cjwbw/t2i-adapter – Run with an API on Replicate

cjwbw / t2i-adapter

Learning Adapters towards Controllable for Text-to-Image Diffusion Models

Cold

Public
3.9K runs
A100 (80GB)
GitHub
Paper
License

Iterate in playground

Run with an API

Playground API Examples README Versions

Input

prompt

string

Shift + Return to add a new line

Input prompt

Default: "A car with flying wings"

negative_prompt

string

Shift + Return to add a new line

ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted faceugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face

Specify things to not see in the output

Default: "ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face"

image

*file

Input image

model_checkpoint

string

Choose a model.

Default: "sd-v1-4"

type_in

string

Choose type of your input. When image is chosen, output will be the extracted sketch and the generated images.

Default: "image"

plms

boolean

Use plms sampling if set to True.

Default: true

dpm_solver

boolean

Use se dpm_solver sampling if set to True.

Default: false

width

integer

Width of output image. Lower the width if out of memory.

Default: 512

height

integer

Height of output image. Lower the height if out of memory.

Default: 512

num_outputs

integer

(minimum: 1, maximum: 4)

Number of images to output.

Default: 1

num_inference_steps

integer

(minimum: 1, maximum: 500)

Number of denoising steps

Default: 50

guidance_scale

number

(minimum: 1, maximum: 20)

Scale for classifier-free guidance

Default: 7.5

Run this model in Node.js with one line of code:

npx create-replicate --model=cjwbw/t2i-adapter

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run cjwbw/t2i-adapter using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "cjwbw/t2i-adapter:cf58e3279848046acb6fdaae3c48e79129b4a60eda8b1506bf58b264bdf58832",
  {
    input: {
      plms: true,
      image: "https://replicate.delivery/pbxt/IKi1ltMO3wmBYVEgffQDRJY6EBFJooB7By06MYiiojUDAAgk/car.png",
      width: 512,
      height: 512,
      prompt: "A car with flying wings",
      type_in: "sketch",
      dpm_solver: false,
      num_outputs: 1,
      guidance_scale: 7.5,
      negative_prompt: "ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face",
      model_checkpoint: "anything-v4.0-pruned",
      num_inference_steps: 50
    }
  }
);

// To access the file URL:
console.log(output[0].url()); //=> "http://example.com"

// To write the file to disk:
fs.writeFile("my-image.png", output[0]);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run cjwbw/t2i-adapter using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "cjwbw/t2i-adapter:cf58e3279848046acb6fdaae3c48e79129b4a60eda8b1506bf58b264bdf58832",
    input={
        "plms": True,
        "image": "https://replicate.delivery/pbxt/IKi1ltMO3wmBYVEgffQDRJY6EBFJooB7By06MYiiojUDAAgk/car.png",
        "width": 512,
        "height": 512,
        "prompt": "A car with flying wings",
        "type_in": "sketch",
        "dpm_solver": False,
        "num_outputs": 1,
        "guidance_scale": 7.5,
        "negative_prompt": "ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face",
        "model_checkpoint": "anything-v4.0-pruned",
        "num_inference_steps": 50
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run cjwbw/t2i-adapter using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

curl -s -X POST \
  -H "Authorization: Bearer $REPLICATE_API_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Prefer: wait" \
  -d $'{
    "version": "cjwbw/t2i-adapter:cf58e3279848046acb6fdaae3c48e79129b4a60eda8b1506bf58b264bdf58832",
    "input": {
      "plms": true,
      "image": "https://replicate.delivery/pbxt/IKi1ltMO3wmBYVEgffQDRJY6EBFJooB7By06MYiiojUDAAgk/car.png",
      "width": 512,
      "height": 512,
      "prompt": "A car with flying wings",
      "type_in": "sketch",
      "dpm_solver": false,
      "num_outputs": 1,
      "guidance_scale": 7.5,
      "negative_prompt": "ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face",
      "model_checkpoint": "anything-v4.0-pruned",
      "num_inference_steps": 50
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

