👗How to Create a Clothing Restyler App (IP Adapter, ControlNet + GPT Vision)

In this video we show you how to create a workflow that allows you to design new fashion trends in the blink of an eye! Using a combination of Controlnet, IP Adapter and GPT Vision, this mini-app can

Check out the example glif here!

Step by step

1. Click Build to start a new glif project.

2. Add two Image Input Blocks.

3. Label one "Original Image" and the other "Style Image".

4. Add a Multipick Input Block.

5. List the integers 0.0 to 1.0 incrementally by 0.1 each step in the previous block.

6. Add a Glif Block.

7. Switch the Glif Block to GPT Vision.

8. Add the parameters outlined below.

9. Add an LLM Block and under "Advanced", select Claude 3.5 Sonnet.

10. Adjust the LLM to the below parameters.

11. (Optional) Add Several Text Combine Blocks to set values for the IPA Strength (1), Mask Blur (30), and Target (outfit).

12. (Optional) Add a Text Combine Block for the prompt: bizarre fashion photograph of person wearing [style description], shot on film.

13. Add a ComfyUI Block.

14. Input below JSON into ComfyUI Block.

15. Publish the glif!

GPT Vision Input

Text Prompt

You're a fashion designer, taking inspiration from all sorts of images. Today you want to create a unisex outfit inspired by this image. Write a brief description of the outfit, focus on the style, theme, colors and fabrics and materials (bizarre is good!). This is for a high end weirdo fashion brand, so any and all materials are possible, you can even go sci-fi. Just go, write a 6-7 word sentence, no more, no yapping:

Image URL

[Style Image]

maxTokens

100

Claude 3.5 Input

Prompt

Name a fashion trend based on an outfit that is the following: [gpt vision] Only return one or two words, no intro, just go, write it in ALL CAPS - make it very unique, not some lame existing genre (no "Cyberpunk"), it needs to super leftfield and weird!

Max Tokens

50

System Prompt

You're a fashion editor, picking up on the most insane and cool trends and subcultures. You name trends, often ending in -core.

