Costura de Vídeo Multi-Keyframe
Cria um vídeo suave usando 6 quadros-chave. Preenche automaticamente o movimento entre quadros e costura os segmentos de forma contínua.
Dentro do workflow
Uma prévia somente leitura do modelo de origem. Abra-o no Cloud para explorar os nós e fazer alterações.
Loading the graph
Open full-size workflow previewArquivos neste workflow
- Saída
- Vídeo
- Autor
- ComfyUI
Chame este modelo pelo seu código
A solicitação abaixo usa o mesmo grafo e as mesmas configurações preparadas no Playground. Cada envio pode iniciar uma execução paga; não tente novamente automaticamente um envio cujo resultado seja desconhecido.
What you need
- Runs on
- Comfy Cloud
- Chave
- COMFY_API_KEY
- Seus arquivos
- Enviados pelo código a partir de suas URLs
Requer um plano pago do Cloud e créditos.
# 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({
"62": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_keyframe_1_FILENAME"
},
"_meta": {
"title": "Start Frame"
}
},
"122": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_keyframe_2_FILENAME"
},
"_meta": {
"title": "End Frame"
}
},
"124": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_keyframe_3_FILENAME"
},
"_meta": {
"title": "End Frame"
}
},
"126": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_keyframe_4_FILENAME"
},
"_meta": {
"title": "End Frame"
}
},
"128": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_keyframe_5_FILENAME"
},
"_meta": {
"title": "End Frame"
}
},
"130": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_keyframe_6_FILENAME"
},
"_meta": {
"title": "End Frame"
}
},
"133": {
"class_type": "ImageBatch",
"inputs": {
"image1": [
"140:8",
0
],
"image2": [
"141:8",
0
]
},
"_meta": {
"title": "Batch Images (DEPRECATED)"
}
},
"134": {
"class_type": "ImageBatch",
"inputs": {
"image1": [
"133",
0
],
"image2": [
"142:8",
0
]
},
"_meta": {
"title": "Batch Images (DEPRECATED)"
}
},
"135": {
"class_type": "ImageBatch",
"inputs": {
"image1": [
"134",
0
],
"image2": [
"143:8",
0
]
},
"_meta": {
"title": "Batch Images (DEPRECATED)"
}
},
"136": {
"class_type": "ImageBatch",
"inputs": {
"image1": [
"135",
0
],
"image2": [
"144:8",
0
]
},
"_meta": {
"title": "Batch Images (DEPRECATED)"
}
},
"145": {
"class_type": "CreateVideo",
"inputs": {
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": 24,
"images": [
"136",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"146": {
"class_type": "SaveVideo",
"inputs": {
"filename_prefix": "video/ComfyUI",
"format": "auto",
"format.codec": "auto",
"video": [
"145",
0
]
},
"_meta": {
"title": "Save Video"
}
},
"140:37": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"140:38": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "umt5_xxl_fp8_e4m3fn_scaled.safetensors",
"device": "default",
"type": "wan"
},
"_meta": {
"title": "Load CLIP"
}
},
"140:39": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "wan_2.1_vae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"140:54": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"140:91",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"140:55": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"140:92",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"140:56": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"140:57": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "enable",
"cfg": 1,
"end_at_step": 2,
"latent_image": [
"140:67",
2
],
"model": [
"140:54",
0
],
"negative": [
"140:67",
1
],
"noise_seed": 106238756028535,
"positive": [
"140:67",
0
],
"return_with_leftover_noise": "enable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 0,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"140:58": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "disable",
"cfg": 1,
"end_at_step": 10000,
"latent_image": [
"140:57",
0
],
"model": [
"140:55",
0
],
"negative": [
"140:67",
1
],
"noise_seed": 0,
"positive": [
"140:67",
0
],
"return_with_leftover_noise": "disable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 2,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"140:6": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"140:38",
0
],
"text": "A time-lapse from a locked-off camera on the same hilltop: the stormy indigo twilight lifts and dawn breaks over the lone hawthorn tree. Clouds race across the sky and thin out, mist rolls through the hollows and burns away, and pale pink blossom bursts open all along the tree's twisted branches. Fresh green shoots spread over the rolling hill and the light warms from cold violet to soft rose and gold. The tree, the hill and the horizon stay exactly in place; only the season, the sky and the light change, smoothly and continuously, as the seasons advance in a single fluid shot with no cuts."
