A LoRa trained with 50 of my own oil paintings. I paint quickly, using the alla prima technique (single layer) and usually on canvas, so the LoRa reproduces that with acceptable fidelity. The brushstrokes are clearly visible and the texture of the cotton canvas can be glimpsed at times.Using a strenght of approximately 0.80-0.90 I achieve a result similar to my real work, with less, a more finished and perfectionist image will be achieved, and with more, a freer but more muddled one. You can use with any checkpoint of course but it gives the strongest results with Juggernaut XL (thanks to <a target="_blank" rel="ugc" href="https://civarchive.com/user/KandooAI">KandooAI) </a>which was used as base for the training proccess. It works particullarly well with DreamShaper XL Turbo. Havent tried any other checkpoints but should work with most XL models. This LoRa has been trained with a variety of images and therefore can properly generate a variety of content, be it landscapes, nudity, still life of whatever.Well, I leave it there and I hope it is of some interest to you.Regards.PD: sorry about the crazy prompts in the provided images as they are random ones generated with the amazing One Buttom Prompt script as i am too lazy to think them myself :P

Quick Oil On Canvas

New version trained with some aditional watercolors and charcoal sketches for a more pronounced and variated artisty effect.

Files with SHA256: 4f88cb8b33bf919bc79d1c84e9bf451189643c9e7b4eec192f75dedba6f3b90e