Tasks

Image-to-Image

Image-to-image is the task of transforming an input image through a variety of possible manipulations and enhancements, such as super-resolution, image inpainting, colorization, and more.

Inputs
Image-to-Image Model
Output

About Image-to-Image

Image-to-image pipelines can also be used in text-to-image tasks, to provide visual guidance to the text-guided generation process.

Use Cases

Image inpainting

Image inpainting is widely used during photography editing to remove unwanted objects, such as poles, wires, or sensor dust.

Image colorization

Old or black and white images can be brought up to life using an image colorization model.

Super Resolution

Super-resolution models increase the resolution of an image, allowing for higher-quality viewing and printing.

Inference

You can use pipelines for image-to-image in 🧨diffusers library to easily use image-to-image models. See an example for StableDiffusionImg2ImgPipeline below.

import torch
from diffusers import AutoPipelineForImage2Image
from diffusers.utils import make_image_grid, load_image

pipeline = AutoPipelineForImage2Image.from_pretrained(
    "stabilityai/stable-diffusion-xl-refiner-1.0", torch_dtype=torch.float16, variant="fp16", use_safetensors=True
)

# this helps us to reduce memory usage- since SDXL is a bit heavy, this could help by
# offloading the model to CPU w/o hurting performance.
pipeline.enable_model_cpu_offload()

# prepare image
url = "https://huggingface.co./datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-sdxl-init.png"
init_image = load_image(url)

prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"

# pass prompt and image to pipeline
image = pipeline(prompt, image=init_image, strength=0.5).images[0]
make_image_grid([init_image, image], rows=1, cols=2)

You can use huggingface.js to infer image-to-image models on Hugging Face Hub.

import { HfInference } from "@huggingface/inference";

const inference = new HfInference(HF_TOKEN);
await inference.imageToImage({
    data: await (await fetch("image")).blob(),
    model: "timbrooks/instruct-pix2pix",
    parameters: {
        prompt: "Deblur this image",
    },
});

Uses Cases for Text Guided Image Generation

Style Transfer

One of the most popular use cases of image-to-image is style transfer. With style transfer models:

  • a regular photo can be transformed into a variety of artistic styles or genres, such as a watercolor painting, a comic book illustration and more.
  • new images can be generated using a text prompt, in the style of a reference input image.

See 🧨diffusers example for style transfer with AutoPipelineForText2Image below.

from diffusers import AutoPipelineForText2Image
from diffusers.utils import load_image
import torch

# load pipeline
pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="sdxl_models", weight_name="ip-adapter_sdxl.bin")

# set the adapter and scales - this is a component that lets us add the style control from an image to the text-to-image model
scale = {
    "down": {"block_2": [0.0, 1.0]},
    "up": {"block_0": [0.0, 1.0, 0.0]},
}
pipeline.set_ip_adapter_scale(scale)

style_image = load_image("https://huggingface.co./datasets/huggingface/documentation-images/resolve/0052a70beed5bf71b92610a43a52df6d286cd5f3/diffusers/rabbit.jpg")

generator = torch.Generator(device="cpu").manual_seed(26)
image = pipeline(
    prompt="a cat, masterpiece, best quality, high quality",
    ip_adapter_image=style_image,
    negative_prompt="text, watermark, lowres, low quality, worst quality, deformed, glitch, low contrast, noisy, saturation, blurry",
    guidance_scale=5,
    num_inference_steps=30,
    generator=generator,
).images[0]
image

ControlNet

Controlling the outputs of diffusion models only with a text prompt is a challenging problem. ControlNet is a neural network model that provides image-based control to diffusion models. Control images can be edges or other landmarks extracted from a source image. Examples

Pix2Pix

Pix2Pix is a popular model used for image-to-image translation tasks. It is based on a conditional-GAN (generative adversarial network) where instead of a noise vector a 2D image is given as input. More information about Pix2Pix can be retrieved from this link where the associated paper and the GitHub repository can be found.

The images below show some examples extracted from the Pix2Pix paper. This model can be applied to various use cases. It is capable of relatively simpler things, e.g., converting a grayscale image to its colored version. But more importantly, it can generate realistic pictures from rough sketches (can be seen in the purse example) or from painting-like images (can be seen in the street and facade examples below).

Examples

Useful Resources

References

[1] P. Isola, J. -Y. Zhu, T. Zhou and A. A. Efros, "Image-to-Image Translation with Conditional Adversarial Networks," 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 5967-5976, doi: 10.1109/CVPR.2017.632.

This page was made possible thanks to the efforts of Paul Gafton and Osman Alenbey.

Compatible libraries

Image-to-Image demo
Inference Examples
Inference API (serverless) has been turned off for this model.
Models for Image-to-Image
Browse Models (752)

Note An image-to-image model to improve image resolution.

Note A model that increases the resolution of an image.

Note A model that creates a set of variations of the input image in the style of DALL-E using Stable Diffusion.

Note A model that generates images based on segments in the input image and the text prompt.

Note A model that takes an image and an instruction to edit the image.

Datasets for Image-to-Image
Browse Datasets (384)

No example dataset is defined for this task.

Note Contribute by proposing a dataset for this task !

Spaces using Image-to-Image

Note Image enhancer application for low light.

Note Style transfer application.

Note An application that generates images based on segment control.

Note Image generation application that takes image control and text prompt.

Note Edit images with instructions.

Metrics for Image-to-Image
PSNR
Peak Signal to Noise Ratio (PSNR) is an approximation of the human perception, considering the ratio of the absolute intensity with respect to the variations. Measured in dB, a high value indicates a high fidelity.
SSIM
Structural Similarity Index (SSIM) is a perceptual metric which compares the luminance, contrast and structure of two images. The values of SSIM range between -1 and 1, and higher values indicate closer resemblance to the original image.
IS
Inception Score (IS) is an analysis of the labels predicted by an image classification model when presented with a sample of the generated images.