<h3 id="hi-everyone!">Hi everyone!</h3>This is the third version VividNS LoRa.The third version attempts to improve details and concepts. However, not all of the new categories are working perfectly just yet.Almost all images for training are either Sensitive or NSFW!<h2 id="parameters-i-used-for-the-generations">Parameters I used for the generations</h2>- Lighting LoRA: 8 steps. <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v/Qwen-Image-2512-Lightning/tree/main">LINK</a>- Steps: 8-10- Sampler: res_2s / euler a / deis_3m- Scheduler: beta57, normal, simple<h3 id="i-highly-recommend-using-lighting-lora-8-steps-or-4-steps.">I highly recommend using Lighting Lora 8 steps or 4 steps.</h3><hr />This LoRA is not meant to be completely photorealistic.The main goal is to explore and unlock things that were not possible with the original model.The training set included 1200 images, and none of them were taken from real people.

Files with SHA256: 3fedcc21ccdc49cbf18a7cf4e6241a6deaf3b564b50c7bffe17014af952c024c