playgroundai / playground-v2.5-1024px-aesthetic

Playground v2.5 is the state-of-the-art open-source model in aesthetic quality

Warm

Public
2.2M runs
A100 (80GB)
GitHub
Weights
Paper
License

Run with an API

Playground API Examples README Versions

Input

prompt

string

Shift + Return to add a new line

Astronaut in a jungle, cold color palette, muted colors, detailed, 8kAstronaut in a jungle, cold color palette, muted colors, detailed, 8k

Input prompt

Default: "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"

negative_prompt

string

Shift + Return to add a new line

ugly, deformed, noisy, blurry, distortedugly, deformed, noisy, blurry, distorted

Negative Input prompt

Default: "ugly, deformed, noisy, blurry, distorted"

image

file

Input image for img2img or inpaint mode

mask

file

Input mask for inpaint mode. Black areas will be preserved, white areas will be inpainted.

width

integer

(minimum: 256, maximum: 1536)

Width of output image

Default: 1024

height

integer

(minimum: 256, maximum: 1536)

Height of output image

Default: 1024

num_outputs

integer

(minimum: 1, maximum: 4)

Number of images to output.

Default: 1

scheduler

string

Scheduler. DPMSolver++ or DPM++2MKarras is recommended for most cases

Default: "DPMSolver++"

num_inference_steps

integer

(minimum: 1, maximum: 60)

Number of denoising steps

Default: 25

guidance_scale

number

(minimum: 0.1, maximum: 20)

Scale for classifier-free guidance

Default: 3

prompt_strength

number

(minimum: 0, maximum: 1)

Prompt strength when using img2img / inpaint. 1.0 corresponds to full destruction of information in image

Default: 0.8

seed

integer

Random seed. Leave blank to randomize the seed

apply_watermark

boolean

Applies a watermark to enable determining if an image is generated in downstream applications. If you have other provisions for generating or deploying images safely, you can use this to disable watermarking.

Default: true

disable_safety_checker

boolean

Disable safety checker for generated images. This feature is only available through the API. See https://replicate.com/docs/how-does-replicate-work#safety

Default: false

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

npx create-replicate --model=playgroundai/playground-v2.5-1024px-aesthetic

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 playgroundai/playground-v2.5-1024px-aesthetic using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "playgroundai/playground-v2.5-1024px-aesthetic:a45f82a1382bed5c7aeb861dac7c7d191b0fdf74d8d57c4a0e6ed7d4d0bf7d24",
  {
    input: {
      width: 1024,
      height: 1024,
      prompt: "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
      scheduler: "DPMSolver++",
      num_outputs: 1,
      guidance_scale: 3,
      apply_watermark: true,
      negative_prompt: "ugly, deformed, noisy, blurry, distorted",
      prompt_strength: 0.8,
      num_inference_steps: 25
    }
  }
);
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 playgroundai/playground-v2.5-1024px-aesthetic using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "playgroundai/playground-v2.5-1024px-aesthetic:a45f82a1382bed5c7aeb861dac7c7d191b0fdf74d8d57c4a0e6ed7d4d0bf7d24",
    input={
        "width": 1024,
        "height": 1024,
        "prompt": "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
        "scheduler": "DPMSolver++",
        "num_outputs": 1,
        "guidance_scale": 3,
        "apply_watermark": True,
        "negative_prompt": "ugly, deformed, noisy, blurry, distorted",
        "prompt_strength": 0.8,
        "num_inference_steps": 25
    }
)
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 playgroundai/playground-v2.5-1024px-aesthetic 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": "a45f82a1382bed5c7aeb861dac7c7d191b0fdf74d8d57c4a0e6ed7d4d0bf7d24",
    "input": {
      "width": 1024,
      "height": 1024,
      "prompt": "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
      "scheduler": "DPMSolver++",
      "num_outputs": 1,
      "guidance_scale": 3,
      "apply_watermark": true,
      "negative_prompt": "ugly, deformed, noisy, blurry, distorted",
      "prompt_strength": 0.8,
      "num_inference_steps": 25
    }
  }' \
  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/playgroundai/playground-v2.5-1024px-aesthetic@sha256:a45f82a1382bed5c7aeb861dac7c7d191b0fdf74d8d57c4a0e6ed7d4d0bf7d24 \
  -i 'width=1024' \
  -i 'height=1024' \
  -i 'prompt="Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"' \
  -i 'scheduler="DPMSolver++"' \
  -i 'num_outputs=1' \
  -i 'guidance_scale=3' \
  -i 'apply_watermark=true' \
  -i 'negative_prompt="ugly, deformed, noisy, blurry, distorted"' \
  -i 'prompt_strength=0.8' \
  -i 'num_inference_steps=25'

