Update:

I found it incredibly useful to stack v0.1 with the v.01 boost together.

"v0.1" is the main lora

"v0.1 BOOST" can make more interesting compositions and deeper textures if used together with v0.1

Good combination is v0.1 = 0.5 weight, v0.1 BOOST = 0.7 weight.

First Anima LoRa attempt, still trying to figure things out.

This LoRa is focusing on the world building concept of the A New Dawm anime movie that just came out this year. Dataset is based on screenshots from trailers, I hope to improve it when the actual movie is available.

Settings I like to use;

Sampler: er_sde (toggle "advanced" and scroll all the way down. er_sde was now added to the tensorArt generator) or dpmpp 2m SDE gpu

Scheduler: Simple

Steps: 30

CFG: 3.5-5

Some of these prompts can be useful depending on the look you're going for:

muted color, pale color, painterly, watercolor \(medium\), cover page, offcial art, detailed foreground, depths of field

It defaults towards a greenish hue. But it was trained on a number of different hues, as the anime offers a variety of vibes. For example you can put "mint theme" or "green theme" in the negative prompt to get a more vibrant blueish pastel touch.