Bernini-R: Image Edit
Re-light and restage one picture from a written instruction: Bernini-R re-renders the frame with matched light instead of layering a filter over it.
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。
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。
# 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({
"100": {
"class_type": "BatchImagesNode",
"inputs": {
"images.image0": [
"104",
0
]
},
"_meta": {
"title": "Batch Images"
}
},
"104": {
"class_type": "LoadImage",
"inputs": {
"image": "example.png"
},
"_meta": {
"title": "Load Image"
}
},
"114": {
"class_type": "LoadImage",
"inputs": {
"image": "UPLOADED_image_FILENAME"
},
"_meta": {
"title": "Load Image"
}
},
"115": {
"class_type": "SaveImage",
"inputs": {
"filename_prefix": "Bernini-R",
"images": [
"76:16",
0
]
},
"_meta": {
"title": "Save Image"
}
},
"116": {
"class_type": "ResizeImageMaskNode",
"inputs": {
"input": [
"114",
0
],
"resize_type": "scale longer dimension",
"resize_type.longer_size": 1280,
"scale_method": "area"
},
"_meta": {
"title": "Resize Image/Mask"
}
},
"117": {
"class_type": "GetImageSize",
"inputs": {
"image": [
"116",
0
]
},
"_meta": {
"title": "Get Image Size"
}
},
"118": {
"class_type": "ImageCompare",
"inputs": {
"compare_view": None,
"image_a": [
"114",
0
],
"image_b": [
"76:16",
0
]
},
"_meta": {
"title": "Compare Images"
}
},
"76:11": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "lightx2v_T2V_14B_cfg_step_distill_v2_lora_rank64_bf16.safetensors",
"model": [
"76:5",
0
],
"strength_model": 3
},
"_meta": {
"title": "Load LoRA"
}
},
"76:12": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_bernini_r_low_noise_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"76:120": {
"class_type": "PrimitiveStringMultiline",
"inputs": {
"value": "make it night"
},
"_meta": {
"title": "Text Multiline (User Prompt)"
}
},
"76:15": {
"class_type": "SamplerCustom",
"inputs": {
"add_noise": False,
"cfg": [
"76:72",
0
],
"latent_image": [
"76:19",
0
],
"model": [
"76:53",
0
],
"negative": [
"76:50",
1
],
"noise_seed": 0,
"positive": [
"76:50",
0
],
"sampler": [
"76:27",
0
],
"sigmas": [
"76:17",
1
]
},
"_meta": {
"title": "SamplerCustom"
}
},
"76:16": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"76:15",
0
],
"vae": [
"76:7",
0
]
},
"_meta": {
"title": "VAE Decode"
}
},
"76:17": {
"class_type": "SplitSigmas",
"inputs": {
"sigmas": [
"76:18",
0
],
"step": [
"76:68",
0
]
},
"_meta": {
"title": "SplitSigmas"
}
},
"76:18": {
"class_type": "BasicScheduler",
"inputs": {
"denoise": 1,
"model": [
"76:12",
0
],
"scheduler": "simple",
"steps": [
"76:64",
0
]
},
"_meta": {
"title": "BasicScheduler"
}
},
"76:19": {
"class_type": "SamplerCustom",
"inputs": {
"add_noise": True,
"cfg": [
"76:72",
0
],
"latent_image": [
"76:50",
2
],
"model": [
"76:52",
0
],
"negative": [
"76:50",
1
],
"noise_seed": 283365432432581,
"positive": [
"76:50",
0
],
"sampler": [
"76:27",
0
],
"sigmas": [
"76:17",
0
]
},
"_meta": {
"title": "SamplerCustom"
}
},
"76:27": {
"class_type": "KSamplerSelect",
"inputs": {
"sampler_name": "res_multistep"
},
"_meta": {
"title": "KSamplerSelect"
}
},
"76:29": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "lightx2v_T2V_14B_cfg_step_distill_v2_lora_rank64_bf16.safetensors",
"model": [
"76:12",
0
],
"strength_model": 1.5
},
"_meta": {
"title": "Load LoRA"
}
},
"76:3": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"76:9",
0
],
"text": [
"76:59",
0
]
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"76:4": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"76:9",
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 (Prompt)"
}
},
"76:5": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "wan2.2_bernini_r_high_noise_fp8_scaled.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"76:50": {
"class_type": "BerniniConditioning",
"inputs": {
"batch_size": 1,
"height": [
"117",
1
],
"length": 1,
"negative": [
"76:4",
0
],
"positive": [
"76:3",
0
],
"ref_max_size": 848,
"source_video": [
"116",
0
],
"vae": [
"76:7",
0
],
"width": [
"117",
0
]
},
"_meta": {
"title": "Bernini Conditioning"
}
},
"76:52": {
"class_type": "ComfySwitchNode",
"inputs": {
"on_false": [
"76:5",
0
],
"on_true": [
"76:11",
0
],
"switch": [
"76:70",
0
]
},
"_meta": {
"title": "Switch (High Noise)"
}
},
"76:53": {
"class_type": "ComfySwitchNode",
"inputs": {
"on_false": [
"76:12",
0
],
"on_true": [
"76:29",
0
],
"switch": [
"76:70",
0
]
},
"_meta": {
"title": "Switch (Low Noise)"
}
},
"76:54": {
"class_type": "CustomCombo",
"inputs": {
"choice": "Default"
},
"_meta": {
"title": "Custom Combo"
}
},
"76:57:1": {
"class_type": "PreviewAny",
"inputs": {
"source": [
"76:57:56",
0
]
},
"_meta": {
"title": "Preview as Text"
}
},
"76:57:2": {
"class_type": "RegexExtract",
"inputs": {
"case_insensitive": False,
"dotall": False,
"group_index": 1,
"mode": "First Group",
"multiline": False,
"regex_pattern": [
"76:57:8",
0
],
"string": "You are a helpful assistant.\nYou are a helpful assistant specialized in text-to-image generation.\nYou are a helpful assistant specialized in text-to-video generation.\nYou are a helpful assistant specialized in image editing.\nYou are a helpful assistant specialized in subject-to-image generation."
