Danbooru - Artist Tags for V4.5 + More! (Patreon)
Content
I've updated the page, and since last time I did an update, there on Patreon, there has been a lot going on, especially under the hood.
Databases! Everything has been moved over to a database. With this I mean that behind the scenes there is a happy little database keeping things spinning around in the correct direction. It saves all image data, metadata, votes, and so on.
User logins! I've opted for a passphrase based login system. That means that when you create a user, you get a passphrase and everything you vote for (and in the future, all your favorites) will be saved to that passphrase. Don't show it to someone else, or they can use it to since the passphrase itself is both your login and password. Akin to how Mullvad VPN only has an account number and no passwords.
Lost your passphrase? Pfh... tough luck! But if you ask nicely I can dig in the database and see if I can't find it.
What's the QR? Take a picture of it, and you can use the link to log you in automatically on your tablet or phone.
Mobile experience: I've changed things like the navbar, etc to better work on your cellphone. I'm not quite done yet with the login window, it is a bit skewed, but we're gettin' there.
Pretty pretty icons: I've changed all Unicode and Emoji based icons to instead use Remix Icon. That way the look is standardized across the board. There were only a few logos and such I needed to make my own SVG-versions of.
Improved layout: Tweaks and improvements to layout. This is always and ongoing. This project for me to rekindle my old roots as a web designer, and get my interest in it back. So there will always be slight tweaking and overworking details.
Artist Tags for V4.5
With the release of NovelAI Diffusion 4.5 I had to do some new testing. This time I wanted to see if there was a difference between the images generated using v4 and v4.5. So I generated a new set of images using the same prompt and settings as before, but this time with the new model.
Initial testing turned out... not so great. So in the end, I had to compose an entirely new prompt to better showcase artist styles. But as before, it is all based on Danbooru artist tags, going from the most popular down to top 150.
Compared to the older model, v4.5 replicates artists a lot better. However, you might see that some artists known for things like realistic drawings aren't represented well since my prompt is missing the 'realistic' tag.
Generation
I wrote a Python script utilizing the NovelAI API to generate all these images. I first locked down a generic prompt to allow the artist tag to influence the image as much as possible and then set it up to iterate through each and every one of the tags. The script is constructed in such a way that it requires my input every time it generates a new image. So there have been a whole lot of button pressing. This is due to the NovelAI ToS, 9.1.5, which I have also confirmed with the NovelAI team.
Prompt:
artist:@@@ (eckzahn), neutral background, white background, transparent background, cowboy shot, -1.5::nipples::
Character 1
girl, soft smile, red lips, hoodie, jeans, orange hair, small breasts
Model used is NAI Diffusion V4.5 Full, Steps 28, Guidance 4, Quality Tags Enabled, UC Preset Heavy, Seed 2710136181.
Files
Previews only