ComfyUI JSON

{
  "38": {
    "inputs": {
      "model_name": "sam_vit_h (2.56GB)"
    },
    "class_type": "SAMModelLoader (segment anything)",
    "_meta": {
      "title": "SAMModelLoader (segment anything)"
    }
  },
  "40": {
    "inputs": {
      "image": "{image-input1}"
    },
    "class_type": "LoadImage",
    "_meta": {
      "title": "Load Image"
    }
  },
  "41": {
    "inputs": {
      "model_name": "GroundingDINO_SwinT_OGC (694MB)"
    },
    "class_type": "GroundingDinoModelLoader (segment anything)",
    "_meta": {
      "title": "GroundingDinoModelLoader (segment anything)"
    }
  },
  "42": {
    "inputs": {
      "prompt": "{target}",
      "threshold": 0.5,
      "sam_model": [
        "38",
        0
      ],
      "grounding_dino_model": [
        "41",
        0
      ],
      "image": [
        "40",
        0
      ]
    },
    "class_type": "GroundingDinoSAMSegment (segment anything)",
    "_meta": {
      "title": "GroundingDinoSAMSegment (segment anything)"
    }
  },
  "47": {
    "inputs": {
      "samples": [
        "52",
        0
      ],
      "vae": [
        "53",
        2
      ]
    },
    "class_type": "VAEDecode",
    "_meta": {
      "title": "VAE Decode"
    }
  },
  "52": {
    "inputs": {
      "seed": 452183510506570,
      "steps": 27,
      "cfg": 5,
      "sampler_name": "euler_ancestral",
      "scheduler": "normal",
      "denoise": 1,
      "model": [
        "73",
        0
      ],
      "positive": [
        "80",
        0
      ],
      "negative": [
        "74",
        1
      ],
      "latent_image": [
        "74",
        2
      ]
    },
    "class_type": "KSampler",
    "_meta": {
      "title": "KSampler"
    }
  },
  "53": {
    "inputs": {
      "ckpt_name": "Juggernaut_X_RunDiffusion.safetensors"
    },
    "class_type": "CheckpointLoaderSimple",
    "_meta": {
      "title": "Load Checkpoint"
    }
  },
  "54": {
    "inputs": {
      "text": "{prompt}",
      "clip": [
        "53",
        1
      ]
    },
    "class_type": "CLIPTextEncode",
    "_meta": {
      "title": "CLIP Text Encode (Prompt)"
    }
  },
  "55": {
    "inputs": {
      "text": "watermark, blurry, bad quality, illustration, pixelated, boring, ugly, wrong, tiled",
      "clip": [
        "53",
        1
      ]
    },
    "class_type": "CLIPTextEncode",
    "_meta": {
      "title": "CLIP Text Encode (Prompt)"
    }
  },
  "67": {
    "inputs": {
      "expand": [
        "88",
        1
      ],
      "incremental_expandrate": 0,
      "tapered_corners": true,
      "flip_input": false,
      "blur_radius": 15,
      "lerp_alpha": 1,
      "decay_factor": 1,
      "fill_holes": false,
      "mask": [
        "42",
        1
      ]
    },
    "class_type": "GrowMaskWithBlur",
    "_meta": {
      "title": "Grow Mask With Blur"
    }
  },
  "72": {
    "inputs": {
      "filename_prefix": "ComfyUI",
      "images": [
        "47",
        0
      ]
    },
    "class_type": "SaveImage",
    "_meta": {
      "title": "Save Image"
    }
  },
  "73": {
    "inputs": {
      "model": [
        "90",
        0
      ]
    },
    "class_type": "DifferentialDiffusion",
    "_meta": {
      "title": "Differential Diffusion"
    }
  },
  "74": {
    "inputs": {
      "positive": [
        "54",
        0
      ],
      "negative": [
        "55",
        0
      ],
      "vae": [
        "53",
        2
      ],
      "pixels": [
        "40",
        0
      ],
      "mask": [
        "67",
        0
      ]
    },
    "class_type": "InpaintModelConditioning",
    "_meta": {
      "title": "InpaintModelConditioning"
    }
  },
  "76": {
    "inputs": {
      "preset": "PLUS (high strength)",
      "model": [
        "53",
        0
      ]
    },
    "class_type": "IPAdapterUnifiedLoader",
    "_meta": {
      "title": "IPAdapter Unified Loader"
    }
  },
  "77": {
    "inputs": {
      "clip_name": "CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors"
    },
    "class_type": "CLIPVisionLoader",
    "_meta": {
      "title": "Load CLIP Vision"
    }
  },
  "78": {
    "inputs": {
      "variable": "{IPAstrength}",
      "fallback": ""
    },
    "class_type": "GlifVariable",
    "_meta": {
      "title": "Glif Variable"
    }
  },
  "79": {
    "inputs": {
      "control_net_name": "mistoLine_fp16.safetensors"
    },
    "class_type": "ControlNetLoader",
    "_meta": {
      "title": "Load ControlNet Model"
    }
  },
  "80": {
    "inputs": {
      "strength": [
        "82",
        2
      ],
      "conditioning": [
        "74",
        0
      ],
      "control_net": [
        "79",
        0
      ],
      "image": [
        "81",
        0
      ]
    },
    "class_type": "ControlNetApply",
    "_meta": {
      "title": "Apply ControlNet"
    }
  },
  "81": {
    "inputs": {
      "image": [
        "40",
        0
      ]
    },
    "class_type": "AnyLinePreprocessor",
    "_meta": {
      "title": "TheMisto.ai Anyline"
    }
  },
  "82": {
    "inputs": {
      "variable": "{cn_strength}",
      "fallback": ""
    },
    "class_type": "GlifVariable",
    "_meta": {
      "title": "Glif Variable"
    }
  },
  "83": {
    "inputs": {
      "image": "{image-input2}"
    },
    "class_type": "LoadImage",
    "_meta": {
      "title": "Load Image"
    }
  },
  "88": {
    "inputs": {
      "variable": "{maskblur}",
      "fallback": "10"
    },
    "class_type": "GlifVariable",
    "_meta": {
      "title": "Glif Variable"
    }
  },
  "90": {
    "inputs": {
      "weight_style": [
        "78",
        2
      ],
      "weight_composition": 0,
      "expand_style": false,
      "combine_embeds": "average",
      "start_at": 0,
      "end_at": 1,
      "embeds_scaling": "V only",
      "model": [
        "76",
        0
      ],
      "ipadapter": [
        "76",
        1
      ],
      "image_style": [
        "83",
        0
      ],
      "image_composition": [
        "83",
        0
      ],
      "clip_vision": [
        "77",
        0
      ]
    },
    "class_type": "IPAdapterStyleComposition",
    "_meta": {
      "title": "IPAdapter Style & Composition SDXL"
    }
  }
}