<h2 id="recommended-settings-8y8g29ndl">Recommended Settings</h2><p><span style="color:rgb(193, 194, 197)">Flux.1 Dev, LoRA weight 1.0</span></p><p></p><p>Also available to use for free on <a rel="ugc" href="https://magic.remix.ai/model-preset/03ec09ad-f569-48e8-bee2-7cae364cafc6">Remix AI</a></p><p></p><h2 id="backstory-3tbmxdyz9">Backstory</h2><p>I wanted to see how well it works to use <a target="_blank" rel="ugc" href="https://civarchive.com/posts/7626113">Midjourney images</a> to train Flux LoRAs, so this is an early experiment in that direction. I focused on images that have lots of particles and inherent motion, as well as crisp focus on any explosions themselves. It works out strangely well as a proxy for high-detail generations.</p>

Files with SHA256: c6a309b44455acb202bf5db3b1686687f0caf701e3a9317ba741f3bc74e3d47e