Wan2.1 SCAIL
基於參考圖構圖的單觸SCAIL姿態控制
下載工作流程在 Cloud 中試用 API 文件
00:00
工作流程內部
來源範本的唯讀預覽。請在 Cloud 中開啟,以探索節點並進行變更。
從您的程式碼呼叫此模型
以下請求使用與 Playground 相同的已準備節點圖和設定。每次提交都可能啟動付費執行;若提交結果不明,請勿自動重試提交。
1 取得 API 金鑰 ↗
What you need
- Runs on
- Comfy Cloud
- 金鑰
- COMFY_API_KEY
- 您的檔案
- 由程式碼透過其 URL 上傳
需要付費 Cloud 方案與點數。
下載 API JSONAPI 文件 ↗
# Python 3.10+: pip install comfy-sdk==0.4.0
import os
from comfy_sdk import Comfy
api_key = os.environ.get("COMFY_API_KEY", "").strip()
if not api_key: raise RuntimeError("Set COMFY_API_KEY")
os.environ.setdefault("COMFY_BASE_URL", "https://cloud.comfy.org")
client = Comfy(api_key=api_key)
workflow = client.workflows.from_json({
"390": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_FILENAME"
},
"_meta": {
"title": "Load Image: Reference"
}
},
"399": {
"class_type": "SaveVideo",
"inputs": {
"filename_prefix": "video/ComfyUI",
"format": "auto",
"format.codec": "auto",
"video": [
"385:395",
0
]
},
"_meta": {
"title": "Save Video"
}
},
"408": {
"class_type": "LoadVideo",
"inputs": {
"file": "UPLOADED_video_FILENAME"
},
"_meta": {
"title": "Load Video"
}
},
"385:107": {
"class_type": "ImageResizeKJv2",
"inputs": {
"crop_position": "center",
"device": "cpu",
"divisible_by": 32,
"height": 480,
"image": [
"390",
0
],
"keep_proportion": "crop",
"pad_color": "255,255,255",
"upscale_method": "lanczos",
"width": 832
},
"_meta": {
"title": "Resize Image v2"
}
},
"385:131": {
"class_type": "ImageResizeKJv2",
"inputs": {
"crop_position": "center",
"device": "cpu",
"divisible_by": 32,
"height": 480,
"image": [
"385:406",
0
],
"keep_proportion": "crop",
"pad_color": "0, 0, 0",
"upscale_method": "lanczos",
"width": 832
},
"_meta": {
"title": "Resize Image v2"
}
},
"385:22": {
"class_type": "WanVideoModelLoader",
"inputs": {
"attention_mode": "sdpa",
"base_precision": "fp16_fast",
"load_device": "offload_device",
"model": "Wan21-14B-SCAIL-preview_comfy_bf16.safetensors",
"quantization": "disabled",
"rms_norm_function": "default"
},
"_meta": {
"title": "WanVideo Model Loader"
}
},
"385:28": {
"class_type": "WanVideoDecode",
"inputs": {
"enable_vae_tiling": False,
"normalization": "default",
"samples": [
"385:348",
0
],
"tile_stride_x": 144,
"tile_stride_y": 128,
"tile_x": 272,
"tile_y": 272,
"vae": [
"385:38",
0
]
},
"_meta": {
"title": "WanVideo Decode"
}
},
"385:324": {
"class_type": "WanVideoAddSCAILPoseEmbeds",
"inputs": {
"embeds": [
"385:350",
0
],
"end_percent": 0.5,
"pose_images": [
"385:362",
0
],
"start_percent": 0,
"strength": 1,
"vae": [
"385:38",
0
]
},
"_meta": {
"title": "WanVideo Add SCAIL Pose Embeds"
}
},
"385:326": {
"class_type": "CLIPVisionLoader",
"inputs": {
"clip_name": "clip_vision_h.safetensors"
},
"_meta": {
"title": "Load CLIP Vision"
}
},
"385:327": {
"class_type": "WanVideoClipVisionEncode",
"inputs": {
"clip_vision": [
"385:326",
0
],
"combine_embeds": "average",
"crop": "center",
"force_offload": True,
"image_1": [
"385:107",
0
],
"ratio": 0.5,
"strength_1": 1,
"strength_2": 1,
"tiles": 0
},
"_meta": {
"title": "WanVideo ClipVision Encode"
}
},
"385:334": {
"class_type": "NLFPredict",
"inputs": {
"images": [
"385:131",
0
],
"model": [
"385:384",
0
],
"per_batch": -1
},
"_meta": {
"title": "NLF Predict"
}
},
"385:348": {
"class_type": "WanVideoSamplerv2",
"inputs": {
"add_noise_to_samples": False,