},
"_meta": {
"title": "CLIP Text Encode (Positive Prompt)"
}
},
"140:60": {
"class_type": "CreateVideo",
"inputs": {
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": 16,
"images": [
"140:8",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"140:67": {
"class_type": "WanFirstLastFrameToVideo",
"inputs": {
"batch_size": 1,
"end_image": [
"122",
0
],
"height": 480,
"length": 25,
"negative": [
"140:7",
0
],
"positive": [
"140:6",
0
],
"start_image": [
"62",
0
],
"vae": [
"140:39",
0
],
"width": 832
},
"_meta": {
"title": "WanFirstLastFrameToVideo"
}
},
"140:7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"140:38",
0
],
"text": "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"
},
"_meta": {
"title": "CLIP Text Encode (Negative Prompt)"
}
},
"140:8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"140:58",
0
],
"vae": [
"140:39",
0
]
},
"_meta": {
"title": "VAE Decode"
}
},
"140:91": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors",
"model": [
"140:37",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"140:92": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors",
"model": [
"140:56",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"141:37": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"141:38": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "umt5_xxl_fp8_e4m3fn_scaled.safetensors",
"device": "default",
"type": "wan"
},
"_meta": {
"title": "Load CLIP"
}
},
"141:39": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "wan_2.1_vae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"141:54": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"141:91",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"141:55": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"141:92",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"141:56": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"141:57": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "enable",
"cfg": 1,
"end_at_step": 2,
"latent_image": [
"141:67",
2
],
"model": [
"141:54",
0
],
"negative": [
"141:67",
1
],
"noise_seed": 332092862784409,
"positive": [
"141:67",
0
],
"return_with_leftover_noise": "enable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 0,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"141:58": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "disable",
"cfg": 1,
"end_at_step": 10000,
"latent_image": [
"141:57",
0
],
"model": [
"141:55",
0
],
"negative": [
"141:67",
1
],
"noise_seed": 0,
"positive": [
"141:67",
0
],
"return_with_leftover_noise": "disable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 2,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"141:6": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"141:38",
0
],
"text": "A time-lapse from a locked-off camera on the same hilltop: spring turns into high summer around the lone hawthorn tree. The pink blossom falls away and drifts on the wind as dense dark-green leaves fill the crown. The grass on the hill grows tall and lush, wildflowers open in waves of white and yellow, and tall white clouds build and sail across a bright blue midday sky, their shadows sweeping over the slopes. The tree, the hill and the horizon stay exactly in place; only the season, the sky and the light change, smoothly and continuously, in a single fluid shot with no cuts."
},
"_meta": {
"title": "CLIP Text Encode (Positive Prompt)"
}
},
"141:60": {
"class_type": "CreateVideo",
"inputs": {
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": 16,
"images": [
"141:8",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"141:67": {
"class_type": "WanFirstLastFrameToVideo",
"inputs": {
"batch_size": 1,
"end_image": [
"124",
0
],
"height": 480,
"length": 25,
"negative": [
"141:7",
0
],
"positive": [
"141:6",
0
],
"start_image": [
"122",
0
],
"vae": [
"141:39",
0
],
"width": 832
},
"_meta": {
"title": "WanFirstLastFrameToVideo"
}
},
"141:7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"141:38",
0
],
"text": "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"
},
"_meta": {
"title": "CLIP Text Encode (Negative Prompt)"
}
},
"141:8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"141:58",
0
],
"vae": [
"141:39",
0
]
},
"_meta": {
"title": "VAE Decode"
}
},
"141:91": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors",
"model": [
"141:37",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"141:92": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors",
"model": [
"141:56",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"142:37": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"142:38": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "umt5_xxl_fp8_e4m3fn_scaled.safetensors",
"device": "default",
"type": "wan"
},
"_meta": {
"title": "Load CLIP"
}
},
