=torch.float16,
+                device="cuda",
+                checkpoint_path=os.path.join(
+                    self.config.model_path,
+                    "models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth",
+                ),
+                tokenizer_path=os.path.join(self.config.model_path, "xlm-roberta-large"),
+            )
+        else:
+            raise ValueError(f"Unsupported model class: {self.config.model_cls}")
+        return image_encoder
+
+    def _run_image_encoder(self, img):
+        if "wan2.1" in self.config.model_cls:
+            img = image_transporter.load_image(img)
+            img = TF.to_tensor(img).sub_(0.5).div_(0.5).cuda()
+            clip_encoder_out = self.image_encoder.visual([img[:, None, :, :]], self.config).squeeze(0).to(torch.bfloat16)
+        else:
+            raise ValueError(f"Unsupported model class: {self.config.model_cls}")
+        return clip_encoder_out
+
+
+def run_image_encoder(message: Message):
+    try:
+        global runner
+        image_encoder_out = runner._run_image_encoder(message.img)
+        assert image_encoder_out is not None
+        ImageEncoderServiceStatus.complete_task(message)
+        return image_encoder_out
+    except Exception as e:
+        logger.error(f"task_id {message.task_id} failed: {str(e)}")
+        ImageEncoderServiceStatus.record_failed_task(message, error=str(e))
+
+
+@app.post("/v1/local/image_encoder/generate")
+def v1_local_image_encoder_generate(message: Message):
+    try:
+        task_id = ImageEncoderServiceStatus.start_task(message)
+        image_encoder_output = run_image_encoder(message)
+        output = tensor_transporter.prepare_tensor(image_encoder_output)
+        del image_encoder_output
+        return {"task_id": task_id, "task_status": "completed", "output": output, "kwargs": None}
+    except RuntimeError as e:
+        return {"error": str(e)}
+
+
+@app.get("/v1/local/image_encoder/generate/service_status")
+async def get_service_status():
+    return ImageEncoderServiceStatus.get_status_service()
+
+
+@app.get("/v1/local/image_encoder/generate/get_all_tasks")
+async def get_all_tasks():
+    return ImageEncoderServiceStatus.get_all_tasks()
+
+
+@app.post("/v1/local/image_encoder/generate/task_status")
+async def get_task_status(message: TaskStatusMessage):
+    return ImageEncoderServiceStatus.get_status_task_id(message.task_id)
+
+
+# =========================
+# Main Entry
+# =========================
+
+if __name__ == "__main__":
+    ProcessManager.register_signal_handler()
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--model_cls", type=str, required=True, choices=["wan2.1", "hunyuan", "wan2.1_causvid", "wan2.1_skyreels_v2_df"], default="hunyuan")
+    parser.add_argument("--task", type=str, choices=["t2v", "i2v"], default="t2v")
+    parser.add_argument("--model_path", type=str, required=True)
+    parser.add_argument("--config_json", type=str, required=True)
+
+    parser.add_argument("--port", type=int, default=9003)
+    args = parser.parse_args()
+    logger.info(f"args: {args}")
+
+    assert args.task == "i2v"
+
+    with ProfilingContext("Init Server Cost"):
+        config = set_config(args)
+        config["mode"] = "split_server"
+        logger.info(f"config:\n{json.dumps(config, ensure_ascii=False, indent=4)}")
+        runner = ImageEncoderRunner(config)
+
+    uvicorn.run(app, host="0.0.0.0", port=config.port, reload=False, workers=1)

