Améliorateur de Peau + Suréchantillonneur
Un flux de travail en un clic pour générer un portrait de personnage, affiner/ajouter des détails de peau, et agrandir en 4k. Le tout en moins de 60 secondes.
À propos
Le workflow « Skin Enhancer + Upscaler » est conçu pour simplifier la création de portraits de personnages de haute qualité en combinant l’amélioration des détails de peau et l’upscaling d’image dans un seul workflow efficace. En utilisant les modèles Z-Image-Turbo et SDXL, ce workflow permet de générer des portraits détaillés avec des textures de peau affinées, puis d’upscaler l’image finale en 4K en moins d’une minute. Des nœuds clés comme LoadImage et SaveImage gèrent l’entrée et la sortie, tandis que des nœuds personnalisés de comfyui-kjnodes et comfyui_essentials s’occupent des tâches plus fines de retouche des détails de peau et d’upscaling. Le workflow est particulièrement utile aux artistes et designers qui doivent produire rapidement des images haute résolution avec des traits du visage renforcés.
Techniquement, le workflow s’appuie sur des nœuds comme fb45acca-28ee-4342-9487-e4e2369c9883 et eeeed90f-caa6-493b-8da1-03f72f13e685 pour analyser et améliorer les traits du visage selon des prompts définis par l’utilisateur. Les utilisateurs peuvent préciser des caractéristiques comme des taches de rousseur, une pilosité faciale ou des préférences de maquillage, qui sont ensuite traitées pour affiner l’apparence du portrait. Les nœuds ComfySwitchNode et ImageCompare permettent de basculer entre différentes options de traitement, offrant de la flexibilité sur le rendu final. Ce workflow est idéal pour améliorer des portraits avec un minimum d’effort et un maximum d’efficacité, ce qui en fait un outil utile pour des projets professionnels comme personnels.
Ce que vous fournissez
- Lecture du workflow
S’exécute dans ComfyUI. Ce workflow ne peut pas s’exécuter sur cette page. Ouvrez-le dans Comfy Cloud, avec tout déjà chargé, ou téléchargez-le pour votre propre ComfyUI.
Essayer dans CloudDans le workflow
Un aperçu en lecture seule du graphe de nœuds. Ouvrez-le dans Comfy Cloud ou téléchargez-le pour modifier les nœuds.
Loading the graph
Open full-size workflow previewFichiers de ce workflow
- Auteur
- Rob
Nécessite des nœuds personnalisés
- Lecture du workflow
Utiliser ce workflow depuis du code
Généré à partir du graphe et des entrées du workflow lorsque cela est possible ; sinon, les exemples enregistrés sont affichés. Vérifiez les identifiants et les points de terminaison avant d’exécuter le code, ce qui peut entraîner des frais.
What you need
- Runs on
- Comfy Cloud
- Clé
- COMFY_API_KEY
Nécessite un forfait Cloud payant et des crédits.
# 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({
"17": {
"class_type": "LoadImage",
"inputs": {
"image": "input_face-1.png"
},
"_meta": {
"title": "Load Image"
}
},
"282": {
"class_type": "SaveImage",
"inputs": {
"filename_prefix": "portrait-upscaled",
"images": [
"283:27",
0
]
},
"_meta": {
"title": "Save Image"
}
},
"286": {
"class_type": "SaveImage",
"inputs": {
"filename_prefix": "portrait-z-image-turbo",
"images": [
"285:8",
0
]
},
"_meta": {
"title": "Save Image"
}
},
"300": {
"class_type": "ComfySwitchNode",
"inputs": {
"on_false": [
"285:8",
0
],
"on_true": [
"17",
0
],
"switch": False
},
"_meta": {
"title": "Image Switch"
}
},
"302": {
"class_type": "ImageCompare",
"inputs": {
"compare_view": {
"__value__": [
"",
""
]
},
"image_a": [
"283:27",
0
],
"image_b": [
"300",
0
]
},
"_meta": {
"title": "Image Compare (Before/After)"
}
},
"283:21": {
"class_type": "GetImageSize",
"inputs": {
"image": [
"284:180",
0
]
},
"_meta": {
"title": "Get Image Size"
}
},
"283:22": {
"class_type": "SeedVR2LoadVAEModel",
"inputs": {
"cache_model": False,
