What do you get when you combine two similar but under-trained LoRAs? Something interesting looking.

This combo was primarily trained on landscape oriented photos and specifically avoided people in order to avoid distorting characters (though untested). I'm happy with the results, personally.

If you wish to support my work, the easiest way to do so it to post your creations directly to this page as it rewards me the most buzz, thanks.

Note: All promo images taken at 1.20 strength with ComfyUi's new "res_multistep" sampler rather than Euler.