AnimateDiff - IPAdapters Looping Animation
Drag and drop then dream through a stack of images
ワークフローをダウンロードCloudで試す API ドキュメント
00:00
ワークフローの内部
ソーステンプレートの読み取り専用プレビューです。Cloud で開くと、ノードを確認して変更できます。
Loading the graph
Open full-size workflow preview- 出力
- 動画
- 作成者
- Purz
コードからこのモデルを呼び出す
以下のリクエストでは、Playground と同じ準備済みグラフと設定を使用します。送信ごとに有料実行が開始される場合があります。結果が不明な送信を自動的に再試行しないでください。
1 API キーを取得 ↗
What you need
- Runs on
- Comfy Cloud
- キー
- COMFY_API_KEY
- ファイル
- コードによって URL からアップロードされます
有料の Cloud プランとクレジットが必要です。
API JSONをダウンロードAPI ドキュメント ↗
# 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({
"14": {
"class_type": "VHS_VideoCombine",
"inputs": {
"crf": 12,
"filename_prefix": "AnimateDiff",
"format": "video/h264-mp4",
"frame_rate": 24,
"images": [
"38:13",
0
],
"loop_count": 0,
"pingpong": False,
"pix_fmt": "yuv420p",
"save_metadata": True,
"save_output": True,
"trim_to_audio": False
},
"_meta": {
"title": "Video Combine 🎥🅥🅗🅢"
}
},
"61": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_1_FILENAME"
},
"_meta": {
"title": "Image 1"
}
},
"62": {
"class_type": "ImageScaleToTotalPixels",
"inputs": {
"image": [
"61",
0
],
"megapixels": 1,
"resolution_steps": 1,
"upscale_method": "lanczos"
},
"_meta": {
"title": "Rescale Image 1"
}
},
"63": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_2_FILENAME"
},
"_meta": {
"title": "Image 2"
}
},
"64": {
"class_type": "ImageScaleToTotalPixels",
"inputs": {
"image": [
"63",
0
],
"megapixels": 1,
"resolution_steps": 1,
"upscale_method": "lanczos"
},
"_meta": {
"title": "Rescale Image 2"
}
},
"65": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_3_FILENAME"
},
"_meta": {
"title": "Image 3"
}
},
"66": {
"class_type": "ImageScaleToTotalPixels",
"inputs": {
"image": [
"65",
0
],
"megapixels": 1,
"resolution_steps": 1,
"upscale_method": "lanczos"
},
"_meta": {
"title": "Rescale Image 3"
}
},
"67": {
"class_type": "ImageBatchMulti",
"inputs": {
"image_1": [
"62",
0
],
"image_2": [
"64",
0
],
"image_3": [
"66",
0
],
"image_4": [
"62",
0
],
"inputcount": 4
},
"_meta": {
"title": "Image Batch Multi"
}
},
"76": {
"class_type": "VHS_VideoCombine",
"inputs": {
"crf": 12,
"filename_prefix": "AnimateDiff",
"format": "video/h264-mp4",
"frame_rate": 24,
"images": [
"77:75",
0
],
"loop_count": 0,
"pingpong": False,
"pix_fmt": "yuv420p",
"save_metadata": True,
"save_output": True,
"trim_to_audio": False
},
"_meta": {
"title": "Video Combine 🎥🅥🅗🅢"
}
},
"38:1": {
"class_type": "ADE_UseEvolvedSampling",
"inputs": {
"beta_schedule": "lcm >> sqrt_linear",
"context_options": [
"38:7",
0
],
"m_models": [
"38:2",
0
],
"model": [
"38:8",
0
],
"sample_settings": [
"38:6",
0
]
},
"_meta": {
"title": "Use Evolved Sampling 🎭🅐🅓②"
}
},
"38:10": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"38:8",
1
],
"text": "nsfw, nude"
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"38:11": {
"class_type": "KSampler",
"inputs": {
"cfg": 2,
"denoise": 1,
"latent_image": [
"38:12",
0
],
"model": [
"38:36",
0
],
"negative": [
"38:10",
0
],
"positive": [
"38:9",
0
],
"sampler_name": "lcm",
"scheduler": "beta",
"seed": 484814760137008,
"steps": 13
},
"_meta": {
"title": "KSampler"
}
},
"38:12": {
"class_type": "EmptyLatentImage",
"inputs": {
"batch_size": [
"38:27",
2
],
"height": 576,
"width": 1024
},
"_meta": {
"title": "Empty Latent Image"
}
},
"38:13": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"38:11",
0
],
"vae": [
"38:8",
2
]
},
"_meta": {
"title": "VAE Decode"
}
},