"142:39": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "wan_2.1_vae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"142:54": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"142:91",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"142:55": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"142:92",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"142:56": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"142:57": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "enable",
"cfg": 1,
"end_at_step": 2,
"latent_image": [
"142:67",
2
],
"model": [
"142:54",
0
],
"negative": [
"142:67",
1
],
"noise_seed": 502081291840272,
"positive": [
"142:67",
0
],
"return_with_leftover_noise": "enable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 0,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"142:58": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "disable",
"cfg": 1,
"end_at_step": 10000,
"latent_image": [
"142:57",
0
],
"model": [
"142:55",
0
],
"negative": [
"142:67",
1
],
"noise_seed": 0,
"positive": [
"142:67",
0
],
"return_with_leftover_noise": "disable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 2,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"142:6": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"142:38",
0
],
"text": "A time-lapse from a locked-off camera on the same hilltop: summer fades into autumn as the sun sinks to golden hour behind the lone hawthorn tree. The leaves turn from green to amber, orange and deep red, a few of them tumbling away in gusts of wind. The tall grass dries to a warm gold, and the tree's long shadow stretches slowly across the hill as the low sun slides sideways. Clouds glow copper and pink. The tree, the hill and the horizon stay exactly in place; only the season, the sky and the light change, smoothly and continuously, in one fluid shot with no cuts."
},
"_meta": {
"title": "CLIP Text Encode (Positive Prompt)"
}
},
"142:60": {
"class_type": "CreateVideo",
"inputs": {
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": 16,
"images": [
"142:8",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"142:67": {
"class_type": "WanFirstLastFrameToVideo",
"inputs": {
"batch_size": 1,
"end_image": [
"126",
0
],
"height": 480,
"length": 25,
"negative": [
"142:7",
0
],
"positive": [
"142:6",
0
],
"start_image": [
"124",
0
],
"vae": [
"142:39",
0
],
"width": 832
},
"_meta": {
"title": "WanFirstLastFrameToVideo"
}
},
"142:7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"142:38",
0
],
"text": "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"
},
"_meta": {
"title": "CLIP Text Encode (Negative Prompt)"
}
},
"142:8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"142:58",
0
],
"vae": [
"142:39",
0
]
},
"_meta": {
"title": "VAE Decode"
}
},
"142:91": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors",
"model": [
"142:37",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"142:92": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors",
"model": [
"142:56",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"143:37": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"143:38": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "umt5_xxl_fp8_e4m3fn_scaled.safetensors",
"device": "default",
"type": "wan"
},
"_meta": {
"title": "Load CLIP"
}
},
"143:39": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "wan_2.1_vae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"143:54": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"143:91",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"143:55": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"143:92",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"143:56": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"143:57": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "enable",
"cfg": 1,
"end_at_step": 2,
"latent_image": [
"143:67",
2
],
"model": [
"143:54",
0
],
"negative": [
"143:67",
1
],
"noise_seed": 282609046210025,
"positive": [
"143:67",
0
],
"return_with_leftover_noise": "enable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 0,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"143:58": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "disable",
"cfg": 1,
"end_at_step": 10000,
"latent_image": [
"143:57",
0
],
"model": [
"143:55",
0
],
"negative": [
"143:67",
1
],
"noise_seed": 0,
"positive": [
"143:67",
0
],
"return_with_leftover_noise": "disable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 2,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"143:6": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"143:38",
0
],
"text": "A time-lapse from a locked-off camera on the same hilltop: autumn gives way to deep winter as dusk falls over the lone hawthorn tree. The last red leaves blow off and the branches stand bare. Snow begins to fall, first in a few flakes, then thickly, settling on every branch and smoothing the whole hill under a soft white blanket. The warm sunset drains into a cold blue-violet sky with a thin band of last light on the horizon. The tree, the hill and the horizon stay exactly in place; only the season, the sky and the light change, smoothly and continuously, in one fluid shot with no cuts."
