22-hours/vintedois-diffusion – Run with an API on Replicate

22-hours / vintedois-diffusion

Generate beautiful images with simple prompts

Cold

Public
247.2K runs
T4

Run with an API

Playground API Examples README Versions

Input

prompt

string

Shift + Return to add a new line

Input prompt

Default: "a photo of an astronaut riding a horse on mars"

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: 640

height

integer

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

Default: 640

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: "K_EULER_ANCESTRAL"

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=22-hours/vintedois-diffusion

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 22-hours/vintedois-diffusion using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "22-hours/vintedois-diffusion:28cea91bdfced0e2dc7fda466cc0a46501c0edc84905b2120ea02e0707b967fd",
  {
    input: {
      seed: 44,
      width: 640,
      height: 640,
      prompt: "victorian city landscape",
      scheduler: "K_EULER_ANCESTRAL",
      num_outputs: 1,
      guidance_scale: 7.5,
      prompt_strength: 1,
      num_inference_steps: 50
    }
  }
);
console.log(output);

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 22-hours/vintedois-diffusion using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "22-hours/vintedois-diffusion:28cea91bdfced0e2dc7fda466cc0a46501c0edc84905b2120ea02e0707b967fd",
    input={
        "seed": 44,
        "width": 640,
        "height": 640,
        "prompt": "victorian city landscape",
        "scheduler": "K_EULER_ANCESTRAL",
        "num_outputs": 1,
        "guidance_scale": 7.5,
        "prompt_strength": 1,
        "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 22-hours/vintedois-diffusion 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": "28cea91bdfced0e2dc7fda466cc0a46501c0edc84905b2120ea02e0707b967fd",
    "input": {
      "seed": 44,
      "width": 640,
      "height": 640,
      "prompt": "victorian city landscape",
      "scheduler": "K_EULER_ANCESTRAL",
      "num_outputs": 1,
      "guidance_scale": 7.5,
      "prompt_strength": 1,
      "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/22-hours/vintedois-diffusion@sha256:28cea91bdfced0e2dc7fda466cc0a46501c0edc84905b2120ea02e0707b967fd \
  -i 'seed=44' \
  -i 'width=640' \
  -i 'height=640' \
  -i 'prompt="victorian city landscape"' \
  -i 'scheduler="K_EULER_ANCESTRAL"' \
  -i 'num_outputs=1' \
  -i 'guidance_scale=7.5' \
  -i 'prompt_strength=1' \
  -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/22-hours/vintedois-diffusion@sha256:28cea91bdfced0e2dc7fda466cc0a46501c0edc84905b2120ea02e0707b967fd
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "seed": 44,
      "width": 640,
      "height": 640,
      "prompt": "victorian city landscape",
      "scheduler": "K_EULER_ANCESTRAL",
      "num_outputs": 1,
      "guidance_scale": 7.5,
      "prompt_strength": 1,
      "num_inference_steps": 50
    }
  }' \
  http://localhost:5000/predictions

