NetaYume Lumina 文生圖
高品質動漫風格圖像產生,角色理解更強、紋理細節更豐富,基於 Danbooru 資料集從 Neta Lumina 微調。
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。
Loading the graph
Open full-size workflow preview此工作流程中的檔案
- 輸出
- 圖片
- 作者
- ComfyUI
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。
What you need
- Runs on
- Comfy Cloud
- 金鑰
- COMFY_API_KEY
需要付費 Cloud 方案與點數。
# 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({
"3": {
"class_type": "KSampler",
"inputs": {
"cfg": 4,
"denoise": 1,
"latent_image": [
"13",
0
],
"model": [
"11",
0
],
"negative": [
"25:7",
0
],
"positive": [
"26:7",
0
],
"sampler_name": "res_multistep",
"scheduler": "simple",
"seed": 942500763821827,
"steps": 30
},
"_meta": {
"title": "KSampler"
}
},
"4": {
"class_type": "CheckpointLoaderSimple",
"inputs": {
"ckpt_name": "NetaYumev35_pretrained_all_in_one.safetensors"
},
"_meta": {
"title": "Load Checkpoint"
}
},
"8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"3",
0
],
"vae": [
"4",
2
]
},
"_meta": {
"title": "VAE Decode"
}
},
"9": {
"class_type": "SaveImage",
"inputs": {
"filename_prefix": "NetaYume_Lumina_3.5",
"images": [
"8",
0
]
},
"_meta": {
"title": "Save Image"
}
},
"11": {
"class_type": "ModelSamplingAuraFlow",
"inputs": {
"model": [
"4",
0
],
"sampling": "flow",
"shift": 4
},
"_meta": {
"title": "ModelSamplingAuraFlow"
}
},
"13": {
"class_type": "EmptySD3LatentImage",
"inputs": {
"batch_size": 1,
"height": 768,
"width": 1344
},
"_meta": {
"title": "EmptySD3LatentImage"
}
},
"25:22": {
"class_type": "StringConcatenate",
"inputs": {
"delimiter": "",
"string_a": [
"25:23",
0
],
"string_b": [
"25:24",
0
]
},
"_meta": {
"title": "Concatenate Text"
}
},
"25:23": {
"class_type": "PrimitiveStringMultiline",
"inputs": {
"value": "You are an assistant designed to generate low-quality images based on textual prompts <Prompt Start> "
},
"_meta": {
"title": "System prompt"
}
},
"25:24": {
"class_type": "PrimitiveStringMultiline",
"inputs": {
"value": "blurry, worst quality, low quality, jpeg artifacts, signature, watermark, username, error, deformed hands, bad anatomy, extra limbs, poorly drawn hands, poorly drawn face, mutation, deformed, extra eyes, extra arms, extra legs, malformed limbs, fused fingers, too many fingers, long neck, cross-eyed, bad proportions, missing arms, missing legs, extra digit, fewer digits, cropped"
},
"_meta": {
"title": "System prompt"
}
},
"25:7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"4",
1
],
"text": [
"25:22",
0
]
},
"_meta": {
"title": "CLIP Text Encode (Negative Prompt)"
}
},
"26:22": {
"class_type": "StringConcatenate",
"inputs": {
"delimiter": "",
"string_a": [
"26:23",
0
],
"string_b": [
"26:24",
0
]
},
"_meta": {
"title": "Concatenate Text"
}
},
"26:23": {
"class_type": "PrimitiveStringMultiline",
"inputs": {
"value": "You are an assistant designed to generate high quality anime images based on textual prompts. <Prompt Start> "
},
"_meta": {
"title": "System prompt"
}
},
"26:24": {
"class_type": "PrimitiveStringMultiline",
"inputs": {
"value": "A richly detailed anime illustration depicts a slender young East Asian herbalist standing atop a sun-drenched coastal terrace. Her long, flowing raven hair is loosely braided with delicate jade pins and fragrant jasmine blossoms. She wears layered, billowing silk robes of pale lavender, saffron yellow, and mint green, embroidered with intricate golden threadwork along the sleeves. In one hand, she cradles an ornate brass astrolabe reflecting warm sunlight, while a woven bamboo basket rests beside her feet. Behind her, weathered ivory marble colonnades are entwined with vibrant violet wisteria and creeping ivy. Bathed in a radiant palette of turquoise ocean waters, peach-tinged clouds, and warm golden hour light, drifting petals dance across an expansive vista of distant mist-shrouded islands and sparkling sea foam."
},
"_meta": {
"title": "Prompt"
}
},
"26:7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"4",
1
],
"text": [
"26:22",
0
]
},
"_meta": {
"title": "CLIP Text Encode (Positive Prompt)"
}
}
})
# 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("9"):
output.to_file(output.name)試試範例
關於此工作流程
NetaYume Lumina 文字轉圖像工作流程專為生成高品質動漫風格圖像而設計,能精確理解角色細節與紋理。此流程採用 OmniGen 模型,該模型以 Neta Lumina 為基礎,並在著名的 Danbooru 動漫圖像資料集上微調。流程包含多個關鍵節點,如 KSampler 用於取樣、CheckpointLoaderSimple 載入模型,以及 VAEDecode 將潛空間解碼為圖像。透過輸入描述性文字提示,使用者可創作細膩且鮮明的動漫圖像。
技術上,流程分為三大步驟:載入模型、設定圖像尺寸、撰寫提示詞。CheckpointLoaderSimple 節點負責載入 OmniGen 模型,KSampler 節點則處理取樣以確保高品質輸出。VAEDecode 節點將潛在圖像資料轉換為最終視覺輸出,並以 SaveImage 節點儲存。此流程特別適合希望精確控制角色特徵與場景構圖的藝術家與創作者,製作動漫風插畫。
問答
建議尺寸為 1024x1024,但也可嘗試更高解析度如 1024x1536。
可以,您可在提示詞中用 @ 符號加入畫師標籤,以影響生成圖像的風格。
可參考 Neta.art 部落格上的 Neta Lumina 提示詞手冊,獲得有效撰寫提示詞的指引。
輸出圖像會儲存在您系統的 'ComfyUI/outputs' 資料夾。