Welcome to V6 of my workflows.With this release, the previous boxing weight-themed workflows (e.g., cruiserweight, lightweight, etc.) are archived in an included zip file. Instead, I've focused on a single workflow.Included in this workflow is a custom Node for Aspect Ratios. To use it, extract and place it in the comfyui/custom_nodes folder. <div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/nnMbx98He8A" start="0"></iframe></div>Now, you can use SDXL or SD 1.5/2.0 models. I've retained the dual sampler approach introduced by SDXL, commonly referred to as base/refiner. However, I've dropped the SDXL 3-stage Prompt Process (positive, supplementary, and negative) for backward compatibility.Key features retained include:<ul><li>ControlNet module</li><li>Ipadaptors module (aka poor man's LORA)</li><li>Face Detailer</li><li>ChatGPT assistant</li><li>FreeU, LORA loaders &amp; a larger styles option.</li></ul>The workflow has been 'simplified' by removing switches. Instead, modules can be disabled using ComfyUI's built-in Bypass options (CTRL+B on a single Node or 'Bypass all Group Nodes' from the right-click options on a Group). The 'daily driver' view in Green (loaded by default) and 'The Engine Room' (nodes that need not be touched regularly) principles have been retained.By default, when downloading from CivitAI, the following are bypassed:<ul><li>ChatGPT (as API account required)</li><li>FreeU</li><li>ControlNet</li><li>Ipadaptor</li><li>Face Detailer</li></ul>Note: This version is arranged for use with a 1440p 32" monitor. Check the numerous notes and links inside the workflow for troubleshooting."Feel free to adjust based on your personal style and preferences!

Winstons Workflow Collection

This version includes an initial cut of an LCM (Latent Consistency Model) enhanced workflow that uses LCM LORAs to reduce generation time by using less steps . For reference on my 1080ti this reduces the time taken to generate an 1792x2034 image (using upscaling) from approx 330 seconds to approx 150 seconds <a target="_blank" rel="ugc" href="https://www.youtube.com/watch?v=pw8eNpmhT5o">https://www.youtube.com/watch?v=pw8eNpmhT5o</a>

Files with SHA256: 57d06996763b53d2557100a199947e1aba883592e0552577e271c15946f71989