lightx2v/common/apis/text_encoder.py

+import argparse
+from typing import Optional
+from fastapi import FastAPI
+from pydantic import BaseModel
+from loguru import logger
+import uvicorn
+import json
+import os
+import torch
+
+from lightx2v.models.input_encoders.hf.t5.model import T5EncoderModel
+from lightx2v.models.input_encoders.hf.llama.model import TextEncoderHFLlamaModel
+from lightx2v.models.input_encoders.hf.clip.model import TextEncoderHFClipModel
+from lightx2v.models.input_encoders.hf.llava.model import TextEncoderHFLlavaModel
+
+from lightx2v.utils.profiler import ProfilingContext
+from lightx2v.utils.set_config import set_config
+from lightx2v.utils.service_utils import TaskStatusMessage, BaseServiceStatus, ProcessManager, TensorTransporter, ImageTransporter
+
+tensor_transporter = TensorTransporter()
+image_transporter = ImageTransporter()
+
+# =========================
+# FastAPI Related Code
+# =========================
+
+runner = None
+
+app = FastAPI()
+
+
+class Message(BaseModel):
+    task_id: str
+    task_id_must_unique: bool = False
+
+    text: str
+    img: Optional[bytes] = None
+
+    def get(self, key, default=None):
+        return getattr(self, key, default)
+
+
+class TextEncoderServiceStatus(BaseServiceStatus):
+    pass
+
+
+class TextEncoderRunner:
+    def __init__(self, config):
+        self.config = config
+        self.text_encoders = self.get_text_encoder_model()
+
+    def get_text_encoder_model(self):
+        if "wan2.1" in self.config.model_cls:
+            text_encoder = T5EncoderModel(
+                text_len=self.config["text_len"],
+                dtype=torch.bfloat16,
+                device="cuda",
+                checkpoint_path=os.path.join(self.config.model_path, "models_t5_umt5-xxl-enc-bf16.pth"),
+                tokenizer_path=os.path.join(self.config.model_path, "google/umt5-xxl"),
+                shard_fn=None,
+            )
+            text_encoders = [text_encoder]
+        elif self.config.model_cls in ["hunyuan"]:
+            if self.config.task == "t2v":
+                text_encoder_1 = TextEncoderHFLlamaModel(os.path.join(self.config.model_path, "text_encoder"), "cuda")
+            else:
+                text_encoder_1 = TextEncoderHFLlavaModel(os.path.join(self.config.model_path, "text_encoder_i2v"), "cuda")
+            text_encoder_2 = TextEncoderHFClipModel(os.path.join(self.config.model_path, "text_encoder_2"), "cuda")
+            text_encoders = [text_encoder_1, text_encoder_2]
+        else:
+            raise ValueError(f"Unsupported model class: {self.config.model_cls}")
+        return text_encoders
+
+    def _run_text_encoder(self, text, img):
+        if "wan2.1" in self.config.model_cls:
+            text_encoder_output = {}
+            n_prompt = self.config.get("negative_prompt", "")
+            context = self.text_encoders[0].infer([text], self.config)
+            context_null = self.text_encoders[0].infer([n_prompt if n_prompt else ""], self.config)
+            text_encoder_output["context"] = context
+            text_encoder_output["context_null"] = context_null
+        elif self.config.model_cls in ["hunyuan"]:
+            text_encoder_output = {}
+            for i, encoder in enumerate(self.text_encoders):
+                if self.config.task == "i2v" and i == 0:
+                    img = image_transporter.load_image(img)
+                    text_state, attention_mask = encoder.infer(text, img, self.config)
+                else:
+                    text_state, attention_mask = encoder.infer(text, self.config)
+                text_encoder_output[f"text_encoder_{i + 1}_text_states"] = text_state.to(dtype=torch.bfloat16)
+                text_encoder_output[f"text_encoder_{i + 1}_attention_mask"] = attention_mask
+        else:
+            raise ValueError(f"Unsupported model class: {self.config.model_cls}")
+        return text_encoder_output
+
+
+def run_text_encoder(message: Message):
+    try:
+        global runner
+        text_encoder_output = runner._run_text_encoder(message.text, message.img)
+        TextEncoderServiceStatus.complete_task(message)
+        return text_encoder_output
+    except Exception as e:
+        logger.error(f"task_id {message.task_id} failed: {str(e)}")
+        TextEncoderServiceStatus.record_failed_task(message, error=str(e))
+
+
+@app.post("/v1/local/text_encoder/generate")
+def v1_local_text_encoder_generate(message: Message):
+    try:
+        task_id = TextEncoderServiceStatus.start_task(message)
+        text_encoder_output = run_text_encoder(message)
+        output = tensor_transporter.prepare_tensor(text_encoder_output)
+        del text_encoder_output
+        return {"task_id": task_id, "task_status": "completed", "output": output, "kwargs": None}
+    except RuntimeError as e:
+        return {"error": str(e)}
+
+
+@app.get("/v1/local/text_encoder/generate/service_status")
+async def get_service_status():
+    return TextEncoderServiceStatus.get_status_service()
+
+
+@app.get("/v1/local/text_encoder/generate/get_all_tasks")
+async def get_all_tasks():
+    return TextEncoderServiceStatus.get_all_tasks()
+
+
+@app.post("/v1/local/text_encoder/generate/task_status")
+async def get_task_status(message: TaskStatusMessage):
+    return TextEncoderServiceStatus.get_status_task_id(message.task_id)
+
+
+# =========================
+# Main Entry
+# =========================
+
+if __name__ == "__main__":
+    ProcessManager.register_signal_handler()
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--model_cls", type=str, required=True, choices=["wan2.1", "hunyuan", "wan2.1_causvid", "wan2.1_skyreels_v2_df"], default="hunyuan")
+    parser.add_argument("--task", type=str, choices=["t2v", "i2v"], default="t2v")
+    parser.add_argument("--model_path", type=str, required=True)
+    parser.add_argument("--config_json", type=str, required=True)
+
+    parser.add_argument("--port", type=int, default=9002)
+    args = parser.parse_args()
+    logger.info(f"args: {args}")
+
+    with ProfilingContext("Init Server Cost"):
+        config = set_config(args)
+        config["mode"] = "split_server"
+        logger.info(f"config:\n{json.dumps(config, ensure_ascii=False, indent=4)}")
+        runner = TextEncoderRunner(config)
+
+    uvicorn.run(app, host="0.0.0.0", port=config.port, reload=False, workers=1)