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/playgroundai/playground-v2.5-1024px-aesthetic@sha256:a45f82a1382bed5c7aeb861dac7c7d191b0fdf74d8d57c4a0e6ed7d4d0bf7d24
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "width": 1024,
      "height": 1024,
      "prompt": "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
      "scheduler": "DPMSolver++",
      "num_outputs": 1,
      "guidance_scale": 3,
      "apply_watermark": true,
      "negative_prompt": "ugly, deformed, noisy, blurry, distorted",
      "prompt_strength": 0.8,
      "num_inference_steps": 25
    }
  }' \
  http://localhost:5000/predictions

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

Output

{
  "completed_at": "2024-03-15T21:00:51.212675Z",
  "created_at": "2024-03-15T20:59:27.831539Z",
  "data_removed": false,
  "error": null,
  "id": "hhow2slbfyrxyaxtbucbhzwcti",
  "input": {
    "width": 1024,
    "height": 1024,
    "prompt": "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
    "scheduler": "DPMSolver++",
    "num_outputs": 1,
    "guidance_scale": 3,
    "apply_watermark": true,
    "negative_prompt": "ugly, deformed, noisy, blurry, distorted",
    "prompt_strength": 0.8,
    "num_inference_steps": 25
  },
  "logs": "Using seed: 2478151516\nPrompt: Astronaut in a jungle, cold color palette, muted colors, detailed, 8k\ntxt2img mode\n  0%|          | 0/25 [00:00<?, ?it/s]\n  4%|▍         | 1/25 [00:00<00:07,  3.06it/s]\n 12%|█▏        | 3/25 [00:00<00:04,  5.13it/s]\n 16%|█▌        | 4/25 [00:00<00:04,  5.08it/s]\n 20%|██        | 5/25 [00:01<00:03,  5.05it/s]\n 24%|██▍       | 6/25 [00:01<00:03,  5.03it/s]\n 28%|██▊       | 7/25 [00:01<00:03,  5.00it/s]\n 32%|███▏      | 8/25 [00:01<00:03,  4.99it/s]\n 36%|███▌      | 9/25 [00:01<00:03,  4.98it/s]\n 40%|████      | 10/25 [00:02<00:03,  4.98it/s]\n 44%|████▍     | 11/25 [00:02<00:02,  4.98it/s]\n 48%|████▊     | 12/25 [00:02<00:02,  4.97it/s]\n 52%|█████▏    | 13/25 [00:02<00:02,  4.96it/s]\n 56%|█████▌    | 14/25 [00:02<00:02,  4.96it/s]\n 60%|██████    | 15/25 [00:03<00:02,  4.97it/s]\n 64%|██████▍   | 16/25 [00:03<00:01,  4.97it/s]\n 68%|██████▊   | 17/25 [00:03<00:01,  4.96it/s]\n 72%|███████▏  | 18/25 [00:03<00:01,  4.96it/s]\n 76%|███████▌  | 19/25 [00:03<00:01,  4.96it/s]\n 80%|████████  | 20/25 [00:04<00:01,  4.96it/s]\n 84%|████████▍ | 21/25 [00:04<00:00,  4.96it/s]\n 88%|████████▊ | 22/25 [00:04<00:00,  4.95it/s]\n 92%|█████████▏| 23/25 [00:04<00:00,  4.95it/s]\n 96%|█████████▌| 24/25 [00:04<00:00,  4.95it/s]\n100%|██████████| 25/25 [00:05<00:00,  4.95it/s]\n100%|██████████| 25/25 [00:05<00:00,  4.94it/s]",
  "metrics": {
    "predict_time": 8.01971,
    "total_time": 83.381136
  },
  "output": [
    "https://replicate.delivery/pbxt/XAK4XRgpjYaCGRrm9yxzO2bacj4XTE1Nl6bwaXKOHKYApJoE/out-0.png"
  ],
  "started_at": "2024-03-15T21:00:43.192965Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/hhow2slbfyrxyaxtbucbhzwcti",
    "cancel": "https://api.replicate.com/v1/predictions/hhow2slbfyrxyaxtbucbhzwcti/cancel"
  },
  "version": "a45f82a1382bed5c7aeb861dac7c7d191b0fdf74d8d57c4a0e6ed7d4d0bf7d24"
}

Generated in

8.0 seconds

Tweak itReport

Examples

View more examples

Run time and cost

This model costs approximately $0.066 to run on Replicate, or 15 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 48 seconds. The predict time for this model varies significantly based on the inputs.