},
"_meta": {
"title": "CLIP Text Encode (Positive Prompt)"
}
},
"143:60": {
"class_type": "CreateVideo",
"inputs": {
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": 16,
"images": [
"143:8",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"143:67": {
"class_type": "WanFirstLastFrameToVideo",
"inputs": {
"batch_size": 1,
"end_image": [
"128",
0
],
"height": 480,
"length": 25,
"negative": [
"143:7",
0
],
"positive": [
"143:6",
0
],
"start_image": [
"126",
0
],
"vae": [
"143:39",
0
],
"width": 832
},
"_meta": {
"title": "WanFirstLastFrameToVideo"
}
},
"143:7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"143:38",
0
],
"text": "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"
},
"_meta": {
"title": "CLIP Text Encode (Negative Prompt)"
}
},
"143:8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"143:58",
0
],
"vae": [
"143:39",
0
]
},
"_meta": {
"title": "VAE Decode"
}
},
"143:91": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors",
"model": [
"143:37",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"143:92": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors",
"model": [
"143:56",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"144:37": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"144:38": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "umt5_xxl_fp8_e4m3fn_scaled.safetensors",
"device": "default",
"type": "wan"
},
"_meta": {
"title": "Load CLIP"
}
},
"144:39": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "wan_2.1_vae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"144:54": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"144:91",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"144:55": {
"class_type": "ModelSamplingSD3",
"inputs": {
"model": [
"144:92",
0
],
"shift": 5
},
"_meta": {
"title": "ModelSamplingSD3"
}
},
"144:56": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_i2v_low_noise_14B_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"144:57": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "enable",
"cfg": 1,
"end_at_step": 2,
"latent_image": [
"144:67",
2
],
"model": [
"144:54",
0
],
"negative": [
"144:67",
1
],
"noise_seed": 1081792442176464,
"positive": [
"144:67",
0
],
"return_with_leftover_noise": "enable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 0,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"144:58": {
"class_type": "KSamplerAdvanced",
"inputs": {
"add_noise": "disable",
"cfg": 1,
"end_at_step": 10000,
"latent_image": [
"144:57",
0
],
"model": [
"144:55",
0
],
"negative": [
"144:67",
1
],
"noise_seed": 0,
"positive": [
"144:67",
0
],
"return_with_leftover_noise": "disable",
"sampler_name": "euler",
"scheduler": "simple",
"start_at_step": 2,
"steps": 4
},
"_meta": {
"title": "KSampler (Advanced)"
}
},
"144:6": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"144:38",
0
],
"text": "A time-lapse from a locked-off camera on the same hilltop: winter dusk deepens into a clear, still night above the bare, snow-dusted hawthorn tree. The last glow on the horizon fades away, the sky darkens to deep navy and black, and stars appear one by one until the faint band of the galaxy arcs over the tree. Moonlight rises from the side and glints on the snow, giving the hill a silver edge, while the stars wheel slowly across the sky. The tree, the hill and the horizon stay exactly in place; only the sky and the light change, smoothly and continuously, with no cuts."
},
"_meta": {
"title": "CLIP Text Encode (Positive Prompt)"
}
},
"144:60": {
"class_type": "CreateVideo",
"inputs": {
"bit_depth": "auto",
"codec": "none",
"color_space": "sRGB",
"fps": 16,
"images": [
"144:8",
0
]
},
"_meta": {
"title": "Create Video"
}
},
"144:67": {
"class_type": "WanFirstLastFrameToVideo",
"inputs": {
"batch_size": 1,
"end_image": [
"130",
0
],
"height": 480,
"length": 25,
"negative": [
"144:7",
0
],
"positive": [
"144:6",
0
],
"start_image": [
"128",
0
],
"vae": [
"144:39",
0
],
"width": 832
},
"_meta": {
"title": "WanFirstLastFrameToVideo"
}
},