},
"_meta": {
"title": "Extract Text"
}
},
"76:57:56": {
"class_type": "PrimitiveInt",
"inputs": {
"value": [
"76:54",
1
]
},
"_meta": {
"title": "Int (line index)"
}
},
"76:57:8": {
"class_type": "StringReplace",
"inputs": {
"find": "index",
"replace": [
"76:57:1",
0
],
"string": "^(?:[^\\n]*\\n){index}([^\\n]*)(?:\\n|$)"
},
"_meta": {
"title": "Replace Text"
}
},
"76:59": {
"class_type": "StringConcatenate",
"inputs": {
"delimiter": "",
"string_a": [
"76:57:2",
0
],
"string_b": [
"76:120",
0
]
},
"_meta": {
"title": "Concatenate Text"
}
},
"76:62": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 40
},
"_meta": {
"title": "Int (Steps)"
}
},
"76:63": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 6
},
"_meta": {
"title": "Int (Steps)"
}
},
"76:64": {
"class_type": "ComfySwitchNode",
"inputs": {
"on_false": [
"76:62",
0
],
"on_true": [
"76:63",
0
],
"switch": [
"76:70",
0
]
},
"_meta": {
"title": "Switch (Steps)"
}
},
"76:66": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 20
},
"_meta": {
"title": "Int (Split Steps)"
}
},
"76:67": {
"class_type": "PrimitiveInt",
"inputs": {
"value": 3
},
"_meta": {
"title": "Int (Split Steps)"
}
},
"76:68": {
"class_type": "ComfySwitchNode",
"inputs": {
"on_false": [
"76:66",
0
],
"on_true": [
"76:67",
0
],
"switch": [
"76:70",
0
]
},
"_meta": {
"title": "Switch (Low Steps)"
}
},
"76:7": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "Wan2_1_VAE_bf16.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"76:70": {
"class_type": "PrimitiveBoolean",
"inputs": {
"value": True
},
"_meta": {
"title": "Boolean (Enable Turbo LoRA?)"
}
},
"76:71": {
"class_type": "PrimitiveFloat",
"inputs": {
"value": 5
},
"_meta": {
"title": "Float (CFG)"
}
},
"76:72": {
"class_type": "ComfySwitchNode",
"inputs": {
"on_false": [
"76:71",
0
],
"on_true": [
"76:73",
0
],
"switch": [
"76:70",
0
]
},
"_meta": {
"title": "Switch (CFG)"
}
},
"76:73": {
"class_type": "PrimitiveFloat",
"inputs": {
"value": 1
},
"_meta": {
"title": "Float (CFG)"
}
},
"76:9": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "umt5_xxl_fp8_e4m3fn_scaled.safetensors",
"device": "default",
"type": "wan"
},
"_meta": {
"title": "Load CLIP"
}
}
})
workflow.set_input("114", "image", client.assets.from_url("https://comfy.org/catalog-media/dd9ca340-ddf2-4513-9fa7-4baa5f2cb8f1/cd05b4c0-2a7f-4985-9b7b-5d1f70cb85f5"))
# 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("115"):
output.to_file(output.name)例を試す
このワークフローについて
Bernini-R is an editing model built on the Wan stack: one image sets the framing and the scene, and the prompt says what that scene should become — night instead of day, a warm interior instead of a grey studio, a different season. Because it generates the frame rather than compositing over it, the light is rebuilt from the new instruction instead of being pasted on top, so shadows on the ground, reflections and the sky all agree with the change. The conditioning node takes the resized source at length 1, so a still costs one frame of sampling, and a distilled two-stage sampler runs 3 steps. A comparison node shows the source against the result in the same run, which is how you judge whether the relight held the geometry.
Graph: LoadImage -> ResizeImageMaskNode (longer side 1280) -> BerniniConditioning (length 1) -> two-stage SamplerCustom (high/low noise, 3 steps, res_multistep) -> VAEDecode -> SaveImage (png), plus ImageCompare of the source against the result.
関連するワークフロー
- BBernini-R
Bernini-R: Video Edit
- Nano Banana Pro, Gemini3 Pro Image Preview, Google
キャラクターターンアラウンドを作成
- Qwen-Image-Edit
素材を変更
- Qwen-Image-Edit
被写体を新しいアングルから見る
- Flux.2 Dev, Flux
製品モックアップを作成
- Nano Banana Pro
キャラクターに衣装を試着
- Seedream
商品を新しいシーンに配置
- SSeedVR2
アップスケールとディテール復元
- Nano Banana Pro
商品写真をシャープに
- Topaz
イラストをアップスケール
- HHitPaw
ポートレートのディテールを復元
- FLUX.1 Fill
選択した領域を編集