Readme

Playground v2.5 – 1024px Aesthetic Model

This repository contains a model that generates highly aesthetic images of resolution 1024x1024, as well as portrait and landscape aspect ratios. You can use the model with Hugging Face 🧨 Diffusers.

image/png

Playground v2.5 is a diffusion-based text-to-image generative model, and a successor to Playground v2.

Playground v2.5 is the state-of-the-art open-source model in aesthetic quality. Our user studies demonstrate that our model outperforms SDXL, Playground v2, PixArt-α, DALL-E 3, and Midjourney 5.2.

For details on the development and training of our model, please refer to our blog post and technical report.

Model Description

Developed by: Playground
Model type: Diffusion-based text-to-image generative model
License: Playground v2.5 Community License
Summary: This model generates images based on text prompts. It is a Latent Diffusion Model that uses two fixed, pre-trained text encoders (OpenCLIP-ViT/G and CLIP-ViT/L). It follows the same architecture as Stable Diffusion XL.

Using the model with 🧨 Diffusers

Install diffusers >= 0.27.0 and the relevant dependencies. For now, you need to install from the main diffusers branch in GitHub until a new release is published in PyPi.

Notes: - The pipeline uses the EDMDPMSolverMultistepScheduler scheduler by default, for crisper fine details. It’s an EDM formulation of the DPM++ 2M Karras scheduler. guidance_scale=3.0 is a good default for this scheduler. - The pipeline also supports the EDMEulerScheduler scheduler. It’s an EDM formulation of the Euler scheduler. guidance_scale=5.0 is a good default for this scheduler.

Using the model with Automatic1111/ComfyUI

Support coming soon. We will update this model card with instructions when ready.

User Studies

This model card only provides a brief summary of our user study results. For extensive details on how we perform user studies, please check out our technical report.

We conducted studies to measure overall aesthetic quality, as well as for the specific areas we aimed to improve with Playground v2.5, namely multi aspect ratios and human preference alignment.

Comparison to State-of-the-Art

image/png

The aesthetic quality of Playground v2.5 dramatically outperforms the current state-of-the-art open source models SDXL and PIXART-α, as well as Playground v2. Because the performance differential between Playground V2.5 and SDXL was so large, we also tested our aesthetic quality against world-class closed-source models like DALL-E 3 and Midjourney 5.2, and found that Playground v2.5 outperforms them as well.

Multi Aspect Ratios

image/png

Similarly, for multi aspect ratios, we outperform SDXL by a large margin.

image/png

Next, we benchmark Playground v2.5 specifically on people-related images, to test Human Preference Alignment. We compared Playground v2.5 against two commonly-used baseline models: SDXL and RealStock v2, a community fine-tune of SDXL that was trained on a realistic people dataset.

Playground v2.5 outperforms both baselines by a large margin.

MJHQ-30K Benchmark

image/png

Model	Overall FID
SDXL-1-0-refiner	9.55
playground-v2-1024px-aesthetic	7.07
playground-v2.5-1024px-aesthetic	4.48

Lastly, we report metrics using our MJHQ-30K benchmark which we open-sourced with the v2 release. We report both the overall FID and per category FID. All FID metrics are computed at resolution 1024x1024. Our results show that Playground v2.5 outperforms both Playground v2 and SDXL in overall FID and all category FIDs, especially in the people and fashion categories. This is in line with the results of the user study, which indicates a correlation between human preferences and the FID score of the MJHQ-30K benchmark.

Input

Output

Examples

Run time and cost

Readme

Playground v2.5 – 1024px Aesthetic Model

Model Description

Using the model with 🧨 Diffusers

Using the model with Automatic1111/ComfyUI

User Studies

Comparison to State-of-the-Art

Multi Aspect Ratios

Human Preference Alignment on People-related images

MJHQ-30K Benchmark