"decode_tile_overlap": 128,
"decode_tile_size": 1024,
"decode_tiled": True,
"device": "cuda:0",
"encode_tile_overlap": 128,
"encode_tile_size": 1024,
"encode_tiled": True,
"model": "ema_vae_fp16.safetensors",
"offload_device": "none",
"tile_debug": "false"
},
"_meta": {
"title": "SeedVR2 (Down)Load VAE Model"
}
},
"283:25": {
"class_type": "SimpleMath+",
"inputs": {
"a": [
"283:43",
0
],
"b": [
"283:30",
0
],
"value": "a*b"
},
"_meta": {
"title": "Upscale to Shortest Edge (pixels)"
}
},
"283:27": {
"class_type": "SeedVR2VideoUpscaler",
"inputs": {
"batch_size": 1,
"color_correction": "lab",
"dit": [
"283:28",
0
],
"enable_debug": False,
"image": [
"284:180",
0
],
"input_noise_scale": 0,
"latent_noise_scale": 0,
"max_resolution": 0,
"offload_device": "cpu",
"prepend_frames": 0,
"resolution": [
"283:25",
0
],
"seed": 74825977,
"temporal_overlap": 0,
"uniform_batch_size": False,
"vae": [
"283:22",
0
]
},
"_meta": {
"title": "SeedVR2 Video Upscaler (v2.5.24)"
}
},
"283:28": {
"class_type": "SeedVR2LoadDiTModel",
"inputs": {
"attention_mode": "sdpa",
"blocks_to_swap": 0,
"cache_model": False,
"device": "cuda:0",
"model": "seedvr2_ema_7b_fp16.safetensors",
"offload_device": "cpu",
"swap_io_components": False
},
"_meta": {
"title": "SeedVR2 (Down)Load DiT Model"
}
},
"283:30": {
"class_type": "PrimitiveFloat",
"inputs": {
"value": 4
},
"_meta": {
"title": "Upscale Factor"
}
},
"283:40": {
"class_type": "ImpactCompare",
"inputs": {
"a": [
"283:21",
0
],
"b": [
"283:21",
1
],
"cmp": "a <= b"
},
"_meta": {
"title": "ImpactCompare"
}
},
"283:43": {
"class_type": "LayerUtility: If ",
"inputs": {
"if_condition": [
"283:40",
0
],
"when_FALSE": [
"283:21",
1
],
"when_TRUE": [
"283:21",
0
]
},
"_meta": {
"title": "IfElse"
}
},
"284:116": {
"class_type": "CheckpointLoaderSimple",
"inputs": {
"ckpt_name": "Juggernaut-XL_v9_RunDiffusionPhoto_v2.safetensors"
},
"_meta": {
"title": "Load Checkpoint"
}
},
"284:117": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"284:116",
1
],
"text": "plastic, smooth, glossy, cartoon, anime, worst quality, low quality, illustration, 3d, 2d, painting, cartoons, sketch, anime"
},
"_meta": {
"title": "Negative Prompt"
}
},
"284:156": {
"class_type": "FaceParsingResultsParser(FaceParsing)",
"inputs": {
"background": False,
"cloth": False,
"ear_r": False,
"eye_g": False,
"hair": False,
"hat": False,
"l_brow": True,
"l_ear": False,
"l_eye": False,
"l_lip": True,
"mouth": False,
"neck": False,
"neck_l": False,
"nose": True,
"r_brow": True,
"r_ear": False,
"r_eye": False,
"result": [
"284:161",
1
],
"skin": True,
"u_lip": True
},
"_meta": {
"title": "FaceParsingResultsParser(FaceParsing)"
}
},
"284:157": {
"class_type": "FaceParsingModelLoader(FaceParsing)",
"inputs": {
"device": "cuda"
},
"_meta": {
"title": "FaceParsingModelLoader(FaceParsing)"
}
},
"284:159": {
"class_type": "BBoxDetectorLoader(FaceParsing)",
"inputs": {
"model_name": "bbox/face_yolov8m.pt"
},
"_meta": {
"title": "BBoxDetectorLoader(FaceParsing)"
}
},
"284:160": {
"class_type": "BBoxDetect(FaceParsing)",
"inputs": {
"bbox_detector": [
"284:159",
0
],
"by_ratio": False,
"dilation": 8,
"dilation_ratio": 0.2,
"image": [
"300",
0
],
"threshold": 0.3
},
"_meta": {
"title": "BBoxDetect(FaceParsing)"
}
},
"284:161": {
"class_type": "FaceParse(FaceParsing)",
"inputs": {
"image": [
"284:163",
0
],
"model": [
"284:157",
0
],
"processor": [
"284:165",
0
]
},
"_meta": {
"title": "FaceParse(FaceParsing)"
}
},
"284:162": {
"class_type": "BBoxListItemSelect(FaceParsing)",
"inputs": {
"bbox_list": [
"284:160",
0
],
"index": 0
},
"_meta": {
"title": "BBoxListItemSelect(FaceParsing)"
}
},
"284:163": {
"class_type": "ImageCropWithBBox(FaceParsing)",
"inputs": {
"bbox": [
"284:162",
0
],
"image": [
"300",
0
]
},
"_meta": {
