<h1 id="silvana-and-tamra's-style-mix">Silvana and Tamra's Style Mix</h1>This one is different from the previous Style Mix Lora as it wasn't trained on a mix of different images of each styles separatly but trained with images produced with @Silvana_ 's Style Mix Version of our Style : <a target="_blank" rel="ugc" href="https://civarchive.com/models/2732862/tamra-and-silvanas-style-mix">https://civarchive.com/models/2732862/tamra-and-silvanas-style-mix</a> How? As i said, it was trained with a dataset produced by the Tamra &amp; Silvana's Style Mix Lora. This dataset of more than 800~images, tagged in Natural Language only. Why ? Because i wanted to reacreate this style, but closer to the way i use it, in the hope to make my prompts lighter. Results?Well i'd say "Judge by yourself, what do you think", but in truth, it's not terrible if i had to say so myself. Yes, i hear you (fictional character in my head) ! It can give some weird results, sometimes... try using only 1 of the tags! or change the order you use them in! use one twice!!!! Who knows what it will do !? Play with it, make it your own ! And above all else, have fun with it !<h3 id="how-to-use">how to use</h3>Mostly, surely, like any Anima Lora i would suggest tho, to use <code>er_sde</code> as the sampler maybe around 3 to 4.5 CFG and around 30 stepsbut whatever works for you already would surely work too ^^

Files with SHA256: 45e1a94e2deec51aa7926acc47a3fff5a22ab0536cb51f4d2df96a9ada1b79ec