lightx2v/common/apis/vae.py

+import argparse
+from fastapi import FastAPI
+from pydantic import BaseModel
+from loguru import logger
+from typing import Optional
+import numpy as np
+import uvicorn
+import json
+import os
+import torch
+import torchvision
+import torchvision.transforms.functional as TF
+
+from lightx2v.models.video_encoders.hf.wan.vae import WanVAE
+from lightx2v.models.video_encoders.hf.autoencoder_kl_causal_3d.model import VideoEncoderKLCausal3DModel
+from lightx2v.models.runners.hunyuan.hunyuan_runner import HunyuanRunner
+
+from lightx2v.utils.profiler import ProfilingContext
+from lightx2v.utils.set_config import set_config
+from lightx2v.utils.service_utils import TaskStatusMessage, BaseServiceStatus, ProcessManager, TensorTransporter, ImageTransporter
+
+tensor_transporter = TensorTransporter()
+image_transporter = ImageTransporter()
+
+# =========================
+# FastAPI Related Code
+# =========================
+
+runner = None
+
+app = FastAPI()
+
+
+class Message(BaseModel):
+    task_id: str
+    task_id_must_unique: bool = False
+
+    img: Optional[bytes] = None
+    latents: Optional[bytes] = None
+
+    def get(self, key, default=None):
+        return getattr(self, key, default)
+
+
+class VAEServiceStatus(BaseServiceStatus):
+    pass
+
+
+class VAEEncoderRunner:
+    def __init__(self, config):
+        self.config = config
+        self.vae_model = self.get_vae_model()
+
+    def get_vae_model(self):
+        if "wan2.1" in self.config.model_cls:
+            vae_model = WanVAE(
+                vae_pth=os.path.join(self.config.model_path, "Wan2.1_VAE.pth"),
+                device="cuda",
+                parallel=self.config.parallel_vae,
+            )
+        elif self.config.model_cls in ["hunyuan"]:
+            vae_model = VideoEncoderKLCausal3DModel(model_path=self.config.model_path, dtype=torch.float16, device="cuda", config=self.config)
+        else:
+            raise ValueError(f"Unsupported model class: {self.config.model_cls}")
+        return vae_model
+
+    def _run_vae_encoder(self, img):
+        img = image_transporter.load_image(img)
+        kwargs = {}
+        if "wan2.1" in self.config.model_cls:
+            img = TF.to_tensor(img).sub_(0.5).div_(0.5).cuda()
+            h, w = img.shape[1:]
+            aspect_ratio = h / w
+            max_area = self.config.target_height * self.config.target_width
+            lat_h = round(np.sqrt(max_area * aspect_ratio) // self.config.vae_stride[1] // self.config.patch_size[1] * self.config.patch_size[1])
+            lat_w = round(np.sqrt(max_area / aspect_ratio) // self.config.vae_stride[2] // self.config.patch_size[2] * self.config.patch_size[2])
+            h = lat_h * self.config.vae_stride[1]
+            w = lat_w * self.config.vae_stride[2]
+
+            self.config.lat_h, kwargs["lat_h"] = lat_h, lat_h
+            self.config.lat_w, kwargs["lat_w"] = lat_w, lat_w
+
+            msk = torch.ones(1, self.config.target_video_length, lat_h, lat_w, device=torch.device("cuda"))
+            msk[:, 1:] = 0
+            msk = torch.concat([torch.repeat_interleave(msk[:, 0:1], repeats=4, dim=1), msk[:, 1:]], dim=1)
+            msk = msk.view(1, msk.shape[1] // 4, 4, lat_h, lat_w)
+            msk = msk.transpose(1, 2)[0]
+            vae_encode_out = self.vae_model.encode(
+                [
+                    torch.concat(
+                        [
+                            torch.nn.functional.interpolate(img[None].cpu(), size=(h, w), mode="bicubic").transpose(0, 1),
+                            torch.zeros(3, self.config.target_video_length - 1, h, w),
+                        ],
+                        dim=1,
+                    ).cuda()
+                ],
+                self.config,
+            )[0]
+            vae_encode_out = torch.concat([msk, vae_encode_out]).to(torch.bfloat16)
+        elif self.config.model_cls in ["hunyuan"]:
+            if self.config.i2v_resolution == "720p":
+                bucket_hw_base_size = 960
+            elif self.config.i2v_resolution == "540p":
+                bucket_hw_base_size = 720
+            elif self.config.i2v_resolution == "360p":
+                bucket_hw_base_size = 480
+            else:
+                raise ValueError(f"self.config.i2v_resolution: {self.config.i2v_resolution} must be in [360p, 540p, 720p]")
+
+            origin_size = img.size
+
+            crop_size_list = HunyuanRunner.generate_crop_size_list(bucket_hw_base_size, 32)
+            aspect_ratios = np.array([round(float(h) / float(w), 5) for h, w in crop_size_list])
+            closest_size, closest_ratio = HunyuanRunner.get_closest_ratio(origin_size[1], origin_size[0], aspect_ratios, crop_size_list)
+
+            self.config.target_height, self.config.target_width = closest_size
+            kwargs["target_height"], kwargs["target_width"] = closest_size
