InDiffusion

InDiffusion uses state of the art deep learning AI to generate images based on text input. Dive into the world of AI-generated art with nothing more than the keyboard on your mobile device.

InDiffusion is based on Stable Diffusion software that has taken the world by storm with its high quality text to image generation. Simply describe a thing you would like to see and the machine will produce a picture you may want to hang on your wall. Include descriptions of the style you want your image to use (like "anime", "renaissance oil painting" or "Unreal 5 raytracing") and prepare to be amazed.
The AI software runs in the cloud and utilizes powerful GPUs, so your images will be ready in seconds.

All images you generate will be saved in your device's gallery so you can access them later to view and share.

If you're feeling adventurous, use the Surprise Me button to generate a random prompt for you using the tags that produce some of the best results. Stable Diffusion engine will then generate an image. Tweak the prompt to get the masterpiece closer to what's in your dreams.

Features:
- Text to image - Provide a text prompt describing a scene and InDiffusion will generate an image matching your request. Experiment with describing media to be used, details, styles, mood etc.
- Negative prompts - a very helpful feature allowing you to describe what you don't want in the picture. Together with the positive prompt this leads to better results
- Image2Image - upload an image or use the one you generated yourself and use it as a reference. The AI will make changes to it based on your prompt and produce a new piece of art.
- In-painting - together with image2image generation, in-painting allows you to paint over areas you would like to be modified. This creates a mask that tells the AI which regions should be re-generated and which should just be kept as is in the reference image. This is a powerful tool to give you precise control over every detail in the image.
- Styles - create a library of styles - shortcuts for the pieces of prompts you use often. When saved as a style this pieces can be easily added to your generation prompt with a single tap. No more repetitive typing!
- Image Size - select from Square, Portrait and Landscape formats to better suit your idea.
- Upscaling - turns this on when doing text to image or image to image generation and the resulting image will be twice the standard resolution (i.e. 1024x1024 instead of 512x512). The image is not simply stretched, it's a new image with 4x times more pixels and with additional new details. When used with text to image, fixing the Seed will help getting a similar, but more detailed image. With image to image, you can control the amount of new details added by adjusting the reference image weight.
- Samplers - you can now choose the sampler to be used from a long list of supported samplers. Sampler defines how the image is calculated between steps, so different samplers may produce different end results.
- Seed - you can set and/or re-use the seed. The seed defines the starting point for the image generation process, so you can use it to tweak the parameters on an image without changing it's subject much, or to reproduce previous works.
- Ads-optional mode. Use in-app purchases to avoid the ads, or use the app completely free with ads, or use both. Completely up to you!

Welcome to the world of AI art!