"cfg": 1,
"extra_args": [
"385:354",
0
],
"force_offload": True,
"image_embeds": [
"385:324",
0
],
"model": [
"385:80",
0
],
"scheduler": [
"385:349",
0
],
"seed": 42,
"text_embeds": [
"385:368",
0
]
},
"_meta": {
"title": "WanVideo Sampler v2"
}
},
"385:349": {
"class_type": "WanVideoSchedulerv2",
"inputs": {
"end_step": -1,
"enhance_hf": False,
"scheduler": "dpm++_sde",
"shift": 7,
"start_step": 0,
"steps": 6
},
"_meta": {
"title": "WanVideo Scheduler v2"
}
},
"385:35": {
"class_type": "WanVideoTorchCompileSettings",
"inputs": {
"allow_unmerged_lora_compile": False,
"backend": "inductor",
"compile_transformer_blocks_only": True,
"dynamic": False,
"dynamo_cache_size_limit": 64,
"dynamo_recompile_limit": 128,
"force_parameter_static_shapes": False,
"fullgraph": False,
"mode": "default"
},
"_meta": {
"title": "WanVideo Torch Compile Settings"
}
},
"385:350": {
"class_type": "WanVideoAddSCAILReferenceEmbeds",
"inputs": {
"clip_embeds": [
"385:327",
0
],
"embeds": [
"385:99",
0
],
"end_percent": 1,
"ref_image": [
"385:107",
0
],
"start_percent": 0,
"strength": 1,
"vae": [
"385:38",
0
]
},
"_meta": {
"title": "WanVideo Add SCAIL Reference Embeds"
}
},
"385:354": {
"class_type": "WanVideoSamplerExtraArgs",
"inputs": {
"context_options": [
"385:355",
0
],
"riflex_freq_index": 0,
"rope_function": "comfy"
},
"_meta": {
"title": "WanVideoSampler v2 Extra Args"
}
},
"385:355": {
"class_type": "WanVideoContextOptions",
"inputs": {
"context_frames": 81,
"context_overlap": 48,
"context_schedule": "uniform_standard",
"context_stride": 4,
"freenoise": True,
"fuse_method": "linear",
"verbose": False
},
"_meta": {
"title": "WanVideo Context Options"
}
},
"385:362": {
"class_type": "RenderNLFPoses",
"inputs": {
"draw_face": True,
"draw_hands": True,
"dw_poses": [
"385:363",
0
],
"height": [
"385:383",
0
],
"nlf_poses": [
"385:334",
0
],
"ref_dw_pose": [
"385:365",
0
],
"render_backend": "taichi",
"render_device": "gpu",
"scale_hands": True,
"width": [
"385:382",
0
]
},
"_meta": {
"title": "Render NLF Poses"
}
},
"385:363": {
"class_type": "PoseDetectionVitPoseToDWPose",
"inputs": {
"images": [
"385:131",
0
],
"vitpose_model": [
"385:364",
0
]
},
"_meta": {
"title": "Pose Detection VitPose to DWPose"
}
},
"385:364": {
"class_type": "OnnxDetectionModelLoader",
"inputs": {
"onnx_device": "CUDAExecutionProvider",
"vitpose_model": "vitpose-l-wholebody.onnx",
"yolo_model": "yolov10m.onnx"
},
"_meta": {
"title": "ONNX Detection Model Loader"
}
},
"385:365": {
"class_type": "PoseDetectionVitPoseToDWPose",
"inputs": {
"images": [
"385:107",
0
],
"vitpose_model": [
"385:364",
0
]
},
"_meta": {
"title": "Pose Detection VitPose to DWPose"
}
},
"385:368": {
"class_type": "WanVideoTextEncodeCached",
"inputs": {
"device": "gpu",
"model_name": "umt5_xxl_fp16.safetensors",
"negative_prompt": "色调艳丽,过曝,静态,细节模糊不清,字幕,风格,作品,画作,画面,静止,整体发灰,最差质量,低质量,JPEG压缩残留,丑陋的,残缺的,多余的手指,画得不好的手部,画得不好的脸部,畸形的,毁容的,形态畸形的肢体,手指融合,静止不动的画面,杂乱的背景,三条腿,背景人很多,倒着走",
"positive_prompt": "The young star-mapper in saffron and indigo robes dances alone on the shining salt flat at twilight, following the dancer's exact moves: she lands lightly from a leap, swivels her heels and kicks, spins with her arms flung wide so her layered robes and fringed sash flare out, hops to the side with a playful shoulder shimmy, and finishes with one arm raised toward the two pale moons. Her braided coronets swing with each turn and her brass cuffs catch the lilac light. The camera holds still and the painterly storybook style, the purple mineral ridges and the soft lavender mist stay exactly as in the reference image.",
