Continuing the AI images generation topic, I've started to work on a new project called Deorum. It is supposed to be a fantasy pantheon generator based on a huge number of precreated images.
It is in pre-alpha with admin part only, so not public as for now. There is a screen available for non-admins, where you can assign/verify tags to portraits.
The screen is open for you to play with, assign tags based on common sense and provide a generic feedback on how it looks and feels. Please note it's not the actual application, only a part of admin tools to manage images. Actual app concept is not finalized yet, I have some ideas and also open for suggestions.
There are already about 6000 images based on 100 unique hand-drawn portraits. And the current plan is to have at least 10000 portraits. All images are created using Stable Diffusion guided image-to-image generation. It took quite a lot of time to make it work as expected, but still lots of portraits are pretty lame and probably will be removed as tools are progressing fast and also our quality expectations are increasing.
Portraits will probably be licensed under CC BY-NC-SA 3.0, so free for non-commercial usage, but requiring payment for commercial needs and "claiming" characters/portraits. Once "claimed", the caracter will be removed from the database to make it truly unique. The idea is to make the tool slightly profitable or at least to cover infrastructure and AI-generation expenses.
Link: https://deorum.vercel.app
Need to sign up first, then click on 'Match' in the navbar. To assign a tag to an image click on 'Yes' to the right. If tag is not accosiated with the portrait, click 'No'. To open a new portrait faster press Enter, or Space, or Arrow.