cjwbw / dreambooth-avatar

Dreambooth finetuning of Stable Diffusion (v1.5.1) on Avatar art style by Lambda Labs

Cold

Public
585 runs
T4

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: ""

negative_prompt

string

Shift + Return to add a new line

Specify things to not see in the output

width

integer

Width of output image. Maximum size is 1024x768 or 768x1024 because of memory limits

Default: 512

height

integer

Height of output image. Maximum size is 1024x768 or 768x1024 because of memory limits

Default: 512

prompt_strength

number

Prompt strength when using init image. 1.0 corresponds to full destruction of information in init image

Default: 0.8

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

scheduler

string

Choose a scheduler.

Default: "DPMSolverMultistep"

seed

integer

Random seed. Leave blank to randomize the seed

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

npx create-replicate --model=cjwbw/dreambooth-avatar

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/dreambooth-avatar using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "cjwbw/dreambooth-avatar:381cc743d5880b92869ac05cafda83834931ae7b5ae3e627ce993dfd9b1194e1",
  {
    input: {
      width: 512,
      height: 512,
      prompt: "Yoda, avatarart style",
      scheduler: "DPMSolverMultistep",
      num_outputs: 1,
      guidance_scale: 7.5,
      prompt_strength: 0.8,
      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/dreambooth-avatar using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "cjwbw/dreambooth-avatar:381cc743d5880b92869ac05cafda83834931ae7b5ae3e627ce993dfd9b1194e1",
    input={
        "width": 512,
        "height": 512,
        "prompt": "Yoda, avatarart style",
        "scheduler": "DPMSolverMultistep",
        "num_outputs": 1,
        "guidance_scale": 7.5,
        "prompt_strength": 0.8,
        "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/dreambooth-avatar 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/dreambooth-avatar:381cc743d5880b92869ac05cafda83834931ae7b5ae3e627ce993dfd9b1194e1",
    "input": {
      "width": 512,
      "height": 512,
      "prompt": "Yoda, avatarart style",
      "scheduler": "DPMSolverMultistep",
      "num_outputs": 1,
      "guidance_scale": 7.5,
      "prompt_strength": 0.8,
      "num_inference_steps": 50
    }
  }' \
  https://api.replicate.com/v1/predictions

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

You can run this model locally using Cog. First, install Cog:

brew install cog

If you don’t have Homebrew, there are other installation options available.

Run this to download the model and run it in your local environment:

cog predict r8.im/chenxwh/dreambooth-avatar@sha256:381cc743d5880b92869ac05cafda83834931ae7b5ae3e627ce993dfd9b1194e1 \
  -i 'width=512' \
  -i 'height=512' \
  -i 'prompt="Yoda, avatarart style"' \
  -i 'scheduler="DPMSolverMultistep"' \
  -i 'num_outputs=1' \
  -i 'guidance_scale=7.5' \
  -i 'prompt_strength=0.8' \
  -i 'num_inference_steps=50'

To learn more, take a look at the Cog documentation.

Run this to download the model and run it in your local environment:

docker run -d -p 5000:5000 --gpus=all r8.im/chenxwh/dreambooth-avatar@sha256:381cc743d5880b92869ac05cafda83834931ae7b5ae3e627ce993dfd9b1194e1
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "width": 512,
      "height": 512,
      "prompt": "Yoda, avatarart style",
      "scheduler": "DPMSolverMultistep",
      "num_outputs": 1,
      "guidance_scale": 7.5,
      "prompt_strength": 0.8,
      "num_inference_steps": 50
    }
  }' \
  http://localhost:5000/predictions

To learn more, take a look at the Cog documentation.