"precision": "bf16",
"quantization": "disabled",
"use_disk_cache": False
},
"_meta": {
"title": "WanVideo TextEncode Cached"
}
},
"385:38": {
"class_type": "WanVideoVAELoader",
"inputs": {
"model_name": "wan_2.1_vae.safetensors",
"precision": "bf16",
"use_cpu_cache": False,
"verbose": False
},
"_meta": {
"title": "WanVideo VAE Loader"
}
},
"385:382": {
"class_type": "SimpleMath+",
"inputs": {
"a": 832,
"b": 0,
"value": "a / 2"
},
"_meta": {
"title": "🔧 Simple Math"
}
},
"385:383": {
"class_type": "SimpleMath+",
"inputs": {
"a": 480,
"b": 0,
"value": "a / 2"
},
"_meta": {
"title": "🔧 Simple Math"
}
},
"385:384": {
"class_type": "LoadNLFModel",
"inputs": {
"nlf_model": "nlf_l_multi_0.3.2.torchscript",
"warmup": True
},
"_meta": {
"title": "Load NLF Model"
}
},
"385:395": {
"class_type": "CreateVideo",
"inputs": {
"audio": [
"385:405",
1
],
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": 24,
"images": [
"385:28",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"385:405": {
"class_type": "GetVideoComponents",
"inputs": {
"video": [
"408",
0
]
},
"_meta": {
"title": "Get Video Components"
}
},
"385:406": {
"class_type": "ImageFromBatch",
"inputs": {
"batch_index": 0,
"image": [
"385:405",
0
],
"length": 121
},
"_meta": {
"title": "Get Image from Batch"
}
},
"385:56": {
"class_type": "WanVideoLoraSelect",
"inputs": {
"lora": "lightx2v_T2V_14B_cfg_step_distill_v2_lora_rank64_bf16.safetensors",
"low_mem_load": False,
"merge_loras": False,
"strength": 1
},
"_meta": {
"title": "WanVideo Lora Select"
}
},
"385:80": {
"class_type": "WanVideoSetLoRAs",
"inputs": {
"lora": [
"385:56",
0
],
"model": [
"385:22",
0
]
},
"_meta": {
"title": "WanVideo Set LoRAs"
}
},
"385:99": {
"class_type": "WanVideoEmptyEmbeds",
"inputs": {
"height": [
"385:107",
2
],
"num_frames": 121,
"width": [
"385:107",
1
]
},
"_meta": {
"title": "WanVideo Empty Embeds"
}
}
})
workflow.set_input("390", "image", client.assets.from_url("https://comfy.org/catalog-media/74255c19-ca65-4b6f-b045-a5c7058292ff/4b4d7584-ec11-463f-a645-910ff0860587"))
workflow.set_input("408", "file", client.assets.from_url("https://comfy.org/catalog-media/e2a9954d-e846-4452-b43f-facd4f913c68/891e0580-1de5-4870-a8a7-31922ced1730"))
# run() also passes api_key to the partner nodes in this workflow.
job = client.run(workflow, api_key=api_key)
for output in job.get_outputs("399"):
output.to_file(output.name)試試範例
關於此工作流程
ComfyUI 中的 Wan2.1 SCAIL 工作流程用於把影片來源的動作與姿勢資訊轉移到參考圖片上,同時維持圖片原本的構圖。此流程使用 Wan2.1 SCAIL 模型,擅長將姿勢資料調整為符合參考圖片的比例與取景。工作流程使用多個自訂節點,包括 ComfyUI-WanAnimatePreprocess 與 ComfyUI-WanVideoWrapper,用於高效率地預處理並封裝影片資料。這些節點與 comfyui-scail-pose 節點協作,負責精準縮放與對齊姿勢資料。透過這些工具,你可以建立保留原始參考圖美術風格的動態動畫。
此工作流程特別適合需要高保真姿勢轉移的應用,例如把靜態圖做成動畫片段,或讓影片內容在風格上保持一致。整合 LoadImage、SaveVideo 與 LoadVideo 等節點後,從輸入到輸出流程順暢,讓你能輕鬆匯入媒體、套用 SCAIL 技術並匯出最終動畫影片。對於複雜的姿勢控制任務,這提供了一個操作簡單的一鍵式方案。
問答
動作明確、姿態清晰的影片最適合,流程需精確擷取與轉移姿態資料。
可以,LoadImage 節點支援多種圖像格式,方便選擇參考圖像。
可以,comfyui-scail-pose 節點可調整姿態縮放與對齊強度。
請使用高解析度的來源影片與參考圖像,並設定 SaveVideo 節點輸出高品質格式。