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

Output

{
  "completed_at": "2023-02-08T16:52:47.560905Z",
  "created_at": "2023-02-08T16:51:57.049366Z",
  "data_removed": false,
  "error": null,
  "id": "7zm7kijennfzvjyqisdiag5x4m",
  "input": {
    "seed": 44,
    "width": 640,
    "height": 640,
    "prompt": "victorian city landscape",
    "scheduler": "K_EULER_ANCESTRAL",
    "num_outputs": 1,
    "guidance_scale": 7.5,
    "prompt_strength": 1,
    "num_inference_steps": "50"
  },
  "logs": "Using seed: 44\n  0%|          | 0/50 [00:00<?, ?it/s]\n  2%|▏         | 1/50 [00:00<00:48,  1.01it/s]\n  4%|▍         | 2/50 [00:01<00:46,  1.03it/s]\n  6%|▌         | 3/50 [00:02<00:45,  1.03it/s]\n  8%|▊         | 4/50 [00:03<00:44,  1.04it/s]\n 10%|█         | 5/50 [00:04<00:43,  1.04it/s]\n 12%|█▏        | 6/50 [00:05<00:42,  1.04it/s]\n 14%|█▍        | 7/50 [00:06<00:41,  1.04it/s]\n 16%|█▌        | 8/50 [00:07<00:40,  1.04it/s]\n 18%|█▊        | 9/50 [00:08<00:39,  1.04it/s]\n 20%|██        | 10/50 [00:09<00:38,  1.04it/s]\n 22%|██▏       | 11/50 [00:10<00:37,  1.04it/s]\n 24%|██▍       | 12/50 [00:11<00:36,  1.04it/s]\n 26%|██▌       | 13/50 [00:12<00:35,  1.04it/s]\n 28%|██▊       | 14/50 [00:13<00:34,  1.04it/s]\n 30%|███       | 15/50 [00:14<00:33,  1.03it/s]\n 32%|███▏      | 16/50 [00:15<00:32,  1.03it/s]\n 34%|███▍      | 17/50 [00:16<00:31,  1.03it/s]\n 36%|███▌      | 18/50 [00:17<00:31,  1.03it/s]\n 38%|███▊      | 19/50 [00:18<00:30,  1.03it/s]\n 40%|████      | 20/50 [00:19<00:29,  1.03it/s]\n 42%|████▏     | 21/50 [00:20<00:28,  1.03it/s]\n 44%|████▍     | 22/50 [00:21<00:27,  1.03it/s]\n 46%|████▌     | 23/50 [00:22<00:26,  1.03it/s]\n 48%|████▊     | 24/50 [00:23<00:25,  1.03it/s]\n 50%|█████     | 25/50 [00:24<00:24,  1.03it/s]\n 52%|█████▏    | 26/50 [00:25<00:23,  1.03it/s]\n 54%|█████▍    | 27/50 [00:26<00:22,  1.03it/s]\n 56%|█████▌    | 28/50 [00:27<00:21,  1.03it/s]\n 58%|█████▊    | 29/50 [00:28<00:20,  1.03it/s]\n 60%|██████    | 30/50 [00:29<00:19,  1.03it/s]\n 62%|██████▏   | 31/50 [00:30<00:18,  1.03it/s]\n 64%|██████▍   | 32/50 [00:31<00:17,  1.03it/s]\n 66%|██████▌   | 33/50 [00:31<00:16,  1.03it/s]\n 68%|██████▊   | 34/50 [00:32<00:15,  1.03it/s]\n 70%|███████   | 35/50 [00:33<00:14,  1.03it/s]\n 72%|███████▏  | 36/50 [00:34<00:13,  1.03it/s]\n 74%|███████▍  | 37/50 [00:35<00:12,  1.03it/s]\n 76%|███████▌  | 38/50 [00:36<00:11,  1.03it/s]\n 78%|███████▊  | 39/50 [00:37<00:10,  1.02it/s]\n 80%|████████  | 40/50 [00:38<00:09,  1.02it/s]\n 82%|████████▏ | 41/50 [00:39<00:08,  1.02it/s]\n 84%|████████▍ | 42/50 [00:40<00:07,  1.02it/s]\n 86%|████████▌ | 43/50 [00:41<00:06,  1.02it/s]\n 88%|████████▊ | 44/50 [00:42<00:05,  1.02it/s]\n 90%|█████████ | 45/50 [00:43<00:04,  1.02it/s]\n 92%|█████████▏| 46/50 [00:44<00:03,  1.02it/s]\n 94%|█████████▍| 47/50 [00:45<00:02,  1.02it/s]\n 96%|█████████▌| 48/50 [00:46<00:01,  1.02it/s]\n 98%|█████████▊| 49/50 [00:47<00:00,  1.02it/s]\n100%|██████████| 50/50 [00:48<00:00,  1.02it/s]\n100%|██████████| 50/50 [00:48<00:00,  1.03it/s]",
  "metrics": {
    "predict_time": 50.451546,
    "total_time": 50.511539
  },
  "output": [
    "https://replicate.delivery/pbxt/j0ALfNdwwfvXG0wLdr69LlpW8TnbuH9M11xZMMWu4GGeqw4gA/out-0.png"
  ],
  "started_at": "2023-02-08T16:51:57.109359Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/7zm7kijennfzvjyqisdiag5x4m",
    "cancel": "https://api.replicate.com/v1/predictions/7zm7kijennfzvjyqisdiag5x4m/cancel"
  },
  "version": "28cea91bdfced0e2dc7fda466cc0a46501c0edc84905b2120ea02e0707b967fd"
}