Output

{
  "completed_at": "2023-02-07T14:06:12.965677Z",
  "created_at": "2023-02-07T14:05:48.776743Z",
  "data_removed": false,
  "error": null,
  "id": "v2f5nb2ea5gnlmny5da5qdus7e",
  "input": {
    "width": 512,
    "height": 512,
    "prompt": "Yoda, avatarart style",
    "scheduler": "DPMSolverMultistep",
    "num_outputs": 1,
    "guidance_scale": 7.5,
    "prompt_strength": 0.8,
    "num_inference_steps": 50
  },
  "logs": "Using seed: 52301\n  0%|          | 0/50 [00:00<?, ?it/s]\n  2%|▏         | 1/50 [00:00<00:24,  2.04it/s]\n  4%|▍         | 2/50 [00:00<00:22,  2.17it/s]\n  6%|▌         | 3/50 [00:01<00:21,  2.21it/s]\n  8%|▊         | 4/50 [00:01<00:20,  2.23it/s]\n 10%|█         | 5/50 [00:02<00:20,  2.24it/s]\n 12%|█▏        | 6/50 [00:02<00:19,  2.24it/s]\n 14%|█▍        | 7/50 [00:03<00:19,  2.24it/s]\n 16%|█▌        | 8/50 [00:03<00:18,  2.24it/s]\n 18%|█▊        | 9/50 [00:04<00:18,  2.24it/s]\n 20%|██        | 10/50 [00:04<00:17,  2.24it/s]\n 22%|██▏       | 11/50 [00:04<00:17,  2.23it/s]\n 24%|██▍       | 12/50 [00:05<00:17,  2.23it/s]\n 26%|██▌       | 13/50 [00:05<00:16,  2.23it/s]\n 28%|██▊       | 14/50 [00:06<00:16,  2.23it/s]\n 30%|███       | 15/50 [00:06<00:15,  2.21it/s]\n 32%|███▏      | 16/50 [00:07<00:15,  2.22it/s]\n 34%|███▍      | 17/50 [00:07<00:14,  2.22it/s]\n 36%|███▌      | 18/50 [00:08<00:14,  2.22it/s]\n 38%|███▊      | 19/50 [00:08<00:13,  2.22it/s]\n 40%|████      | 20/50 [00:08<00:13,  2.22it/s]\n 42%|████▏     | 21/50 [00:09<00:13,  2.22it/s]\n 44%|████▍     | 22/50 [00:09<00:12,  2.21it/s]\n 46%|████▌     | 23/50 [00:10<00:12,  2.21it/s]\n 48%|████▊     | 24/50 [00:10<00:11,  2.22it/s]\n 50%|█████     | 25/50 [00:11<00:11,  2.21it/s]\n 52%|█████▏    | 26/50 [00:11<00:10,  2.21it/s]\n 54%|█████▍    | 27/50 [00:12<00:10,  2.21it/s]\n 56%|█████▌    | 28/50 [00:12<00:09,  2.20it/s]\n 58%|█████▊    | 29/50 [00:13<00:09,  2.20it/s]\n 60%|██████    | 30/50 [00:13<00:09,  2.20it/s]\n 62%|██████▏   | 31/50 [00:13<00:08,  2.20it/s]\n 64%|██████▍   | 32/50 [00:14<00:08,  2.19it/s]\n 66%|██████▌   | 33/50 [00:14<00:07,  2.19it/s]\n 68%|██████▊   | 34/50 [00:15<00:07,  2.19it/s]\n 70%|███████   | 35/50 [00:15<00:06,  2.19it/s]\n 72%|███████▏  | 36/50 [00:16<00:06,  2.19it/s]\n 74%|███████▍  | 37/50 [00:16<00:05,  2.19it/s]\n 76%|███████▌  | 38/50 [00:17<00:05,  2.19it/s]\n 78%|███████▊  | 39/50 [00:17<00:05,  2.19it/s]\n 80%|████████  | 40/50 [00:18<00:04,  2.19it/s]\n 82%|████████▏ | 41/50 [00:18<00:04,  2.18it/s]\n 84%|████████▍ | 42/50 [00:19<00:03,  2.18it/s]\n 86%|████████▌ | 43/50 [00:19<00:03,  2.18it/s]\n 88%|████████▊ | 44/50 [00:19<00:02,  2.18it/s]\n 90%|█████████ | 45/50 [00:20<00:02,  2.18it/s]\n 92%|█████████▏| 46/50 [00:20<00:01,  2.18it/s]\n 94%|█████████▍| 47/50 [00:21<00:01,  2.17it/s]\n 96%|█████████▌| 48/50 [00:21<00:00,  2.17it/s]\n 98%|█████████▊| 49/50 [00:22<00:00,  2.17it/s]\n100%|██████████| 50/50 [00:22<00:00,  2.16it/s]\n100%|██████████| 50/50 [00:22<00:00,  2.20it/s]",
  "metrics": {
    "predict_time": 24.123844,
    "total_time": 24.188934
  },
  "output": [
    "https://replicate.delivery/pbxt/maedOFyCeapXeIGbiQmDKZxbRgRxGw1hxXkeuhUXoHnTNDwBB/out-0.png"
  ],
  "started_at": "2023-02-07T14:05:48.841833Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/v2f5nb2ea5gnlmny5da5qdus7e",
    "cancel": "https://api.replicate.com/v1/predictions/v2f5nb2ea5gnlmny5da5qdus7e/cancel"
  },
  "version": "381cc743d5880b92869ac05cafda83834931ae7b5ae3e627ce993dfd9b1194e1"
}

