Hello welcome!
Here I leave a small introduction of the LoRA:
My idea was to have a concept with detailed backgrounds that simulate watercolor drawings, sketches or quite detailed backgrounds. But I made a mistake and now I'm here to correct it...
IL | V1:
This version works well with other Loras, adds nice detailed backgrounds, and since it wasn't trained in just one style, it adapts well to other styles. I recommend it alongside pastel styles, or just play around with it as you like. It's fun to experiment with style changes ~ XD
Upgrade | WDB | V2:
To try to correct the errors in the faces, I added a greater number of data with more detailed faces without changing the style much, I changed the main trigger and moderately suppressed the "detailed background" trigger to avoid some errors from the previous version.
WIP | WDB | V1:
The idea of this was to train it as a concept but I ended up training it as a style, that was a mistake on my part because I always use faces in the foreground to avoid a lack of details in complex landscapes and the dataset did not have many examples like that, I still upload it because It works well for me with other styles and I plan to fix it later, let's just say it's a wip ~ 7w7)/
Description
I like this lora, but I needed to refine it a bit more since some details were lost in the original (updated) version, however, it is trained on Agga's so I don't know how well it works with other models :3
FAQ
Comments (5)
I can't make this one work either, tried with all triggerwords, together and not... i dunno why your recent LORA don't work as much aas the ones you did before (not blaming you, just wondering what changed, the haruhime and the ones before it works perfectly for exemple)
Hi, it's about the training model. This version is trained on Utopia, and Utopia is a fusion of Noob, Illustrious, and Pony. I've also trained some Lora on Illustrious v2.0, but many models were made with fusions of Noob AI along with Illustrious v0.1, so that can fail. However, the previous version is trained on Illustrious 0.1 (the base), so it should serve you better.
PS: I've only tried it on wai nsfw and it works pretty well there.
@Herrscher_AGGA i see, thanks for the info!
I used forge and used the same UtopiaxlV30ALTR as you and this set of lora, and then copied all the parameters of the example diagram The generated result is very different from the example diagram，I can't reproduce the example diagram
Hi, I made the images in Civitai so I can't help you with that, but I think it uses a1111 with normal, I'm not sure.