Spaces:
Sleeping
Sleeping
improve: richer garment description + stronger prompt + 16 steps + 2MP ref- describe_garment now detects sleeve type, neckline, and special features (tie/bow, buttons, ruffles, etc.) using CLIP zero-shot — gives the model a much more specific caption to work from- Prompts now explicitly instruct the model to preserve sleeve shape, neckline, silhouette, and decorative elements without simplifying- Steps: 12 → 16 (better detail at ~25% more inference time)- Clothing reference: 1.5 → 2.0 megapixels (more detail in the conditioning)
e1b8d5e | { | |
| "76": { | |
| "inputs": { | |
| "image": "target.png" | |
| }, | |
| "class_type": "LoadImage", | |
| "_meta": { "title": "Load Target Image" } | |
| }, | |
| "81": { | |
| "inputs": { | |
| "image": "clothing.png" | |
| }, | |
| "class_type": "LoadImage", | |
| "_meta": { "title": "Load Clothing Reference Image" } | |
| }, | |
| "99": { | |
| "inputs": { | |
| "sampler_name": "euler" | |
| }, | |
| "class_type": "KSamplerSelect", | |
| "_meta": { "title": "KSamplerSelect" } | |
| }, | |
| "100": { | |
| "inputs": { | |
| "steps": 16, | |
| "width": ["115", 0], | |
| "height": ["115", 1] | |
| }, | |
| "class_type": "Flux2Scheduler", | |
| "_meta": { "title": "Flux2Scheduler" } | |
| }, | |
| "101": { | |
| "inputs": { | |
| "cfg": 1, | |
| "model": ["116", 0], | |
| "positive": ["112:139", 0], | |
| "negative": ["112:137", 0] | |
| }, | |
| "class_type": "CFGGuider", | |
| "_meta": { "title": "CFGGuider" } | |
| }, | |
| "102": { | |
| "inputs": { | |
| "noise": ["104", 0], | |
| "guider": ["101", 0], | |
| "sampler": ["99", 0], | |
| "sigmas": ["100", 0], | |
| "latent_image": ["120", 0] | |
| }, | |
| "class_type": "SamplerCustomAdvanced", | |
| "_meta": { "title": "SamplerCustomAdvanced" } | |
| }, | |
| "103": { | |
| "inputs": { | |
| "samples": ["102", 0], | |
| "vae": ["108", 0] | |
| }, | |
| "class_type": "VAEDecode", | |
| "_meta": { "title": "VAE Decode" } | |
| }, | |
| "104": { | |
| "inputs": { | |
| "noise_seed": 248575667574619 | |
| }, | |
| "class_type": "RandomNoise", | |
| "_meta": { "title": "RandomNoise" } | |
| }, | |
| "106": { | |
| "inputs": { | |
| "clip_name": "qwen_3_4b.safetensors", | |
| "type": "flux2", | |
| "device": "default" | |
| }, | |
| "class_type": "CLIPLoader", | |
| "_meta": { "title": "Load CLIP" } | |
| }, | |
| "107": { | |
| "inputs": { | |
| "text": "start with Picture 1 as the base image, keeping its lighting, environment, and background. remove the cloth from Picture 1 completely and replace it with the cloth from Picture 2, strictly preserving the cloth color, material and design of Picture 2. copy the cloth from Picture 2, match the pose from Picture 1 to subject's entire body, high quality, sharp details, 4k. Describe the expression in Picture 1 and copy it to the new image.", | |
| "clip": ["106", 0] | |
| }, | |
| "class_type": "CLIPTextEncode", | |
| "_meta": { "title": "Prompt" } | |
| }, | |
| "108": { | |
| "inputs": { | |
| "vae_name": "flux2-vae.safetensors" | |
| }, | |
| "class_type": "VAELoader", | |
| "_meta": { "title": "Load VAE" } | |
| }, | |
| "109": { | |
| "inputs": { | |
| "upscale_method": "nearest-exact", | |
| "megapixels": 1, | |
| "resolution_steps": 1, | |
| "image": ["76", 0] | |
| }, | |
| "class_type": "ImageScaleToTotalPixels", | |
| "_meta": { "title": "Scale Target Image" } | |
| }, | |
| "110": { | |
| "inputs": { | |
| "upscale_method": "lanczos", | |
| "megapixels": 2.0, | |
| "resolution_steps": 1, | |
| "image": ["135:133", 0] | |
| }, | |
| "class_type": "ImageScaleToTotalPixels", | |
| "_meta": { "title": "Scale Clothing Image" } | |
| }, | |
| "113": { | |
| "inputs": { | |
| "conditioning": ["107", 0] | |
| }, | |
| "class_type": "ConditioningZeroOut", | |
| "_meta": { "title": "ConditioningZeroOut" } | |
| }, | |
| "115": { | |
| "inputs": { | |
| "image": ["109", 0] | |
| }, | |
| "class_type": "GetImageSize", | |
| "_meta": { "title": "Get Image Size" } | |
| }, | |