"title": "ImageCropWithBBox(FaceParsing)"
}
},
"284:165": {
"class_type": "FaceParsingProcessorLoader(FaceParsing)",
"inputs": {},
"_meta": {
"title": "FaceParsingProcessorLoader(FaceParsing)"
}
},
"284:176": {
"class_type": "ImageCompositeMasked",
"inputs": {
"destination": [
"284:163",
0
],
"mask": [
"284:179",
0
],
"resize_source": False,
"source": [
"284:196",
0
],
"x": 0,
"y": 0
},
"_meta": {
"title": "Image Composite Masked"
}
},
"284:179": {
"class_type": "GrowMaskWithBlur",
"inputs": {
"blur_radius": 8,
"decay_factor": 1,
"expand": -4,
"fill_holes": False,
"flip_input": False,
"incremental_expandrate": 0,
"lerp_alpha": 1,
"mask": [
"284:156",
0
],
"tapered_corners": True
},
"_meta": {
"title": "Grow Mask With Blur"
}
},
"284:180": {
"class_type": "ImageInsertWithBBox(FaceParsing)",
"inputs": {
"bbox": [
"284:162",
0
],
"image": [
"284:176",
0
],
"image_src": [
"300",
0
]
},
"_meta": {
"title": "ImageInsertWithBBox(FaceParsing)"
}
},
"284:193": {
"class_type": "GetImageSize",
"inputs": {
"image": [
"284:163",
0
]
},
"_meta": {
"title": "Get Image Size"
}
},
"284:196": {
"class_type": "ImageResizeKJv2",
"inputs": {
"crop_position": "center",
"device": "cpu",
"divisible_by": 2,
"height": [
"284:193",
1
],
"image": [
"284:76",
0
],
"keep_proportion": "stretch",
"pad_color": "0, 0, 0",
"upscale_method": "lanczos",
"width": [
"284:193",
0
]
},
"_meta": {
"title": "Resize Image v2"
}
},
"284:227": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "sdxl-skin_realism_acne_skin_details_imperfections.safetensors",
"model": [
"284:116",
0
],
"strength_model": 0.6
},
"_meta": {
"title": "Load LoRA"
}
},
"284:72": {
"class_type": "UpscaleModelLoader",
"inputs": {
"model_name": "8xNMKDFaces160000G_v10.pt"
},
"_meta": {
"title": "Load Upscale Model"
}
},
"284:76": {
"class_type": "UltimateSDUpscale",
"inputs": {
"cfg": 4,
"denoise": 0.2,
"force_uniform_tiles": True,
"image": [
"284:163",
0
],
"mask_blur": 32,
"mode_type": "Linear",
"model": [
"284:227",
0
],
"negative": [
"284:117",
0
],
"positive": [
"284:78",
0
],
"sampler_name": "dpmpp_2m_sde",
"scheduler": "karras",
"seam_fix_denoise": 1,
"seam_fix_mask_blur": 8,
"seam_fix_mode": "None",
"seam_fix_padding": 16,
"seam_fix_width": 64,
"seed": 900811593556971,
"steps": 35,
"tile_height": 1024,
"tile_padding": 32,
"tile_width": 1024,
"tiled_decode": False,
"upscale_by": [
"284:98",
0
],
"upscale_model": [
"284:72",
0
],
"vae": [
"284:116",
2
]
},
"_meta": {
"title": "Ultimate SD Upscale"
}
},
"284:78": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"284:116",
1
],
"text": "Close-up professional photograph of a woman, detailed face skin and visible skin pores, detailed natural skin and blemishes without-makeup, detailed skin texure, detailed lips, natural skin pores, subtle skin imperfections, visible skin pores, slightly cracked lips, skin discoloration, intricate detailed iris, perfect eyes, soft freckles"
},
"_meta": {
"title": "Positive Prompt"
}
},
"284:98": {
"class_type": "PrimitiveFloat",
"inputs": {
"value": 2
},
"_meta": {
"title": "Upscale By"
}
},
"285:11": {
"class_type": "ModelSamplingAuraFlow",
"inputs": {
"model": [
"285:287",
0
],
"sampling": "flow",
"shift": 3
},
"_meta": {
"title": "ModelSamplingAuraFlow"
}
},
"285:13": {
"class_type": "EmptySD3LatentImage",
"inputs": {
"batch_size": 1,
"height": 1152,
"width": 896
},
"_meta": {
"title": "EmptySD3LatentImage"
}
},
"285:287": {
"class_type": "LoraLoaderModelOnly",
"inputs": {
"lora_name": "zimage-grainscape_ultrareal.safetensors",
"model": [
"285:289",
0
],
"strength_model": 1
},
"_meta": {
"title": "Load LoRA"
}
},
"285:288": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "qwen_3_4b.safetensors",
"device": "default",
"type": "lumina2"
},
"_meta": {
"title": "Load CLIP"
}
},
