Comic style SDXL - Jhyd SDXL - v1.0 Review

eldritchadam

lora

sdxl

comics

excellent style with a lot of promise and minor downsides

it's crucial at the moment that you use some negative prompts for this LoRA - photo, blur, bokeh

At least when using base SDXL. I suspect if you want to use this with a model fine-tuned on anime or 3d art, you will want those terms in your negative. The strength of this LoRA model is somewhat low, and easily overridden by the content of the prompt subject. So use negative prompts to keep it on track. I think it likely the model can get stronger with different training parameters. But even with the little bit of a necessary workaround, it is a beautiful, flexible style.

Minimal effort in the samples below - just ran a prompt once with two outputs and chose the better of the two so I don't need to cherry-pick a lot to get decent samples. With just a little more effort, I'm sure this can go far.

I give it 5 stars, though I would want to hold back 1/2 star - and it's rounding up. I'll give that final 1/2 star in my mind if the model sees a little more contrast and gets over the need for the negative prompt.