This was a special bonus prize included in the very first bounty I put out myself, found here https://civarchive.com/bounties/695?highlight=190231#comments
Congratulations to fran235 again for the winning entry and all who participated.
Yes I know what your thinking, a 1.5 lora that includes book titles, specifically the words "Advanced Dungeons and Dragons", that's crazy talk, it should be an XL model. Well I am not proficient in XL yet, but hear me out, this actually works pretty well.
Weight can be anywhere between 0.6 to 1, for closest to the old school book style I recommend 0.9. The lower you go, obviously the more the text disappears. Works really well with my Brothers Hildebrandt Style found here https://civarchive.com/models/212915/brothers-hildebrandt-style-artists,( but you want to keep that lora pretty low) as well as other loras.
You can get pretty creative with this one but some key words to use a book cover, novel illustration, and I think adding "advanced dungeons and dragons" helps with the text.
Description
FAQ
Comments (10)
Thank you!
I always wanted a LORA / Model that could do the styles of DnD Novel covers. I adore the styles.
Another idea if you have time for requests would be "Weird Tales" cover art as well. Similar to this but further back in time. Lol.
Thank you very much for this awesome reward! The lora works very vell with different checkpoints
This was a great request, kudos
No worries
Wow, this is fantastic! Really love fantasy styles like these. Thank you for sharing.
I see this was a special prize request, but if you're ever bored please consider a style lora based on Keith Parkinson's works!
By far my favorite fantasy artist, and I think you'd do a great job.
what tag makes the logo show up?
1.5 is not very good a doing text, so the logo showing up is a bit of a luck of the draw situation. It mostly shows up at higher weights so from anywhere from 0.8 -1. Technically the only prompt you need for it to show up is the keyword advdnd2023. I hope that helps. It also helps to just write "Advanced Dungeons and Dragons".
@Wolf_Systems That's probably it, then. When mixing multiple models during image generation, it's usually a good idea to mix them at medium-low levels so the image doesn't overdrive. Incidentally, I looked at the dataset info - did you really only use two tags? "style", and "advdnd2023 style" ...?
@Talboc Yeah, I kept this pretty simple. Yes I agree about Lora mixing, helps to avoid the burn effect. So I would try to put this Lora at a 0.7 and another lora at maybe 0.5. But easiest thing to do is to impaint your character afterwards at a slightly higher weight if you want ideal results. In saying that, this is an earlier lora of mine,its been a while since I generated anything with it, so I would looking at updating it soon to either XL, Pony or a better version of 1.5. XL will do much better text, as you can see in some of my other work such as the comic book maker 9000, logo maker 9000, etc etc.
@Wolf_Systems Kewl deal. When it comes time to update, I hope you update the 1.5 version, because I've never managed to get XL or Pony to work.