Stable Diffusion AI is an easy-to-use app that lets you quickly generate images from text or other images with just a few clicks. With this app, you can communicate with your own server and generate high-quality images in seconds.
- Can use server environment powered by AI Horde (a crowdsourced distributed cluster of Stable Diffusion workers)
- Can use server environment powered by Stable-Diffusion-WebUI (AUTOMATIC1111)
- Can use server environment powered by SwarmUI
- Can use server envitonment powered by Hugging Face Inference API.
- Can use server environment powered by OpenAI (DALL-E-2, DALL-E-3).
- Can use server environment powered by Stability AI.
- Can use local environment powered by LocalDiffusion (Beta)
- Supports original Txt2Img, Img2Img modes
- Positive and negative prompt support
- Support dynamic size in range from 64 to 2048 px (for width and height)
- Selection of different sampling methods (available samplers are loaded from server)
- Unique seed input
- Dynamic sampling steps in range from 1 to 150
- Dynamic CFG scale in range from 1.0 to 30.0
- Restore faces option
- ( Img2Img ONLY ) : Image selection from device gallery (requires user permission)
- ( Img2Img ONLY ) : Capture input image from camera (requires user permission)
- ( Img2Img ONLY ) : Fetching random image for the input
- ( Img2Img ONLY ) : Inpaint (for A1111)
- Mask blur (1 to 64)
- Mask mode (Masked, not masked)
- Masked content (Fill, Original, Latent noise, Latent nothing)
- Inpaint area (Whole picture, only masked)
- Only masked padding (0 to 256 px)
- Batch generation with maximum of 20 images (for A1111 and Horde)
- Lora picker (for A1111)
- Textual inversion picker (for A1111)
- Hypernetworks picker (for A1111)
- SD Model picker (for A1111)
- In-app Gallery, stored locally, contains all AI generated images
- Displays generated images grid
- Image detail view: Zoom, Pinch, Generation Info.
- Export all gallery to .zip file
- Export single photo to .zip file
- Settings
- WebUI server URL
- Active SD Model selection
- Server availability monitoring (http-ping method)
- Enable/Disable auto-saving of generated images
- Enable/Disable saving generated images to
Download/SDAIandroid MediaStore folder - Clear gallery / app cache
This requires you to have the AUTOMATIC1111 WebUI that is running in server mode.
You can have it running either on your own hardware with modern GPU from Nvidia or AMD, or running it using Google Colab.
- Follow the setup instructions on Stable-Diffusion-WebUI repository.
- Add the arguments
--api --listento the command line arguments of WebUI launch script. - After running the server, get the IP address, or URL of your WebUI server.
- On the first launch, app will ask you for the server URL, enter it and press "Connect" button. If you want to change the server URL, go to Settings tab, choose "Configure" option and repeat the setup flow.
If for some reason you have no ability to run your server instance, you can toggle the Demo mode switch on server setup page: it will allow you to test the app and get familiar with it, but it will return some mock images instead of AI-generated ones.
This requires you to have the SwarmUI that is running in server mode.
You can have it running either on your own hardware with modern GPU from Nvidia or AMD, or running it using Google Colab.
Please refer to the SwarmUI documentation for installation instructions.
AI Horde is a crowdsourced distributed cluster of Image generation workers and text generation workers.
AI Horde requires to use API KEY, this mobile app alows to use either default API KEY (which is "0000000000"), or type your own. You can sign up and get your own AI Horde API KEY here.
Hugging Face Inference API allows to test and evaluate, over 150,000 publicly accessible machine learning models, or your own private models, via simple HTTP requests, with fast inference hosted on Hugging Face shared infrastructure. This service is free, but is rate-limited.
Hugging Face Inference requires to use API KEY, which can be created in Hugging Face account settings.
OpenAI provides a service for text to image generation using DALLE-2 or DALLE-3 models. This service is paid.
OpenAI requires to use API KEY, which can be created in OpenAI API Key settings.
StabilityAI is the image generation service provided by DreamStudio.
StabilityAI requires to use API KEY, which can be created in API Keys page.
Only txt2img mode is supported.
Allows to use phone resources to generate images.
App uses the language provided by OS default settings.
User interface of the app is translated for languages listed in this table:
Language
Since version
Status
English
0.1.0
Translated
Ukrainian
0.1.0
Translated
Turkish
0.4.1
Translated
Russian
0.5.5
Translated
Chinese (Simplified)
0.6.2
Translated
Any contributions to the translations are welcome.
As Google Play has some policies that app needs to be compliant with in order to be allowed to publish on Google Play there are some differences between builds distributed via Google Play and F-Droid/GitHub releases, listed in this table:
Feature
Google Play build
F-Droid/GitHub build
Reason
Sideloading LocalDiffusion custom model
❌
✅
Google Play does not allow publishing apps with android.permission.MANAGE_EXTERNAL_STORAGE permission, which is required to read custom model files from external storage directly.
This software is open source, provided with no warranty, and you are welcome to use it for free.
In case you find this software valuable, and you'd like to say thanks and show a little support, here is the button:
