<h1 id="flux-release:-nc9aj9c9s">Flux Release:</h1><p>I learned how to use the on-site trainer. Full release now available. I'm pretty happy with the results. Might be a bit strong, lmk if you are interested in a less strong version. Trigger with Mandelbulb fractal just as before. My examples are a bit blurry, that's not from the LoRA, I'm still working out kinks in my Comfy workflow.</p><p></p><p>(Sorry about the creepy Teletubbies things lol, they pop up whenever you try to make something cute)</p><p></p><h1 id="trypophobia-warning!-zfxkd2jcb">Trypophobia Warning!</h1><p>This model produces images which may trigger trypophobia. I've tagged the model with trypophobia so that those who are sensitive may avoid it.</p><h1 id="-v3pk7w6t7"></h1><h1 id="mandelbulb-xl:-e7xnk98k2">Mandelbulb XL:</h1><p>Is a model trained on aesthetically pleasing mandelbulb images. Stable diffusion produces mandelbulb imagery but I find it produces a fractal mud rather than beautiful geometric shapes. This model was trained on many different looking mandelbulb shapes, so by default it will come out kind of ugly. You should tell it what shape and color you're looking for and then it will actually work well. Mandelbulb, fractal, detailed, intricate, metallic, colorful, fog, shadow, depth of field, mushrooms, natural, closeup, distant... were all used in the tagging process and are therefore most likely to have the most effective impact. Might play well with humans with some trial and error. Plays very well with cats and flowers.</p><p></p><p>Credits:</p><p>Model Trained on Base SDXL model.</p><p>Example images produced with <a target="_blank" rel="ugc" href="https://civarchive.com/models/425521?modelVersionId=481390">Voight Kampff</a> by FreedomForMars and <a target="_blank" rel="ugc" href="https://civarchive.com/models/163139?modelVersionId=481390">Boomer Art Model</a> by ndimensional.</p>

Files with SHA256: d6df1b0b6353fca76cc8bc703ba4c4b75be963b960582a70db0c8496efc61ba5