"38:2": {
"class_type": "ADE_ApplyAnimateDiffModelSimple",
"inputs": {
"effect_multival": [
"38:5",
0
],
"motion_model": [
"38:3",
0
],
"scale_multival": [
"38:4",
0
]
},
"_meta": {
"title": "Apply AnimateDiff Model 🎭🅐🅓②"
}
},
"38:21": {
"class_type": "IPAdapterUnifiedLoader",
"inputs": {
"model": [
"38:1",
0
],
"preset": "PLUS (high strength)"
},
"_meta": {
"title": "IPAdapter Unified Loader"
}
},
"38:27": {
"class_type": "IPAdapterWeights",
"inputs": {
"add_ending_frames": 16,
"add_starting_frames": 16,
"end_frame": 9999,
"frames": 48,
"image": [
"67",
0
],
"method": "alternate batches",
"start_frame": 0,
"timing": "ease_in",
"weights": "1.0, 0.0"
},
"_meta": {
"title": "IPAdapter Weights"
}
},
"38:3": {
"class_type": "ADE_LoadAnimateDiffModel",
"inputs": {
"model_name": "AnimateLCM_sd15_t2v.ckpt"
},
"_meta": {
"title": "Load AnimateDiff Model 🎭🅐🅓②"
}
},
"38:32": {
"class_type": "IPAdapterBatch",
"inputs": {
"embeds_scaling": "V only",
"encode_batch_size": 0,
"end_at": 1,
"image": [
"38:27",
3
],
"ipadapter": [
"38:21",
1
],
"model": [
"38:21",
0
],
"start_at": 0,
"weight": [
"38:27",
0
],
"weight_type": "ease in"
},
"_meta": {
"title": "IPAdapter Batch (Adv.)"
}
},
"38:33": {
"class_type": "IPAdapterBatch",
"inputs": {
"embeds_scaling": "V only",
"encode_batch_size": 0,
"end_at": 1,
"image": [
"38:27",
4
],
"ipadapter": [
"38:21",
1
],
"model": [
"38:32",
0
],
"start_at": 0,
"weight": [
"38:27",
1
],
"weight_type": "ease out"
},
"_meta": {
"title": "IPAdapter Batch (Adv.)"
}
},
"38:36": {
"class_type": "PatchModelAddDownscale",
"inputs": {
"block_number": 3,
"downscale_after_skip": True,
"downscale_factor": 2,
"downscale_method": "bicubic",
"end_percent": 0.35,
"model": [
"38:33",
0
],
"start_percent": 0,
"upscale_method": "bicubic"
},
"_meta": {
"title": "PatchModelAddDownscale (Kohya Deep Shrink)"
}
},
"38:4": {
"class_type": "ADE_MultivalDynamic",
"inputs": {
"float_val": 1
},
"_meta": {
"title": "Multival 🎭🅐🅓"
}
},
"38:5": {
"class_type": "ADE_MultivalDynamic",
"inputs": {
"float_val": 1
},
"_meta": {
"title": "Multival 🎭🅐🅓"
}
},
"38:6": {
"class_type": "ADE_AnimateDiffSamplingSettings",
"inputs": {
"adapt_denoise_steps": False,
"batch_offset": 0,
"noise_type": "repeated_context",
"seed_gen": "comfy",
"seed_offset": 0,
"seed_override": 0
},
"_meta": {
"title": "Sample Settings 🎭🅐🅓"
}
},
"38:7": {
"class_type": "ADE_LoopedUniformContextOptions",
"inputs": {
"closed_loop": False,
"context_length": 16,
"context_overlap": 4,
"context_stride": 1,
"fuse_method": "pyramid",
"guarantee_steps": 1,
"start_percent": 0,
"use_on_equal_length": False
},
"_meta": {
"title": "Context Options◆Looped Uniform 🎭🅐🅓"
}
},
"38:8": {
"class_type": "CheckpointLoaderSimple",
"inputs": {
"ckpt_name": "realismByStableYogi_v4LCM.safetensors"
},
"_meta": {
"title": "Load Checkpoint"
}
},
"38:9": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"38:8",
1
],
"text": "A slow, hypnotic dream in which a giant hourglass and its tiny cloaked wanderer drift through changing worlds: molten golden sand pours and glows, then freezes into shimmering blue ice under falling snow, then softens into moss, vines and blooming flowers in warm sunset light. Gentle continuous morphing, swirling particles of light and drifting mist, rippling reflections on the water, subtle camera breathing, painterly brushwork, rich saturated colour, smooth seamless loop with no cuts, ethereal and calm atmosphere, like a half-remembered dream."