+
+            resize_param = min(closest_size)
+            center_crop_param = closest_size
+
+            ref_image_transform = torchvision.transforms.Compose(
+                [torchvision.transforms.Resize(resize_param), torchvision.transforms.CenterCrop(center_crop_param), torchvision.transforms.ToTensor(), torchvision.transforms.Normalize([0.5], [0.5])]
+            )
+
+            semantic_image_pixel_values = [ref_image_transform(img)]
+            semantic_image_pixel_values = torch.cat(semantic_image_pixel_values).unsqueeze(0).unsqueeze(2).to(torch.float16).to(torch.device("cuda"))
+
+            vae_encode_out = self.vae_model.encode(semantic_image_pixel_values, self.config).mode()
+
+            scaling_factor = 0.476986
+            vae_encode_out.mul_(scaling_factor)
+        else:
+            raise ValueError(f"Unsupported model class: {self.config.model_cls}")
+        return vae_encode_out, kwargs
+
+    def _run_vae_decoder(self, latents):
+        latents = tensor_transporter.load_tensor(latents)
+        images = self.vae_model.decode(latents, generator=None, config=self.config)
+        return images
+
+
+def run_vae_encoder(message: Message):
+    try:
+        global runner
+        vae_encode_out, kwargs = runner._run_vae_encoder(message.img)
+        VAEServiceStatus.complete_task(message)
+        return vae_encode_out, kwargs
+    except Exception as e:
+        logger.error(f"task_id {message.task_id} failed: {str(e)}")
+        VAEServiceStatus.record_failed_task(message, error=str(e))
+
+
+def run_vae_decoder(message: Message):
+    try:
+        global runner
+        images = runner._run_vae_decoder(message.latents)
+        VAEServiceStatus.complete_task(message)
+        return images
+    except Exception as e:
+        logger.error(f"task_id {message.task_id} failed: {str(e)}")
+        VAEServiceStatus.record_failed_task(message, error=str(e))
+
+
+@app.post("/v1/local/vae_model/encoder/generate")
+def v1_local_vae_model_encoder_generate(message: Message):
+    try:
+        task_id = VAEServiceStatus.start_task(message)
+        vae_encode_out, kwargs = run_vae_encoder(message)
+        output = tensor_transporter.prepare_tensor(vae_encode_out)
+        del vae_encode_out
+        return {"task_id": task_id, "task_status": "completed", "output": output, "kwargs": kwargs}
+    except RuntimeError as e:
+        return {"error": str(e)}
+
+
+@app.post("/v1/local/vae_model/decoder/generate")
+def v1_local_vae_model_decoder_generate(message: Message):
+    try:
+        task_id = VAEServiceStatus.start_task(message)
+        vae_decode_out = run_vae_decoder(message)
+        output = tensor_transporter.prepare_tensor(vae_decode_out)
+        del vae_decode_out
+        return {"task_id": task_id, "task_status": "completed", "output": output, "kwargs": None}
+    except RuntimeError as e:
+        return {"error": str(e)}
+
+
+@app.get("/v1/local/vae_model/encoder/generate/service_status")
+async def get_service_status():
+    return VAEServiceStatus.get_status_service()
+
+
+@app.get("/v1/local/vae_model/decoder/generate/service_status")
+async def get_service_status():
+    return VAEServiceStatus.get_status_service()
+
+
+@app.get("/v1/local/vae_model/encoder/generate/get_all_tasks")
+async def get_all_tasks():
+    return VAEServiceStatus.get_all_tasks()
+
+
+@app.get("/v1/local/vae_model/decoder/generate/get_all_tasks")
+async def get_all_tasks():
+    return VAEServiceStatus.get_all_tasks()
+
+
+@app.post("/v1/local/vae_model/encoder/generate/task_status")
+async def get_task_status(message: TaskStatusMessage):
+    return VAEServiceStatus.get_status_task_id(message.task_id)
+
+
+@app.post("/v1/local/vae_model/decoder/generate/task_status")
+async def get_task_status(message: TaskStatusMessage):
+    return VAEServiceStatus.get_status_task_id(message.task_id)
+
+
+# =========================
+# Main Entry
+# =========================
+
+if __name__ == "__main__":
+    ProcessManager.register_signal_handler()
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--model_cls", type=str, required=True, choices=["wan2.1", "hunyuan", "wan2.1_causvid", "wan2.1_skyreels_v2_df"], default="hunyuan")
+    parser.add_argument("--task", type=str, choices=["t2v", "i2v"], default="t2v")
+    parser.add_argument("--model_path", type=str, required=True)
+    parser.add_argument("--config_json", type=str, required=True)
+
+    parser.add_argument("--port", type=int, default=9004)
+    args = parser.parse_args()
+    logger.info(f"args: {args}")
+
+    with ProfilingContext("Init Server Cost"):
+        config = set_config(args)
+        config["mode"] = "split_server"
+        logger.info(f"config:\n{json.dumps(config, ensure_ascii=False, indent=4)}")
+        runner = VAEEncoderRunner(config)
+
+    uvicorn.run(app, host="0.0.0.0", port=config.port, reload=False, workers=1)