Output

{
  "completed_at": "2023-02-18T03:13:25.528567Z",
  "created_at": "2023-02-18T03:13:18.837061Z",
  "data_removed": false,
  "error": null,
  "id": "5eoir46gdbcdppicbgwbinrxw4",
  "input": {
    "plms": true,
    "image": "https://replicate.delivery/pbxt/IKi1ltMO3wmBYVEgffQDRJY6EBFJooB7By06MYiiojUDAAgk/car.png",
    "width": 512,
    "height": 512,
    "prompt": "A car with flying wings",
    "type_in": "sketch",
    "num_outputs": "1",
    "guidance_scale": 7.5,
    "negative_prompt": "ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face",
    "model_checkpoint": "anything-v4.0-pruned",
    "num_inference_steps": 50
  },
  "logs": "Data shape for PLMS sampling is (1, 4, 64, 64)\nRunning PLMS Sampling with 50 timesteps\nPLMS Sampler:   0%|          | 0/50 [00:00<?, ?it/s]\nPLMS Sampler:   2%|▏         | 1/50 [00:00<00:12,  3.88it/s]\nPLMS Sampler:   4%|▍         | 2/50 [00:00<00:08,  5.75it/s]\nPLMS Sampler:   6%|▌         | 3/50 [00:00<00:06,  6.88it/s]\nPLMS Sampler:   8%|▊         | 4/50 [00:00<00:06,  7.52it/s]\nPLMS Sampler:  10%|█         | 5/50 [00:00<00:05,  7.99it/s]\nPLMS Sampler:  12%|█▏        | 6/50 [00:00<00:05,  8.30it/s]\nPLMS Sampler:  14%|█▍        | 7/50 [00:00<00:05,  8.52it/s]\nPLMS Sampler:  16%|█▌        | 8/50 [00:01<00:04,  8.67it/s]\nPLMS Sampler:  18%|█▊        | 9/50 [00:01<00:04,  8.78it/s]\nPLMS Sampler:  20%|██        | 10/50 [00:01<00:04,  8.64it/s]\nPLMS Sampler:  22%|██▏       | 11/50 [00:01<00:04,  8.74it/s]\nPLMS Sampler:  24%|██▍       | 12/50 [00:01<00:04,  8.83it/s]\nPLMS Sampler:  26%|██▌       | 13/50 [00:01<00:04,  8.79it/s]\nPLMS Sampler:  28%|██▊       | 14/50 [00:01<00:04,  8.85it/s]\nPLMS Sampler:  30%|███       | 15/50 [00:01<00:03,  8.91it/s]\nPLMS Sampler:  32%|███▏      | 16/50 [00:01<00:03,  8.95it/s]\nPLMS Sampler:  34%|███▍      | 17/50 [00:02<00:03,  8.98it/s]\nPLMS Sampler:  36%|███▌      | 18/50 [00:02<00:03,  9.00it/s]\nPLMS Sampler:  38%|███▊      | 19/50 [00:02<00:03,  9.01it/s]\nPLMS Sampler:  40%|████      | 20/50 [00:02<00:03,  8.91it/s]\nPLMS Sampler:  42%|████▏     | 21/50 [00:02<00:03,  8.97it/s]\nPLMS Sampler:  44%|████▍     | 22/50 [00:02<00:03,  9.00it/s]\nPLMS Sampler:  46%|████▌     | 23/50 [00:02<00:03,  8.96it/s]\nPLMS Sampler:  