<h1 id="insanium-v1-j9dx56a6i">INSANIUM v1 </h1>Crafted through precision block-layer merging LoRAs from personally curated datasets in order to invoke surprising fidelity, style versatility, and prompt adherence. Built on top of the versatile powerhouse; <a rel="ugc" href="https://civarchive.com/models/1369089/illustrious-xl-20">Illustrious 2.0</a>, this model excels when generating at higher resolutions. 1.25 - 1.5 megapixels will remove your need for detailers, and create deep, intricate scenes. For best results, Use this model with my <a rel="ugc" href="https://github.com/jurdnf/ComfyUI-JurdnsIterativeNoiseKSampler">Iterative Noise KSampler Node</a>! It uses a unique method of noise injection per a user-determined step interval which enables Illustrious models, and this model in particular, to create far more fine details with impressive coherence. my preferred settings: 5.5 cfg 40 steps Sampler: Euler Ancestral Scheduler: sgm_uniform add noise per steps: 2 starting noise: 0.24 ending noise: 0.00 1280 x 1280 total pixels.

Files with SHA256: 6155de0fd51138f892995c5815b1abf2b1bbf3e588d5ee333203a3afc00c6e96