It is a character LoRA of Senko-san from Sewayaki Kitsune no Senko-san.
While v3A (which I personally preferred before I made v4) tends to work good only with some specific models, v4 seems to be more universal. I still couldn't get any good result with 2.5D models but maybe it's just because I didn't have any experience with them.
I had to delete tags from metadata of preview images because they contain some words civitai doesn't like. Here are couple examples of images which contain PNG-info:
https://files.catbox.moe/tfu9h5.png
https://files.catbox.moe/g4uj7n.png
https://files.catbox.moe/tvxd9d.png
My favorite model for this LoRA is Anything v5 but it should work fine with other 2D models.
I tried to train new version of the LoRA using different base models, but ended with that original NAI works the best as a base model for learning.
Description
I recommend to use this LoRA with weight about 1.4-2.6 (depends on your model it can be required to increase a weight). The bad stuff about this LoRA is that extension sd-webui-additional-networks supports max weight up to 2.0 which means you sometimes will have to setup LoRA directly in your prompt like <lora:senkoLora_v1:2.5>
FAQ
Comments (8)
<3
At last
Can you do a lower dim, safetensor version please?
I've just created another version of this LoRA with a lower dim, I will do some more experiments and publish 144MB safetensors version it in next couple of days.
Here you go.
@NeuroSenko Thank you!
256 dimensions and 2.5 weight? Are you training with a 1e-10 lr or something?
This LoRA is a result of merging three different LoRAs I've made locally (they used a bit different datasets and were training on different models). Every of base LoRAs were created using dim128 and 8e-4 learning rate. For some reasons all my initial LoRAs negatively affects on backgrounds in general, hovewer, after merging they work much better.
Anyway, I see the issue and today after playing with merging parameters I've been able to create a LoRA with 144 MB size which works good below 2.0 weights. I'm planning to play with merge parameters a bit just in order to find an optimal proportions. I'm going to publish another version in next couple of days.