48%|████▊     | 24/50 [00:02<00:02,  8.99it/s]\nPLMS Sampler:  50%|█████     | 25/50 [00:02<00:02,  9.02it/s]\nPLMS Sampler:  52%|█████▏    | 26/50 [00:03<00:02,  9.04it/s]\nPLMS Sampler:  54%|█████▍    | 27/50 [00:03<00:02,  9.05it/s]\nPLMS Sampler:  56%|█████▌    | 28/50 [00:03<00:02,  9.04it/s]\nPLMS Sampler:  58%|█████▊    | 29/50 [00:03<00:02,  9.06it/s]\nPLMS Sampler:  60%|██████    | 30/50 [00:03<00:02,  9.07it/s]\nPLMS Sampler:  62%|██████▏   | 31/50 [00:03<00:02,  9.07it/s]\nPLMS Sampler:  64%|██████▍   | 32/50 [00:03<00:01,  9.07it/s]\nPLMS Sampler:  66%|██████▌   | 33/50 [00:03<00:01,  9.05it/s]\nPLMS Sampler:  68%|██████▊   | 34/50 [00:03<00:01,  9.00it/s]\nPLMS Sampler:  70%|███████   | 35/50 [00:04<00:01,  9.02it/s]\nPLMS Sampler:  72%|███████▏  | 36/50 [00:04<00:01,  9.04it/s]\nPLMS Sampler:  74%|███████▍  | 37/50 [00:04<00:01,  8.96it/s]\nPLMS Sampler:  76%|███████▌  | 38/50 [00:04<00:01,  8.91it/s]\nPLMS Sampler:  78%|███████▊  | 39/50 [00:04<00:01,  8.95it/s]\nPLMS Sampler:  80%|████████  | 40/50 [00:04<00:01,  9.00it/s]\nPLMS Sampler:  82%|████████▏ | 41/50 [00:04<00:00,  9.02it/s]\nPLMS Sampler:  84%|████████▍ | 42/50 [00:04<00:00,  9.05it/s]\nPLMS Sampler:  86%|████████▌ | 43/50 [00:04<00:00,  9.07it/s]\nPLMS Sampler:  88%|████████▊ | 44/50 [00:05<00:00,  9.09it/s]\nPLMS Sampler:  90%|█████████ | 45/50 [00:05<00:00,  9.09it/s]\nPLMS Sampler:  92%|█████████▏| 46/50 [00:05<00:00,  9.09it/s]\nPLMS Sampler:  94%|█████████▍| 47/50 [00:05<00:00,  8.97it/s]\nPLMS Sampler:  96%|█████████▌| 48/50 [00:05<00:00,  8.92it/s]\nPLMS Sampler:  98%|█████████▊| 49/50 [00:05<00:00,  8.99it/s]\nPLMS Sampler: 100%|██████████| 50/50 [00:05<00:00,  9.03it/s]\nPLMS Sampler: 100%|██████████| 50/50 [00:05<00:00,  8.75it/s]",
  "metrics": {
    "predict_time": 6.623156,
    "total_time": 6.691506
  },
  "output": [
    "https://replicate.delivery/pbxt/KlQ9UXTfyR3KPKw2497DmNwra9tr0FqTT7tfecr4gT6oie9BB/out_0.png"
  ],
  "started_at": "2023-02-18T03:13:18.905411Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/5eoir46gdbcdppicbgwbinrxw4",
    "cancel": "https://api.replicate.com/v1/predictions/5eoir46gdbcdppicbgwbinrxw4/cancel"
  },
  "version": "cf58e3279848046acb6fdaae3c48e79129b4a60eda8b1506bf58b264bdf58832"
}

