This is a more extreme version of JBX Beautiful Brushwork. The LoRA was trained by cycling the best results from Beautiful Brushwork back into the training image set, along with some specific images from the original training run.

The result is a much stronger style. It's a little harder to use, and you have to finesse the weights a bit to avoid it going over the top and producing images that are too loose and painterly.

I would recommend starting with a weight of 0.5 in most cases, and then working up from there. A full weight of 1 can work, but it's probably not ideal.

This time the LoRA was trained using the CyberRealistic model rather than Avalon TruVision as before.

I've tested it using both Avalon TruVision and Cyberrealistic and both work well. The style is effective for a range of images, as per the example. Characters, locations, and objects all work well.