The world of AI Art is moving at such a fast pace that it’s hard to keep track of what tools are available out there. Hence I am putting together the list of tools I have found on the internet
A collection of Free AI tools and resources that are hosted on Hugging Face for you to try and experiment in various aspects of AI Art creation. This collection is always growing as new tools are discovered so make sure you bookmark this page.
- CLIP Interrogator – A great tool by @pharmapsychotic that allows you to upload and image and figure out the prompt that would result in similar images using AI. I did a dedicated post on it if you are curious check it out as well.
- CLIP Interrogator 2 – This version is specialized for producing nice prompts for use with Stable Diffusion 2.0 using the ViT-H-14 OpenCLIP model, created by @pharmapsychotic
- Diffusion Space – Demo for Nitrosocke’s fine-tuned model for creating uniquely styled AI Art.
- Openjourney – Demo of Openjourney fine-tuned model by akhaliq
- 4x-UltraSharp – An upscaling model for generating high resolution images from low resolution source images. This model can produce some very nice details especially in portraits. The file is only 67 Mb in size, place in your ESRGAN folder to use with Automatic1111.
Quick links to download various models so you can get to their downloads page.
Custom Fine Tuned Models
A collection of fine-tuned AI models for Stable Diffusion that are available for use for Free or a small fee.
- Future Diffusion – a custom model for Stable Diffusion 2.0 created by Nitrosocke
- Ghibli Diffusion – This is the fine-tuned Stable Diffusion 1.5 model by Nitrosocke trained on images from modern anime feature films from Studio Ghibli.
- Openjourney – This is an open source Stable Diffusion 1.5 fine tuned model on Midjourney images, by PromptHero
- Halloween Diffusion – This Stable Diffusion 1.5 model was trained on halloween vibes to assist you in the spooky season 🦇 by @proximasan
- Oldfilm Artstyle – Paid version. Gives things an ethereal vibe, a custom fine tuned Stable Diffusion 1.5 model by @proximasan
- Darkvictorian Artstyle – Trained on dark, moody, “victorian” imagery. The download includes 2 versions of the model fine tuned Stable Diffusion 1.5 model by @proximasan
- Fantassified Icons 1.0 – A custom finetuned model for Stable Diffusion 1.5 to generate icons inspired by fantasy game by @proximasan. Sample images and instructions provided, just use the token ‘fantassified icon’ to generate such icons. Refer to the instructions for further tips.
- 4xNMKD Upscaler – 4x NMKD Superscale (4x_NMKD-Superscale-SP178000_G) available via Civitai website (caution NSFW)
Google Colabs Notebooks
- CLIP Interrogator – by @pharmapsychotic figure out what a good prompt might be to create new images like an existing one? The CLIP Interrogator is here to get you answers! This version is specialised for producing nice prompts for use with Stable Diffusion 2.0 using the ViT-H-14 OpenCLIP model!
- DreamBooth Stable Diffusion 2.0 – A notebook designed to allow you to train your own model and create a CKPT file based on Stable Diffusion. This notebook is a fork from the original DreamBooth notebook.
- Stable Diffusion Notebook – Stable diffusion notebook created by Engineering of Stuff which you can download and import into your Paperspace environment. I have a blog post that walks through on how to run it.
- Dygen – A painterly conversion for any image (PNG). This tool is created by a very talented AI artist @dymokomi
- Ultimate SD Upscale extension for AUTOMATIC1111 – a diffusion based tile approach to upscaling an image in Automatic1111 using Img2Img tab. Once installed look at the Scripts drop down in that tab.
- SD Prompt Reader – A stand alone tool that let’s your read prompt information from any PNG created using Automatic1111, ComfyUI and InvokeAI. Its quite useful if you are often switching from one UI to another. There is a video explaining more about this which you can watch.
- ControlNet v1 – Very first version of ControlNet released containing 8x models. Download these models via this link.
- ControlNet v1.1 – More models and lighter model files of ControlNet released after v1.
- Posemy.art – a free easy to use character posing app that can be used in ControlNet. Export the character in OpenPose format.