lightx2v/infer.py

@@ -50,6 +50,7 @@ if __name__ == "__main__":
 
     with ProfilingContext("Total Cost"):
         config = set_config(args)
+        config["mode"] = "infer"
         logger.info(f"config:\n{json.dumps(config, ensure_ascii=False, indent=4)}")
         runner = init_runner(config)
 

lightx2v/models/runners/default_runner.py

+import asyncio
 import gc
+import aiohttp
 import torch
 import torch.distributed as dist
+import torchvision.transforms.functional as TF
+from PIL import Image
 from lightx2v.utils.profiler import ProfilingContext4Debug, ProfilingContext
 from lightx2v.utils.utils import save_videos_grid, cache_video
-from lightx2v.utils.prompt_enhancer import PromptEnhancer
+from lightx2v.utils.generate_task_id import generate_task_id
 from lightx2v.utils.envs import *
+from lightx2v.utils.service_utils import TensorTransporter, ImageTransporter
 from loguru import logger
 
 
 class DefaultRunner:
     def __init__(self, config):
         self.config = config
-        if self.config.prompt_enhancer is not None and self.config.task == "t2v":
-            self.load_prompt_enhancer()
-        self.model, self.text_encoders, self.vae_model, self.image_encoder = self.load_model()
-
-    @ProfilingContext("Load prompt enhancer")
-    def load_prompt_enhancer(self):
-        gpu_count = torch.cuda.device_count()
-        if gpu_count == 1:
-            logger.info("Only one GPU, use prompt enhancer cpu offload")
-            raise NotImplementedError("prompt enhancer cpu offload is not supported.")
-        self.prompt_enhancer = PromptEnhancer(model_name=self.config.prompt_enhancer, device_map="cuda:1")
-        self.config["use_prompt_enhancer"] = True  # Set use_prompt_enhancer to True now. (Default is False)
+        # TODO: implement prompt enhancer
+        self.has_prompt_enhancer = False
+        # if self.config.prompt_enhancer is not None and self.config.task == "t2v":
+        #     self.config["use_prompt_enhancer"] = True
+        #     self.has_prompt_enhancer = True
+        if self.config["mode"] == "split_server":
+            self.model = self.load_transformer()
+            self.text_encoders, self.vae_model, self.image_encoder = None, None, None
+            self.tensor_transporter = TensorTransporter()
+            self.image_transporter = ImageTransporter()
+        else:
+            self.model, self.text_encoders, self.vae_model, self.image_encoder = self.load_model()
 
     def set_inputs(self, inputs):
         self.config["prompt"] = inputs.get("prompt", "")
-        self.config["use_prompt_enhancer"] = inputs.get("use_prompt_enhancer", False)  # Reset use_prompt_enhancer from clinet side.
+        if self.has_prompt_enhancer and self.config["mode"] != "infer":
+            self.config["use_prompt_enhancer"] = inputs.get("use_prompt_enhancer", False)  # Reset use_prompt_enhancer from clinet side.
         self.config["negative_prompt"] = inputs.get("negative_prompt", "")
         self.config["image_path"] = inputs.get("image_path", "")
         self.config["save_video_path"] = inputs.get("save_video_path", "")
 
-    def run_input_encoder(self):
+    async def post_encoders(self, prompt, img=None, i2v=False):
+        tasks = []
+        img_byte = self.image_transporter.prepare_image(img) if img is not None else None
+        if i2v:
+            if "wan2.1" in self.config["model_cls"]:
+                tasks.append(
+                    asyncio.create_task(
+                        self.post_task(task_type="image_encoder", urls=self.config["sub_servers"]["image_encoder"], message={"task_id": generate_task_id(), "img": img_byte}, device="cuda")
+                    )
+                )
+            tasks.append(
+                asyncio.create_task(
+                    self.post_task(task_type="vae_model/encoder", urls=self.config["sub_servers"]["vae_model"], message={"task_id": generate_task_id(), "img": img_byte}, device="cuda")
+                )
+            )
+        tasks.append(
+            asyncio.create_task(
+                self.post_task(task_type="text_encoder", urls=self.config["sub_servers"]["text_encoders"], message={"task_id": generate_task_id(), "text": prompt, "img": img_byte}, device="cuda")
+            )
+        )
+        results = await asyncio.gather(*tasks)
+        # clip_encoder, vae_encoder, text_encoder
+        if not i2v:
+            return None, None, results[0]
+        if "wan2.1" in self.config["model_cls"]:
+            return results[0], results[1], results[2]
+        else:
+            return None, results[0], results[1]
+
+    async def run_input_encoder(self):
         image_encoder_output = None
-        if self.config["task"] == "i2v":
-            with ProfilingContext("Run Img Encoder"):
-                image_encoder_output = self.run_image_encoder(self.config, self.image_encoder, self.vae_model)
-        with ProfilingContext("Run Text Encoder"):
-            prompt = self.config["prompt_enhanced"] if self.config["use_prompt_enhancer"] else self.config["prompt"]
-            text_encoder_output = self.run_text_encoder(prompt, self.text_encoders, self.config, image_encoder_output)
+        prompt = self.config["prompt_enhanced"] if self.config["use_prompt_enhancer"] else self.config["prompt"]
+        i2v = self.config["task"] == "i2v"
+        img = Image.open(self.config["image_path"]).convert("RGB") if i2v else None
+        with ProfilingContext("Run Encoders"):
+            if self.config["mode"] == "split_server":
+                clip_encoder_out, vae_encode_out, text_encoder_output = await self.post_encoders(prompt, img, i2v)
+                if i2v:
+                    if self.config["model_cls"] in ["hunyuan"]:
+                        image_encoder_output = {"img": img, "img_latents": vae_encode_out}
+                    elif "wan2.1" in self.config["model_cls"]:
+                        image_encoder_output = {"clip_encoder_out": clip_encoder_out, "vae_encode_out": vae_encode_out}
+                    else:
+                        raise ValueError(f"Unsupported model class: {self.config['model_cls']}")
+            else:
+                if i2v:
+                    image_encoder_output = self.run_image_encoder(self.config, self.image_encoder, self.vae_model)
+                text_encoder_output = self.run_text_encoder(prompt, self.text_encoders, self.config, image_encoder_output)
         self.set_target_shape()
         self.inputs = {"text_encoder_output": text_encoder_output, "image_encoder_output": image_encoder_output}
 
@@ -60,9 +106,9 @@ class DefaultRunner:
 
         return self.model.scheduler.latents, self.model.scheduler.generator
 
-    def run_step(self, step_index=0):
+    async def run_step(self, step_index=0):
         self.init_scheduler()
-        self.run_input_encoder()
+        await self.run_input_encoder()
         self.model.scheduler.prepare(self.inputs["image_encoder_output"])
         self.model.scheduler.step_pre(step_index=step_index)
         self.model.infer(self.inputs)
@@ -74,8 +120,16 @@ class DefaultRunner:
         torch.cuda.empty_cache()
 
     @ProfilingContext("Run VAE")
-    def run_vae(self, latents, generator):
-        images = self.vae_model.decode(latents, generator=generator, config=self.config)
+    async def run_vae(self, latents, generator):
+        if self.config["mode"] == "split_server":
+            images = await self.post_task(
+                task_type="vae_model/decoder",
+                urls=self.config["sub_servers"]["vae_model"],
+                message={"task_id": generate_task_id(), "latents": self.tensor_transporter.prepare_tensor(latents)},
+                device="cpu",
+            )
+        else:
+            images = self.vae_model.decode(latents, generator=generator, config=self.config)
         return images
 
     @ProfilingContext("Save video")
@@ -86,15 +140,30 @@ class DefaultRunner:
             else:
                 save_videos_grid(images, self.config.save_video_path, fps=self.config.get("fps", 24))
 
-    def run_pipeline(self):
+    async def post_task(self, task_type, urls, message, device="cuda"):
+        while True:
+            for url in urls:
+                async with aiohttp.ClientSession() as session:
+                    async with session.get(f"{url}/v1/local/{task_type}/generate/service_status") as response:
+                        status = await response.json()
+                    if status["service_status"] == "idle":
+                        async with session.post(f"{url}/v1/local/{task_type}/generate", json=message) as response:
+                            result = await response.json()
+                            if result["kwargs"] is not None:
+                                for k, v in result["kwargs"].items():
+                                    setattr(self.config, k, v)
+                            return self.tensor_transporter.load_tensor(result["output"], device)
+            await asyncio.sleep(0.1)
+
+    async def run_pipeline(self):
         if self.config["use_prompt_enhancer"]:
             self.config["prompt_enhanced"] = self.prompt_enhancer(self.config["prompt"])
         self.init_scheduler()
-        self.run_input_encoder()
+        await self.run_input_encoder()
         self.model.scheduler.prepare(self.inputs["image_encoder_output"])
         latents, generator = self.run()
         self.end_run()
-        images = self.run_vae(latents, generator)
+        images = await self.run_vae(latents, generator)
         self.save_video(images)
         del latents, generator, images
         gc.collect()

lightx2v/models/runners/hunyuan/hunyuan_runner.py

@@ -21,6 +21,13 @@ class HunyuanRunner(DefaultRunner):
     def __init__(self, config):
         super().__init__(config)
 
+    def load_transformer(self):
+        if self.config.cpu_offload:
+            init_device = torch.device("cpu")
+        else:
+            init_device = torch.device("cuda")
+        return HunyuanModel(self.config.model_path, self.config, init_device, self.config)
+
     @ProfilingContext("Load models")
     def load_model(self):
         if self.config["parallel_attn_type"]:
@@ -64,7 +71,8 @@ class HunyuanRunner(DefaultRunner):
             text_encoder_output[f"text_encoder_{i + 1}_attention_mask"] = attention_mask
         return text_encoder_output
 
-    def get_closest_ratio(self, height: float, width: float, ratios: list, buckets: list):
+    @staticmethod
+    def get_closest_ratio(height: float, width: float, ratios: list, buckets: list):
         aspect_ratio = float(height) / float(width)
         diff_ratios = ratios - aspect_ratio
 
@@ -79,7 +87,8 @@ class HunyuanRunner(DefaultRunner):
 
         return closest_size, closest_ratio
 
-    def generate_crop_size_list(self, base_size=256, patch_size=32, max_ratio=4.0):
+    @staticmethod
+    def generate_crop_size_list(base_size=256, patch_size=32, max_ratio=4.0):
         num_patches = round((base_size / patch_size) ** 2)
         assert max_ratio >= 1.0
         crop_size_list = []

lightx2v/models/runners/wan/wan_causvid_runner.py

@@ -29,6 +29,13 @@ class WanCausVidRunner(WanRunner):
         self.infer_blocks = self.model.config.num_blocks
         self.num_fragments = self.model.config.num_fragments
 
+    def load_transformer(self):
+        if self.config.cpu_offload:
+            init_device = torch.device("cpu")
+        else:
+            init_device = torch.device("cuda")
+        return WanCausVidModel(self.config.model_path, self.config, init_device)
+
     @ProfilingContext("Load models")
     def load_model(self):
         if self.config["parallel_attn_type"]:

lightx2v/models/runners/wan/wan_runner.py

@@ -25,6 +25,19 @@ class WanRunner(DefaultRunner):
     def __init__(self, config):
         super().__init__(config)
 