"285:289": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "z_image_turbo_bf16.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"285:29": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "ae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"285:290": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"285:288",
0
],
"text": "A clean, high-key studio portrait features a central subject captured in a sharp headshot composition against a solid, vibrant cerulean blue backdrop. The lighting is soft and even, characteristic of a professional beauty dish or large softbox, which minimizes harsh shadows and creates a subtle, healthy glow across the skin. The focus is tack-sharp on the eyes, with a shallow depth of field that gently softens the texture of the hair as it recedes toward the shoulders.\n\nThe subject possesses a symmetrical, heart-shaped face with a neutral, poised expression and a direct gaze. Their hair is a bright, cool-toned platinum blonde, styled in a sleek, center-parted bob that falls just to the collarbone with slightly textured ends. Their complexion is smooth and radiant, complemented by a light application of makeup that emphasizes the natural arch of the well-defined, dark blonde eyebrows. The eyes are a warm hazel-brown, framed by dark, delicate lashes, while the lips are a soft, natural rose-pink with a slight satin finish.\n\nThe subject wears a crisp, minimalist white turtleneck sweater made of a fine-ribbed, lightweight knit fabric. The garment's clean lines and bright tone provide a high-contrast transition between the subject and the saturated blue background. The overall atmosphere is modern and clinical yet approachable, defined by the precision of the studio environment and the clarity of the facial features."
},
"_meta": {
"title": "CLIP Text Encode (Prompt)"
}
},
"285:3": {
"class_type": "KSampler",
"inputs": {
"cfg": 1,
"denoise": 1,
"latent_image": [
"285:13",
0
],
"model": [
"285:11",
0
],
"negative": [
"285:33",
0
],
"positive": [
"285:290",
0
],
"sampler_name": "res_multistep",
"scheduler": "simple",
"seed": 223832286985235,
"steps": 8
},
"_meta": {
"title": "KSampler"
}
},
"285:33": {
"class_type": "ConditioningZeroOut",
"inputs": {
"conditioning": [
"285:290",
0
]
},
"_meta": {
"title": "Conditioning Zero Out"
}
},
"285:8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"285:3",
0
],
"vae": [
"285:29",
0
]
},
"_meta": {
"title": "VAE Decode"
}
}
})
# run() also passes api_key to the partner nodes in this workflow.
job = client.run(workflow, api_key=api_key)Questions-réponses
Les portraits avec des traits du visage nets sont idéaux, car le workflow est optimisé pour l’amélioration et l’upscaling des détails du visage.
Oui, tant que le personnage possède des traits du visage à améliorer, le workflow peut être appliqué efficacement.
Des valeurs de débruitage plus élevées entraînent des modifications plus marquées du visage, permettant des améliorations plus prononcées.
Essayez d’ajuster les options de traitement et les paramètres de débruitage, ou affinez votre prompt pour mieux cibler le rendu souhaité.
Workflows associés
- ZZ-Image-Turbo
1.1 Débutant - Texte vers Image
- ZZ-Image-Turbo
2.1 Creator - Bases de la Diffusion
- ZZ-Image-Turbo
2.2 Creator - Diffusion Guidance
- ZZ-Image-Turbo
2.3 Creator : Upscaling et Analyse d’Image
- ZZ-Image-Turbo
Agrandir l'image : Z-Image-Turbo 2K
- Wan2.2, Z-Image-Turbo
AIrt MAchIne
- Google Gemini, Kling 3.0, Z-Image-Turbo
AIrt MAchIne (version API)
- ZZ-Image-Turbo
Changer de nœud
- ZZ-Image-Turbo
Génération par lots de prompts multiples
- Grok, Z-Image-Turbo
Générer des variations réalistes
- Kling 2.6, Nano Banana Pro, Z-Image-Turbo
image vers vidéo
- Grok, Z-Image-Turbo
Images réalistes 2K - Variations rapides