v2.1 Updated 19/5/23
If you enjoy the models I create, please consider buying me a coffee!
I am pleased to announce the release of DarkestDungeon v2 (lora)
What is it?
DarkestDungeon is lora revolving around the artstyle from 'Darkest Dungeon', one of my favourite games of all time. Applies a style to whatever you want, portraits to landscapes. It does it all.
It has been trained partly with modders/game devs in mind. Because of this, you can now make images with a plain white background, such as characters, items, weapons and armour and anything you can think of. (See trigger words below, but it works for ddstyle too).
How do I use it?
To recreate it's atmospheric artstyle, there are a couple of simple rules that I recommend following:
1- The trigger word "ddstyle" goes at the front of your prompt.
2- Keep negatives simple. I use "(worst quality, low quality, normal quality:1.4)"
3- Try adding "darkest dungeon"
4- Play around with the Lora strength if you are finding it hard to get what you want, and try to remove conflicting keywords.
5- Use booru tags as this was trained on a NAI based model: https://danbooru.donmai.us/tags
What models are best?
- It's intended to work with any anime model. Realistic models aren't as effective
Trigger words:
"ddstyle"
"ddtrinketicon"
"crimson court"
"the Cove"
"the Weald"
"the Warrens"
"the Ruins"
Some have been removed, but I'll add them back in soon.
Description
Better at prompting, you can now be more creative.
Overall style improvements
FAQ
Comments (22)
keep putting out quality content and you'll blow up in no time
THE BLOOD PUMPS, THE LIMBS OBEY
This will go a long way towards helping make mods.
If you post any, I’d be very interested in trying them out!
I will be adding in characters and items with plain backgrounds (for easy editing) so that you will be able to make your own easier!
Absolutely. I recently returned to the game and paid attention to mods. I dream of a mod with a complete rework of all the backgrounds in the game to be more beautiful and detailed, but at the same time in harmony with the characters and with a dark atmosphere. With this Lora, we can do something similar quite quickly.
Nice. More trinkets and baubles for my collection.
One neat thing about this LoRA is that 512x512 images will have thicker lines than 1024x1024, so 512x512 is more evocative of DD1 whereas 1024x1024 is more evocative of DD2.
This is a really awesome LORA. Thanks :)
Congrats on release, beautiful work.
Thank you!
love the lora, but you should keep v1.1 also, it had a desired effect when mixed with comic babes
Thanks for pointing that out. I still have it, I'll re-add it soon
Hey, I just updated from v2 to v2.1, and now I cant use this LORA. Getting:
SafeTensorError: Error while deserializing header: MetadataIncompleteBuffer.
'AttributeError: 'NoneType' object has no attribute 'items'.
Ideally I'd like to rollback to V2, but thats been deleted from Civitai for some reason, so if you could reupload v2 thatd be great.
Hey! Not sure what may be causing that, might be unrelated. But I’ll upload the old versions again in a couple of days
First thing when having problems with models/loras - delete them and re-download first.
Idk why it is sometimes like this.. but it is..
It helped so many times, not just for me, but a lot ppl on CivitAI discord.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.