+    def load_transformer(self):
+        if self.config.cpu_offload:
+            init_device = torch.device("cpu")
+        else:
+            init_device = torch.device("cuda")
+        model = WanModel(self.config.model_path, self.config, init_device)
+        if self.config.lora_path:
+            lora_wrapper = WanLoraWrapper(model)
+            lora_name = lora_wrapper.load_lora(self.config.lora_path)
+            lora_wrapper.apply_lora(lora_name, self.config.strength_model)
+            logger.info(f"Loaded LoRA: {lora_name}")
+        return model
+
     @ProfilingContext("Load models")
     def load_model(self):
         if self.config["parallel_attn_type"]:

lightx2v/utils/service_utils.py

+import sys
+import psutil
+import signal
+import base64
+from PIL import Image
+from loguru import logger
+from typing import Optional, List
+from datetime import datetime
+from pydantic import BaseModel
+import threading
+import torch
+import io
+
+
+class ProcessManager:
+    @staticmethod
+    def kill_all_related_processes():
+        """Kill the current process and all its child processes"""
+        current_process = psutil.Process()
+        children = current_process.children(recursive=True)
+        for child in children:
+            try:
+                child.kill()
+            except Exception as e:
+                logger.info(f"Failed to kill child process {child.pid}: {e}")
+        try:
+            current_process.kill()
+        except Exception as e:
+            logger.info(f"Failed to kill main process: {e}")
+
+    @staticmethod
+    def signal_handler(sig, frame):
+        logger.info("\nReceived Ctrl+C, shutting down all related processes...")
+        ProcessManager.kill_all_related_processes()
+        sys.exit(0)
+
+    @staticmethod
+    def register_signal_handler():
+        """Register the signal handler for SIGINT"""
+        signal.signal(signal.SIGINT, ProcessManager.signal_handler)
+
+
+class TaskStatusMessage(BaseModel):
+    task_id: str
+
+
+class BaseServiceStatus:
+    _lock = threading.Lock()
+    _current_task = None
+    _result_store = {}
+
+    @classmethod
+    def start_task(cls, message):
+        with cls._lock:
+            if cls._current_task is not None:
+                raise RuntimeError("Service busy")
+            if message.task_id_must_unique and message.task_id in cls._result_store:
+                raise RuntimeError(f"Task ID {message.task_id} already exists")
+            cls._current_task = {"message": message, "start_time": datetime.now()}
+            return message.task_id
+
+    @classmethod
+    def complete_task(cls, message):
+        with cls._lock:
+            cls._result_store[message.task_id] = {"success": True, "message": message, "start_time": cls._current_task["start_time"], "completion_time": datetime.now()}
+            cls._current_task = None
+
+    @classmethod
+    def record_failed_task(cls, message, error: Optional[str] = None):
+        """Record a failed task with an error message."""
+        with cls._lock:
+            cls._result_store[message.task_id] = {"success": False, "message": message, "start_time": cls._current_task["start_time"], "error": error}
+            cls._current_task = None
+
+    @classmethod
+    def clean_stopped_task(cls):
+        with cls._lock:
+            if cls._current_task:
+                message = cls._current_task["message"]
+                error = "Task stopped by user"
+                cls._result_store[message.task_id] = {"success": False, "message": message, "start_time": cls._current_task["start_time"], "error": error}
+                cls._current_task = None
+
+    @classmethod
+    def get_status_task_id(cls, task_id: str):
+        with cls._lock:
+            if cls._current_task and cls._current_task["message"].task_id == task_id:
+                return {"task_status": "processing"}
+            if task_id in cls._result_store:
+                return {"task_status": "completed", **cls._result_store[task_id]}
+            return {"task_status": "not_found"}
+
+    @classmethod
+    def get_status_service(cls):
+        with cls._lock:
+            if cls._current_task:
+                return {"service_status": "busy", "task_id": cls._current_task["message"].task_id}
+            return {"service_status": "idle"}
+
+    @classmethod
+    def get_all_tasks(cls):
+        with cls._lock:
+            return cls._result_store
+
+
+class TensorTransporter:
+    def __init__(self):
+        self.buffer = io.BytesIO()
+
+    def to_device(self, data, device):
+        if isinstance(data, dict):
+            return {key: self.to_device(value, device) for key, value in data.items()}
+        elif isinstance(data, torch.Tensor):
+            return data.to(device)
+        else:
+            return data
+
+    def prepare_tensor(self, data: torch.Tensor) -> bytes:
+        self.buffer.seek(0)
+        self.buffer.truncate()
+        torch.save(self.to_device(data, "cpu"), self.buffer)
+        return base64.b64encode(self.buffer.getvalue()).decode("utf-8")
+
+    def load_tensor(self, tensor_base64: str, device="cuda") -> torch.Tensor:
+        tensor_bytes = base64.b64decode(tensor_base64)
+        with io.BytesIO(tensor_bytes) as buffer:
+            return self.to_device(torch.load(buffer), device)
+
+
+class ImageTransporter:
+    def __init__(self):
+        self.buffer = io.BytesIO()
+
+    def prepare_image(self, image: Image.Image) -> bytes:
+        self.buffer.seek(0)
+        self.buffer.truncate()
+        image.save(self.buffer, format="PNG")
+        return base64.b64encode(self.buffer.getvalue()).decode("utf-8")
+
+    def load_image(self, image_base64: bytes) -> Image.Image:
+        image_bytes = base64.b64decode(image_base64)
+        with io.BytesIO(image_bytes) as buffer:
+            return Image.open(buffer).convert("RGB")