| "116": { | |
| "inputs": { | |
| "unet_name": "flux-2-klein-4b-Q8_0.gguf" | |
| }, | |
| "class_type": "UnetLoaderGGUF", | |
| "_meta": { "title": "Unet Loader (GGUF)" } | |
| }, | |
| "120": { | |
| "inputs": { | |
| "width": 832, | |
| "height": 1216, | |
| "batch_size": 1 | |
| }, | |
| "class_type": "EmptyFlux2LatentImage", | |
| "_meta": { "title": "Empty Flux 2 Latent" } | |
| }, | |
| "123": { | |
| "inputs": { | |
| "lora_name": "aura-pvc-2-_00010e_074520s.safetensors", | |
| "strength_model": 1 | |
| }, | |
| "class_type": "LoraLoaderModelOnly", | |
| "_meta": { "title": "Load LoRA" } | |
| }, | |
| "124": { | |
| "inputs": { | |
| "lora_name": "aura-pvc-2-_00010e_074520s.safetensors", | |
| "strength_model": 1 | |
| }, | |
| "class_type": "LoraLoaderModelOnly", | |
| "_meta": { "title": "Load LoRA" } | |
| }, | |
| "128": { | |
| "inputs": { | |
| "precision": "auto", | |
| "compile": false | |
| }, | |
| "class_type": "LoadSAM3Model", | |
| "_meta": { "title": "SAM3 Model" } | |
| }, | |
| "140": { | |
| "inputs": { | |
| "filename_prefix": "cloth_swap", | |
| "images": ["103", 0] | |
| }, | |
| "class_type": "SaveImage", | |
| "_meta": { "title": "Save Result" } | |
| }, | |
| "135:133": { | |
| "inputs": { | |
| "force_resize_width": 0, | |
| "force_resize_height": 0, | |
| "image": ["135:131", 0], | |
| "mask": ["135:130", 0] | |
| }, | |
| "class_type": "Cut By Mask", | |
| "_meta": { "title": "Cut By Mask" } | |
| }, | |
| "135:132": { | |
| "inputs": { | |
| "mask": ["135:130", 1] | |
| }, | |
| "class_type": "InvertMask", | |
| "_meta": { "title": "Invert Mask" } | |
| }, | |
| "135:131": { | |
| "inputs": { | |
| "color": "255, 255, 255", | |
| "device": "cpu", | |
| "image": ["135:130", 0], | |
| "mask": ["135:132", 0] | |
| }, | |
| "class_type": "DrawMaskOnImage", | |
| "_meta": { "title": "Draw Mask On Image" } | |
| }, | |
| "135:129": { | |
| "inputs": { | |
| "confidence_threshold": 0.2, | |
| "text_prompt": "clothing, garment, outfit, shirt, pants, dress, jacket", | |
| "max_detections": -1, | |
| "sam3_model_config": ["128", 0], | |
| "image": ["81", 0] | |
| }, | |
| "class_type": "SAM3Grounding", | |
| "_meta": { "title": "SAM3 Text Segmentation" } | |
| }, | |
| "135:130": { | |
| "inputs": { | |
| "face": false, | |
| "hair": false, | |
| "body": false, | |
| "clothes": true, | |
| "accessories": false, | |
| "background": false, | |
| "confidence": 0.45, | |
| "detail_method": "VITMatte", | |
| "detail_erode": 6, | |
| "detail_dilate": 6, | |
| "black_point": 0.01, | |
| "white_point": 0.99, | |
| "process_detail": true, | |
| "device": "cpu", | |
| "max_megapixels": 2, | |
| "images": ["135:129", 1] | |
| }, | |
| "class_type": "LayerMask: PersonMaskUltra V2", | |
| "_meta": { "title": "LayerMask: PersonMaskUltra V2" } | |
| }, | |
| "111:136": { | |
| "inputs": { | |
| "conditioning": ["113", 0], | |
| "latent": ["111:78", 0] | |
| }, | |
| "class_type": "ReferenceLatent", | |
| "_meta": { "title": "ReferenceLatent" } | |
| }, | |
| "111:78": { | |
| "inputs": { | |
| "pixels": ["109", 0], | |
| "vae": ["108", 0] | |
| }, | |
| "class_type": "VAEEncode", | |
| "_meta": { "title": "VAE Encode" } | |
| }, | |
| "111:77": { | |
| "inputs": { | |
| "conditioning": ["107", 0], | |
| "latent": ["111:78", 0] | |
| }, | |
| "class_type": "ReferenceLatent", | |
| "_meta": { "title": "ReferenceLatent" } | |
| }, | |
| "112:137": { | |
| "inputs": { | |
| "conditioning": ["111:136", 0], | |
| "latent": ["112:138", 0] | |
| }, | |
| "class_type": "ReferenceLatent", | |
| "_meta": { "title": "ReferenceLatent" } | |
| }, | |
| "112:138": { | |
| "inputs": { | |
| "pixels": ["110", 0], | |
| "vae": ["108", 0] | |
| }, | |
| "class_type": "VAEEncode", | |
| "_meta": { "title": "VAE Encode" } | |
| }, | |
| "112:139": { | |
| "inputs": { | |
| "conditioning": ["111:77", 0], | |
| "latent": ["112:138", 0] | |
| }, | |
| "class_type": "ReferenceLatent", | |
| "_meta": { "title": "ReferenceLatent" } | |
| } | |
| } | |