Crimson Comics Effect V2

This V2 was created following a suggestion from a user (thanks pester41). The dataset is mostly the same as in V1, with some additional images added.

As you can see from the examples, the effect doesn’t perform at its best when the background is complex. For this reason, I also generated an image using only the effect, so you can compare it with the original images.

I recommend trying to use it with inpainting instead of generating the image directly, because this way you can control the effect much better.

What do you think about this update?

Crimson Comics Effect V1
You’re probably wondering what this is just from the name, so let’s go in order. About a month ago, a user commented on one of my LoRAs and asked whether it would be possible to create a LoRA that replicates the climax effects from Crimson’s works (specifically, those lightning-like energy effects you often see). Since then, the idea has been stuck in my head telling me: “As soon as I can, I need to give this a try.” So here I am.

Let’s talk a bit about what I tested. The works I used to train this LoRA are:

Crimson Girl

Crimson Cats

Crimson Gakuen

Crimson Spider 

Crimson Butterfly

Crimson Queen

I chose these because one thing I’ve learned is that the style has to be consistent — otherwise, you risk ending up with a concept LoRA that doesn’t work properly. And I realized very quickly that this kind of LoRA is much harder to make than a style or character LoRA.

To be safe, in addition to the base activation tag crimsoneffect, I’m also including the tags I used to help the model understand what it needed to learn:
invisible lightning effect, lightning veins, electric crackle overlay, glowing skin veins.

The reason is that sometimes the base tag alone isn’t enough to trigger the effect, and even when using these extra tags, there are cases where the result still doesn’t match the original Crimson effect very well. I’m not entirely sure why — after all, this is my first attempt at creating a concept LoRA.

As I’ve said many times, I’m not particularly good at prompting, so the sample images I generated probably don’t show the effect at its best. That’s why I’m leaving it to you: let me know how I can improve this LoRA, and whether it might be worth trying to create the other visual effects Crimson-sensei has used over the years as her style evolved.

With that said, I’ll hand it over to you — let me know what you think in the comments.