scripts/deploy/start_image_encoder_server.sh

+#!/bin/bash
+
+# set path and first
+lightx2v_path=
+model_path=
+
+# check section
+if [ -z "${CUDA_VISIBLE_DEVICES}" ]; then
+    cuda_devices=0
+    echo "Warn: CUDA_VISIBLE_DEVICES is not set, using default value: ${cuda_devices}, change at shell script or set env variable."
+    export CUDA_VISIBLE_DEVICES=${cuda_devices}
+fi
+
+if [ -z "${lightx2v_path}" ]; then
+    echo "Error: lightx2v_path is not set. Please set this variable first."
+    exit 1
+fi
+
+if [ -z "${model_path}" ]; then
+    echo "Error: model_path is not set. Please set this variable first."
+    exit 1
+fi
+
+export TOKENIZERS_PARALLELISM=false
+
+export PYTHONPATH=${lightx2v_path}:$PYTHONPATH
+
+export ENABLE_PROFILING_DEBUG=true
+export ENABLE_GRAPH_MODE=false
+
+python -m lightx2v.common.apis.image_encoder \
+--model_cls wan2.1 \
+--task i2v \
+--model_path $model_path \
+--config_json ${lightx2v_path}/configs/deploy/wan_i2v.json \
+--port 9003

scripts/deploy/start_server.sh

+#!/bin/bash
+
+# set path and first
+lightx2v_path=
+model_path=
+
+# check section
+if [ -z "${CUDA_VISIBLE_DEVICES}" ]; then
+    cuda_devices=0
+    echo "Warn: CUDA_VISIBLE_DEVICES is not set, using default value: ${cuda_devices}, change at shell script or set env variable."
+    export CUDA_VISIBLE_DEVICES=${cuda_devices}
+fi
+
+if [ -z "${lightx2v_path}" ]; then
+    echo "Error: lightx2v_path is not set. Please set this variable first."
+    exit 1
+fi
+
+if [ -z "${model_path}" ]; then
+    echo "Error: model_path is not set. Please set this variable first."
+    exit 1
+fi
+
+export TOKENIZERS_PARALLELISM=false
+
+export PYTHONPATH=${lightx2v_path}:$PYTHONPATH
+
+export ENABLE_PROFILING_DEBUG=true
+export ENABLE_GRAPH_MODE=false
+
+python -m lightx2v.api_server \
+--model_cls wan2.1 \
+--task i2v \
+--split \
+--model_path $model_path \
+--config_json ${lightx2v_path}/configs/deploy/wan_i2v.json \
+--port 8000

scripts/deploy/start_text_encoder_server.sh

+#!/bin/bash
+
+# set path and first
+lightx2v_path=
+model_path=
+
+# check section
+if [ -z "${CUDA_VISIBLE_DEVICES}" ]; then
+    cuda_devices=0
+    echo "Warn: CUDA_VISIBLE_DEVICES is not set, using default value: ${cuda_devices}, change at shell script or set env variable."
+    export CUDA_VISIBLE_DEVICES=${cuda_devices}
+fi
+
+if [ -z "${lightx2v_path}" ]; then
+    echo "Error: lightx2v_path is not set. Please set this variable first."
+    exit 1
+fi
+
+if [ -z "${model_path}" ]; then
+    echo "Error: model_path is not set. Please set this variable first."
+    exit 1
+fi
+
+export TOKENIZERS_PARALLELISM=false
+
+export PYTHONPATH=${lightx2v_path}:$PYTHONPATH
+
+export ENABLE_PROFILING_DEBUG=true
+export ENABLE_GRAPH_MODE=false
+
+python -m lightx2v.common.apis.text_encoder \
+--model_cls wan2.1 \
+--task i2v \
+--model_path $model_path \
+--config_json ${lightx2v_path}/configs/deploy/wan_i2v.json \
+--port 9002

scripts/deploy/start_vae_server.sh

+#!/bin/bash
+
+# set path and first
+lightx2v_path=
+model_path=
+
+# check section
+if [ -z "${CUDA_VISIBLE_DEVICES}" ]; then
+    cuda_devices=0
+    echo "Warn: CUDA_VISIBLE_DEVICES is not set, using default value: ${cuda_devices}, change at shell script or set env variable."
+    export CUDA_VISIBLE_DEVICES=${cuda_devices}
+fi
+
+if [ -z "${lightx2v_path}" ]; then
+    echo "Error: lightx2v_path is not set. Please set this variable first."
+    exit 1
+fi
+
+if [ -z "${model_path}" ]; then
+    echo "Error: model_path is not set. Please set this variable first."
+    exit 1
+fi
+
+export TOKENIZERS_PARALLELISM=false
+
+export PYTHONPATH=${lightx2v_path}:$PYTHONPATH
+
+export ENABLE_PROFILING_DEBUG=true
+export ENABLE_GRAPH_MODE=false
+
+python -m lightx2v.common.apis.vae \
+--model_cls wan2.1 \
+--task i2v \
+--model_path $model_path \
+--config_json ${lightx2v_path}/configs/deploy/wan_i2v.json \
+--port 9004