<h2 id="welcome-to-chad!-the-collection-of-homoerotic-attire-for-dudes.-mv5mcxu9g">Welcome to CHAD! the Collection of Homoerotic Attire for Dudes.</h2>These LoRAs have been created to help add a bit of sass to the clothing that male characters wear on their sexy, generative bodies. My appologies to anyone who is unfamiliar with the euphemism, "Wife Beater" as it is sometimes used to describe a certain type of white tank top in the US. That term has fallen out of favor, but since Trumpy cancelled my cultural sensitivity training, I'm sticking with it. This particular LoRA, Wife Beater, like the other CHAD LoRAs works best using the euler sampler, the beta scheduler, 16-30 steps, and a strength between .5 and 1 depending on what you want. With a strength around 1 or greater, you will lose guidance and some artifacts could appear. Play around with strength and guidance until you get what you are looking for. Stick to the euler or deis samplers, others like dpm_fast create consistently awful images. None of the reference images contained frontal nudity, so you can expect the usual mangled genitalia that Flux produces if you don't add a NSFW LoRA or checkpoint. As far as the prompts go, the trigger words have been added to the LoRAs, but they don't seem to have much effect; the details of your prompt shape the image.All of the CHAD LoRAs were created using the Flux-LoRA-Trainer-2.0 in the ComfyUI web app. The trainer workflow can be found on the CivitAI site, here: <a target="_blank" rel="ugc" href="https://civarchive.com/models/1180262/flux-lora-trainer-20">https://civarchive.com/models/1180262/flux-lora-trainer-20</a>The trainer can be kind of a memory hog, so if you are working with less than 24GB of VRAM, you should check-out this brilliant article on managing VRAM for this program: <a target="_blank" rel="ugc" href="https://civarchive.com/articles/9487/managing-vram-to-optimize-performance-for-flux-training">https://civarchive.com/articles/9487/managing-vram-to-optimize-performance-for-flux-training</a>

Files with SHA256: fe06cbd5c8b5443c0a71b9eaed6c68b8e530da0ffbb220ccb9a229111015b780