Generated in

50.5 seconds

Tweak itReport

Using seed: 44
  0%|          | 0/50 [00:00<?, ?it/s]
  2%|▏         | 1/50 [00:00<00:48,  1.01it/s]
  4%|▍         | 2/50 [00:01<00:46,  1.03it/s]
  6%|▌         | 3/50 [00:02<00:45,  1.03it/s]
  8%|▊         | 4/50 [00:03<00:44,  1.04it/s]
 10%|█         | 5/50 [00:04<00:43,  1.04it/s]
 12%|█▏        | 6/50 [00:05<00:42,  1.04it/s]
 14%|█▍        | 7/50 [00:06<00:41,  1.04it/s]
 16%|█▌        | 8/50 [00:07<00:40,  1.04it/s]
 18%|█▊        | 9/50 [00:08<00:39,  1.04it/s]
 20%|██        | 10/50 [00:09<00:38,  1.04it/s]
 22%|██▏       | 11/50 [00:10<00:37,  1.04it/s]
 24%|██▍       | 12/50 [00:11<00:36,  1.04it/s]
 26%|██▌       | 13/50 [00:12<00:35,  1.04it/s]
 28%|██▊       | 14/50 [00:13<00:34,  1.04it/s]
 30%|███       | 15/50 [00:14<00:33,  1.03it/s]
 32%|███▏      | 16/50 [00:15<00:32,  1.03it/s]
 34%|███▍      | 17/50 [00:16<00:31,  1.03it/s]
 36%|███▌      | 18/50 [00:17<00:31,  1.03it/s]
 38%|███▊      | 19/50 [00:18<00:30,  1.03it/s]
 40%|████      | 20/50 [00:19<00:29,  1.03it/s]
 42%|████▏     | 21/50 [00:20<00:28,  1.03it/s]
 44%|████▍     | 22/50 [00:21<00:27,  1.03it/s]
 46%|████▌     | 23/50 [00:22<00:26,  1.03it/s]
 48%|████▊     | 24/50 [00:23<00:25,  1.03it/s]
 50%|█████     | 25/50 [00:24<00:24,  1.03it/s]
 52%|█████▏    | 26/50 [00:25<00:23,  1.03it/s]
 54%|█████▍    | 27/50 [00:26<00:22,  1.03it/s]
 56%|█████▌    | 28/50 [00:27<00:21,  1.03it/s]
 58%|█████▊    | 29/50 [00:28<00:20,  1.03it/s]
 60%|██████    | 30/50 [00:29<00:19,  1.03it/s]
 62%|██████▏   | 31/50 [00:30<00:18,  1.03it/s]
 64%|██████▍   | 32/50 [00:31<00:17,  1.03it/s]
 66%|██████▌   | 33/50 [00:31<00:16,  1.03it/s]
 68%|██████▊   | 34/50 [00:32<00:15,  1.03it/s]
 70%|███████   | 35/50 [00:33<00:14,  1.03it/s]
 72%|███████▏  | 36/50 [00:34<00:13,  1.03it/s]
 74%|███████▍  | 37/50 [00:35<00:12,  1.03it/s]
 76%|███████▌  | 38/50 [00:36<00:11,  1.03it/s]
 78%|███████▊  | 39/50 [00:37<00:10,  1.02it/s]
 80%|████████  | 40/50 [00:38<00:09,  1.02it/s]
 82%|████████▏ | 41/50 [00:39<00:08,  1.02it/s]
 84%|████████▍ | 42/50 [00:40<00:07,  1.02it/s]
 86%|████████▌ | 43/50 [00:41<00:06,  1.02it/s]
 88%|████████▊ | 44/50 [00:42<00:05,  1.02it/s]
 90%|█████████ | 45/50 [00:43<00:04,  1.02it/s]
 92%|█████████▏| 46/50 [00:44<00:03,  1.02it/s]
 94%|█████████▍| 47/50 [00:45<00:02,  1.02it/s]
 96%|█████████▌| 48/50 [00:46<00:01,  1.02it/s]
 98%|█████████▊| 49/50 [00:47<00:00,  1.02it/s]
100%|██████████| 50/50 [00:48<00:00,  1.02it/s]
100%|██████████| 50/50 [00:48<00:00,  1.03it/s]

Examples

View more examples

Run time and cost

This model costs approximately $0.031 to run on Replicate, or 32 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 140 seconds. The predict time for this model varies significantly based on the inputs.

Readme

This model was trained on a large amount of high quality images with simple prompts to generate beautiful images without a lot of prompt engineering.

Intended use and ethical considerations

Everything from Stable Diffusion v1-5, plus the fact that this is being built by two indie devs, so it was not extensively tested for new biases.

You can use this model commercially or whatever, but we are not liable if you do messed up stuff with it.

Caveats and recommendations

You can enforce style by prepending your prompt with estilovintedois if it is not good enough.

It should also be very dreamboothable, being able to generate high fidelity faces with a little amount of steps.

Thanks for the Google Developer Expert program for providing us with a GCP credits grant.