Generated in

24.1 seconds

Tweak itReport

Using seed: 52301
  0%|          | 0/50 [00:00<?, ?it/s]
  2%|▏         | 1/50 [00:00<00:24,  2.04it/s]
  4%|▍         | 2/50 [00:00<00:22,  2.17it/s]
  6%|▌         | 3/50 [00:01<00:21,  2.21it/s]
  8%|▊         | 4/50 [00:01<00:20,  2.23it/s]
 10%|█         | 5/50 [00:02<00:20,  2.24it/s]
 12%|█▏        | 6/50 [00:02<00:19,  2.24it/s]
 14%|█▍        | 7/50 [00:03<00:19,  2.24it/s]
 16%|█▌        | 8/50 [00:03<00:18,  2.24it/s]
 18%|█▊        | 9/50 [00:04<00:18,  2.24it/s]
 20%|██        | 10/50 [00:04<00:17,  2.24it/s]
 22%|██▏       | 11/50 [00:04<00:17,  2.23it/s]
 24%|██▍       | 12/50 [00:05<00:17,  2.23it/s]
 26%|██▌       | 13/50 [00:05<00:16,  2.23it/s]
 28%|██▊       | 14/50 [00:06<00:16,  2.23it/s]
 30%|███       | 15/50 [00:06<00:15,  2.21it/s]
 32%|███▏      | 16/50 [00:07<00:15,  2.22it/s]
 34%|███▍      | 17/50 [00:07<00:14,  2.22it/s]
 36%|███▌      | 18/50 [00:08<00:14,  2.22it/s]
 38%|███▊      | 19/50 [00:08<00:13,  2.22it/s]
 40%|████      | 20/50 [00:08<00:13,  2.22it/s]
 42%|████▏     | 21/50 [00:09<00:13,  2.22it/s]
 44%|████▍     | 22/50 [00:09<00:12,  2.21it/s]
 46%|████▌     | 23/50 [00:10<00:12,  2.21it/s]
 48%|████▊     | 24/50 [00:10<00:11,  2.22it/s]
 50%|█████     | 25/50 [00:11<00:11,  2.21it/s]
 52%|█████▏    | 26/50 [00:11<00:10,  2.21it/s]
 54%|█████▍    | 27/50 [00:12<00:10,  2.21it/s]
 56%|█████▌    | 28/50 [00:12<00:09,  2.20it/s]
 58%|█████▊    | 29/50 [00:13<00:09,  2.20it/s]
 60%|██████    | 30/50 [00:13<00:09,  2.20it/s]
 62%|██████▏   | 31/50 [00:13<00:08,  2.20it/s]
 64%|██████▍   | 32/50 [00:14<00:08,  2.19it/s]
 66%|██████▌   | 33/50 [00:14<00:07,  2.19it/s]
 68%|██████▊   | 34/50 [00:15<00:07,  2.19it/s]
 70%|███████   | 35/50 [00:15<00:06,  2.19it/s]
 72%|███████▏  | 36/50 [00:16<00:06,  2.19it/s]
 74%|███████▍  | 37/50 [00:16<00:05,  2.19it/s]
 76%|███████▌  | 38/50 [00:17<00:05,  2.19it/s]
 78%|███████▊  | 39/50 [00:17<00:05,  2.19it/s]
 80%|████████  | 40/50 [00:18<00:04,  2.19it/s]
 82%|████████▏ | 41/50 [00:18<00:04,  2.18it/s]
 84%|████████▍ | 42/50 [00:19<00:03,  2.18it/s]
 86%|████████▌ | 43/50 [00:19<00:03,  2.18it/s]
 88%|████████▊ | 44/50 [00:19<00:02,  2.18it/s]
 90%|█████████ | 45/50 [00:20<00:02,  2.18it/s]
 92%|█████████▏| 46/50 [00:20<00:01,  2.18it/s]
 94%|█████████▍| 47/50 [00:21<00:01,  2.17it/s]
 96%|█████████▌| 48/50 [00:21<00:00,  2.17it/s]
 98%|█████████▊| 49/50 [00:22<00:00,  2.17it/s]
100%|██████████| 50/50 [00:22<00:00,  2.16it/s]
100%|██████████| 50/50 [00:22<00:00,  2.20it/s]

Examples

View more examples

Run time and cost

This model costs approximately $0.035 to run on Replicate, or 28 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 T4 GPU hardware. Predictions typically complete within 3 minutes. The predict time for this model varies significantly based on the inputs.

Readme

weights: https://huggingface.co/lambdalabs/dreambooth-avatar

Dreambooth style: Avatar

Dreambooth finetuning of Stable Diffusion (v1.5.1) on Avatar art style by Lambda Labs.

About

This text-to-image stable diffusion model was trained with dreambooth.
Put in a text prompt and generate your own Avatar style image!

Model description

Base model is Stable Diffusion v1.5 and was trained using Dreambooth with 60 input images sized 512x512 displaying Avatar character images. The model is learning to associate Avatar images with the style tokenized as ‘avatarart style’. Prior preservation was used during training using the class ‘Person’ to avoid training bleeding into the representations for that class. Training ran on 2xA6000 GPUs on Lambda GPU Cloud for 700 steps, batch size 4 (a couple hours, at a cost of about $4).

Author: Eole Cervenka