Generated in

6.6 seconds

Tweak itReport

Data shape for PLMS sampling is (1, 4, 64, 64)
Running PLMS Sampling with 50 timesteps
PLMS Sampler:   0%|          | 0/50 [00:00<?, ?it/s]
PLMS Sampler:   2%|▏         | 1/50 [00:00<00:12,  3.88it/s]
PLMS Sampler:   4%|▍         | 2/50 [00:00<00:08,  5.75it/s]
PLMS Sampler:   6%|▌         | 3/50 [00:00<00:06,  6.88it/s]
PLMS Sampler:   8%|▊         | 4/50 [00:00<00:06,  7.52it/s]
PLMS Sampler:  10%|█         | 5/50 [00:00<00:05,  7.99it/s]
PLMS Sampler:  12%|█▏        | 6/50 [00:00<00:05,  8.30it/s]
PLMS Sampler:  14%|█▍        | 7/50 [00:00<00:05,  8.52it/s]
PLMS Sampler:  16%|█▌        | 8/50 [00:01<00:04,  8.67it/s]
PLMS Sampler:  18%|█▊        | 9/50 [00:01<00:04,  8.78it/s]
PLMS Sampler:  20%|██        | 10/50 [00:01<00:04,  8.64it/s]
PLMS Sampler:  22%|██▏       | 11/50 [00:01<00:04,  8.74it/s]
PLMS Sampler:  24%|██▍       | 12/50 [00:01<00:04,  8.83it/s]
PLMS Sampler:  26%|██▌       | 13/50 [00:01<00:04,  8.79it/s]
PLMS Sampler:  28%|██▊       | 14/50 [00:01<00:04,  8.85it/s]
PLMS Sampler:  30%|███       | 15/50 [00:01<00:03,  8.91it/s]
PLMS Sampler:  32%|███▏      | 16/50 [00:01<00:03,  8.95it/s]
PLMS Sampler:  34%|███▍      | 17/50 [00:02<00:03,  8.98it/s]
PLMS Sampler:  36%|███▌      | 18/50 [00:02<00:03,  9.00it/s]
PLMS Sampler:  38%|███▊      | 19/50 [00:02<00:03,  9.01it/s]
PLMS Sampler:  40%|████      | 20/50 [00:02<00:03,  8.91it/s]
PLMS Sampler:  42%|████▏     | 21/50 [00:02<00:03,  8.97it/s]
PLMS Sampler:  44%|████▍     | 22/50 [00:02<00:03,  9.00it/s]
PLMS Sampler:  46%|████▌     | 23/50 [00:02<00:03,  8.96it/s]
PLMS Sampler:  48%|████▊     | 24/50 [00:02<00:02,  8.99it/s]
PLMS Sampler:  50%|█████     | 25/50 [00:02<00:02,  9.02it/s]
PLMS Sampler:  52%|█████▏    | 26/50 [00:03<00:02,  9.04it/s]
PLMS Sampler:  54%|█████▍    | 27/50 [00:03<00:02,  9.05it/s]
PLMS Sampler:  56%|█████▌    | 28/50 [00:03<00:02,  9.04it/s]
PLMS Sampler:  58%|█████▊    | 29/50 [00:03<00:02,  9.06it/s]
PLMS Sampler:  60%|██████    | 30/50 [00:03<00:02,  9.07it/s]
PLMS Sampler:  62%|██████▏   | 31/50 [00:03<00:02,  9.07it/s]
PLMS Sampler:  64%|██████▍   | 32/50 [00:03<00:01,  9.07it/s]
PLMS Sampler:  66%|██████▌   | 33/50 [00:03<00:01,  9.05it/s]
PLMS Sampler:  68%|██████▊   | 34/50 [00:03<00:01,  9.00it/s]
PLMS Sampler:  70%|███████   | 35/50 [00:04<00:01,  9.02it/s]
PLMS Sampler:  72%|███████▏  | 36/50 [00:04<00:01,  9.04it/s]
PLMS Sampler:  74%|███████▍  | 37/50 [00:04<00:01,  8.96it/s]
PLMS Sampler:  76%|███████▌  | 38/50 [00:04<00:01,  8.91it/s]
PLMS Sampler:  78%|███████▊  | 39/50 [00:04<00:01,  8.95it/s]
PLMS Sampler:  80%|████████  | 40/50 [00:04<00:01,  9.00it/s]
PLMS Sampler:  82%|████████▏ | 41/50 [00:04<00:00,  9.02it/s]
PLMS Sampler:  84%|████████▍ | 42/50 [00:04<00:00,  9.05it/s]
PLMS Sampler:  86%|████████▌ | 43/50 [00:04<00:00,  9.07it/s]
PLMS Sampler:  88%|████████▊ | 44/50 [00:05<00:00,  9.09it/s]
PLMS Sampler:  90%|█████████ | 45/50 [00:05<00:00,  9.09it/s]
PLMS Sampler:  92%|█████████▏| 46/50 [00:05<00:00,  9.09it/s]
PLMS Sampler:  94%|█████████▍| 47/50 [00:05<00:00,  8.97it/s]
PLMS Sampler:  96%|█████████▌| 48/50 [00:05<00:00,  8.92it/s]
PLMS Sampler:  98%|█████████▊| 49/50 [00:05<00:00,  8.99it/s]
PLMS Sampler: 100%|██████████| 50/50 [00:05<00:00,  9.03it/s]
PLMS Sampler: 100%|██████████| 50/50 [00:05<00:00,  8.75it/s]

Examples

View more examples

Run time and cost

This model costs approximately $0.037 to run on Replicate, or 27 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia A100 (80GB) GPU hardware. Predictions typically complete within 27 seconds. The predict time for this model varies significantly based on the inputs.

Readme

Official implementation of T2I-Adapter: Learning Adapters to Dig out More Controllable Ability for Text-to-Image Diffusion Models.

We propose T2I-Adapter, a simple and small (~70M parameters, ~300M storage space) network that can provide extra guidance to pre-trained text-to-image models while freezing the original large text-to-image models.

T2I-Adapter aligns internal knowledge in T2I models with external control signals. We can train various adapters according to different conditions, and achieve rich control and editing effects.