Jan 26, 2023 19:00:00

``OpenJourney'' that reproduces the image generation AI ``Midjourney'' with open source appears

A project aimed at providing people with open source equivalents to AI /

Midjourney that automatically generates images from input character strings (prompts) is ' OpenJourney 'is. OpenJourney adjusts Stable Diffusion v1.5, which is developed open source with the same image generation AI, and is developed to operate close to Midjourney, and the repository is published on Hugging Face, an online AI platform.

Open Journey
https://open-journey.github.io/

prompthero/openjourney Hugging Face
https://huggingface.co/prompthero/openjourney

The OpenJourney development team said, “The main idea behind OpenJourney is that we should have full control over the tools and materials generated. , and you must be able to sell the art you generate in any way you like.”

Since OpenJourney uses NVIDIA CUDA , an NVIDIA GPU is required to install and use it in the local environment. However, it is possible to touch the demo on Hugging Face.

prompthero/openjourney – Run with an API on Replicate
https://replicate.com/prompthero/openjourney

In the Input on the above page, 'promput', 'width (width of output image)', 'height (height of output image)', 'num_outputs (number of output images)', 'num_inderence_steps (number of generation steps)' and 'guidance_scale ( CFG Scale )' 'seed (seed value)' and click 'Submit'.

During generation, a log is displayed on the right side of the screen.

The generated image is shown on the right.

This time, I entered 'a photo of beautiful woman playing piano and old candlestick, Granblue Fantasy, golden lighting, Nikon D4 and Nikon Lens 50mm F1.4' at the prompt to draw 'a woman playing the piano'. Although there is an impression that the sexiness of women is somewhat emphasized, the image was output as per the prompt.