SCROLL DOWN TO SEE VERSION DETAILSWould appreciate comments, posts, reviews, and requestsif you'd like to pay for a commission or just send a dollar or two -&gt; <a target="_blank" rel="ugc" href="https://www.paypal.com/ncp/payment/U32UKLQUV9796">PAYPAL </a>or<a target="_blank" rel="ugc" href="https://www.paypal.com/ncp/payment/U32UKLQUV9796"> </a><a target="_blank" rel="ugc" href="https://ko-fi.com/bloodsplash/commissions">Kofi</a>merged some loras I trained into Rectified Flow noob (loras trained on the model) - <a target="_blank" rel="ugc" href="https://huggingface.co/CabalResearch/NoobAI-RectifiedFlow-Experimental">CabalResearch/NoobAI-RectifiedFlow-Experimental · Hugging Face</a> use the above link for recommended settings and how to get it to work, isn't just drag and drop but is near it. Comfyui workflow is in the files tab or you can click on the above link and download it thereworks best with Comfyui/Swarm or Reforgesampler: Euler A Comfy RF | or any with RF or Comfy (if using Forge, comfyui switches automatically)schedule: simple or NormalVAE: none or one with padding for reflection (not sure which vae to use so I mostly use none, that should use the baked vae of the RF model)need SD3 sampling in the <code>Advanced Model Sampling for reForge</code> tab on reforge to be enabled. 2.5 should work.pos: <code>masterpiece, best quality, absurdres, highres</code>neg: <code>worst quality, normal quality, bad anatomy, text,signature, artist name, watermark</code>Attempted to stabilize it and add a general style boost while maintaining flexibility<hr />Version 1.A - Aestheticmerged loras trained with experimental settings and then attempted to stabilize via a block merge with MiceandDoeV3, anatomy can be hit or miss but the colors come out clean. RF Loras work better than base noob, dunno about v-pred

Files with SHA256: e8ee774af5386200400aededaa48f320c50bd7d86abcaf906fc7e88465d10612