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"77:69": {
"class_type": "RepeatImageBatch",
"inputs": {
"amount": [
"77:70",
3
],
"image": [
"62",
0
]
},
"_meta": {
"title": "Repeat Image Batch"
}
},
"77:70": {
"class_type": "GetImageSizeAndCount",
"inputs": {
"image": [
"38:13",
0
]
},
"_meta": {
"title": "Get Image Size & Count"
}
},
"77:71": {
"class_type": "RepeatImageBatch",
"inputs": {
"amount": [
"77:70",
3
],
"image": [
"64",
0
]
},
"_meta": {
"title": "Repeat Image Batch"
}
},
"77:72": {
"class_type": "RepeatImageBatch",
"inputs": {
"amount": [
"77:70",
3
],
"image": [
"66",
0
]
},
"_meta": {
"title": "Repeat Image Batch"
}
},
"77:73": {
"class_type": "ImageStitch",
"inputs": {
"direction": "right",
"image1": [
"77:69",
0
],
"image2": [
"77:71",
0
],
"match_image_size": True,
"spacing_color": "black",
"spacing_width": 10
},
"_meta": {
"title": "Stitch Images"
}
},
"77:74": {
"class_type": "ImageStitch",
"inputs": {
"direction": "right",
"image1": [
"77:73",
0
],
"image2": [
"77:72",
0
],
"match_image_size": True,
"spacing_color": "black",
"spacing_width": 10
},
"_meta": {
"title": "Stitch Images"
}
},
"77:75": {
"class_type": "ImageStitch",
"inputs": {
"direction": "down",
"image1": [
"77:74",
0
],
"image2": [
"38:13",
0
],
"match_image_size": True,
"spacing_color": "black",
"spacing_width": 10
},
"_meta": {
"title": "Stitch Images"
}
}
})
workflow.set_input("61", "image", client.assets.from_url("https://comfy.org/catalog-media/44f132a2-b86b-423a-9dea-cd45e8dc8a45/137fbc6b-7de9-4656-a813-7b6666d8b3da"))
workflow.set_input("63", "image", client.assets.from_url("https://comfy.org/catalog-media/0672b137-0632-4be0-80be-38150ed1f9d8/b3faec0a-3433-4529-9937-75cd0029a35b"))
workflow.set_input("65", "image", client.assets.from_url("https://comfy.org/catalog-media/746bafc1-ced6-4d5c-8460-302f786532b2/1d5e894e-b82a-439c-98c4-021685a5852c"))
# 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("14"):
output.to_file(output.name)
for output in job.get_outputs("76"):
output.to_file(output.name)例を試す
このワークフローについて
This ComfyUI workflow, titled 'AnimateDiff - IPAdapters Looping Animation,' is designed to create seamless animated transitions between a series of images. It leverages the power of Stable Diffusion 1.5 (SD1.5) along with custom nodes such as 'comfyui-animatediff-evolved' and 'comfyui_ipadapter_plus' to achieve fluid animations.
The workflow uses a combination of nodes like 'VHS_VideoCombine' and 'ImageBatchMulti' to process multiple images in sequence, creating a looping animation effect. By utilizing Weighted IPAdapters, the workflow ensures that each image transition maintains a consistent style and theme, enhancing the overall visual coherence of the animation.
Technically, the workflow begins with loading images into the 'LoadImage' node, which are then scaled appropriately using the 'ImageScaleToTotalPixels' node. The 'ImageBatchMulti' node is crucial for handling multiple images, allowing them to be processed as a batch. The 'VHS_VideoCombine' node plays a pivotal role in merging these images into a continuous video loop. This setup is particularly useful for artists and content creators looking to produce engaging visual content that smoothly transitions between different scenes or concepts.
The integration of the 'comfyui-videohelpersuite' further aids in refining the video output, ensuring high-quality animations.
よくある質問
The 'Weighted IPAdapters' help maintain a consistent visual style across all images, ensuring smooth transitions in the animation.
This workflow is specifically designed for SD1.5. While it may work with other versions, results can vary and are not guaranteed.
Ensure that images are similar in style and theme. Adjust the 'Weighted IPAdapters' settings to fine-tune the transitions.
The workflow focuses on visual animation. You can add audio using video editing software after exporting the animation.