"144:7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"144:38",
0
],
"text": "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"
},
"_meta": {
"title": "CLIP Text Encode (Negative Prompt)"
}
},
"144:8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"144:58",
0
],
"vae": [
"144:39",
0
]
},
"_meta": {
"title": "VAE Decode"
}
},
"144:91": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors",
"model": [
"144:37",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"144:92": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors",
"model": [
"144:56",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
}
})
workflow.set_input("62", "image", client.assets.from_url("https://comfy.org/catalog-media/355e43fc-9314-423c-9bb4-60e392f6b905/409d71e0-812f-475d-a889-049ee879c67f"))
workflow.set_input("122", "image", client.assets.from_url("https://comfy.org/catalog-media/e2928720-cfd8-44d8-88c1-5600eee2df6f/bbe0e7d2-dc7b-4fa2-b232-10a9463b7480"))
workflow.set_input("124", "image", client.assets.from_url("https://comfy.org/catalog-media/4ee3c115-a4fd-4fc8-a5de-66a1043ff111/014a35ec-33e3-452f-bbd2-5781cf7d92ca"))
workflow.set_input("126", "image", client.assets.from_url("https://comfy.org/catalog-media/103c269e-16ee-410f-94e9-7a1bb568299e/2fa4c753-28e4-4e34-a3da-e5457e6bc939"))
workflow.set_input("128", "image", client.assets.from_url("https://comfy.org/catalog-media/deac9ebe-14cb-4897-b6f6-064b209d708b/48acce99-e35a-4fe7-94ec-286f0db330a4"))
workflow.set_input("130", "image", client.assets.from_url("https://comfy.org/catalog-media/f0d772c2-0333-499d-ab54-6aa9f60ada1d/e2765e2c-73ad-4b0f-a727-6b2db2e34b2e"))
# 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("146"):
output.to_file(output.name)Experimente um exemplo
Sobre este fluxo de trabalho
O fluxo de trabalho Multi-Keyframe Video Stitching no ComfyUI foi projetado para transformar uma série de imagens estáticas em uma sequência de vídeo fluida. Utilizando seis quadros-chave, este fluxo gera automaticamente os quadros intermediários necessários para criar um vídeo contínuo. No centro desse processo está o modelo Wan2.2, que se destaca na interpretação do movimento entre imagens para garantir transições suaves. O fluxo começa com os nós LoadImage e ImageBatch, que preparam os quadros-chave para processamento. Os nós intermediários, como 11ac0cb0-38fb-41d4-9a55-7c44422b2516 e 1c5f4533-5d1e-4449-97ca-9027b9ea9c11, lidam com a interpolação dos quadros prevendo e preenchendo os caminhos de movimento. Por fim, os nós CreateVideo e SaveVideo compilam os quadros em um arquivo de vídeo coeso.
Este fluxo é especialmente útil para criadores que desejam produzir animações ou conteúdo em vídeo a partir de imagens estáticas sem exigir edição manual extensa. Ao automatizar o processo de interpolação, economiza tempo e garante consistência no movimento entre os quadros. O uso de nós avançados e do modelo Wan2.2 permite uma saída de vídeo de alta qualidade que mantém a integridade das imagens originais enquanto proporciona um fluxo natural de movimento.
Perguntas e respostas
Este fluxo é otimizado para seis quadros-chave para garantir transições suaves e processamento eficiente.
Sim, você pode ajustar a velocidade modificando as configurações de taxa de quadros no nó CreateVideo.
O nó SaveVideo suporta formatos comuns como MP4 e AVI para a saída final.
Não é necessário experiência prévia. O fluxo foi projetado para ser amigável e intuitivo para iniciantes.
Workflows relacionados
- Qwen-Image-Edit, Wan2.2
1 entrada de imagem Pilha Dividida - Qwen Multiângulo + Wan 2.2
- Wan2.2
1.2 Inicial – Imagem para Vídeo
- Wan2.2, Z-Image-Turbo
AIrt MAchIne
- Wan, Wan2.2
Imagem para Vídeo (Novo)
- Nano Banana, Google Gemini, Wan2.2
Restilizador em Loop
- Wan2.2
Substituição de Céu
- Wan2.2, Wan
Wan 2.2 14B Fun Control
- Wan2.2, Wan
Wan 2.2 14B Fun Controle de Câmera
- Wan2.2, Wan
Wan 2.2 14B Fun Inp
- Wan, Wan2.2
Wan 2.2 14B imagem para vídeo
- Wan2.2, Wan
Wan 2.2 14B Primeiro-Último quadro para Vídeo
- Wan2.2, Wan
Wan 2.2 14B texto para vídeo