So I getting a little bit in-depth about FLUX model and it seems that this model emphasize natural language rather than using a simple tags and the good thing is FLUX able to produce much more clearer image compared to SDXL models which I used to do it.Anyway, I made a FLUX LORA specifically to generate an image of woman with super silky very long hair, usually from waist length up to full body length if got lucky. Since I'm still a little bit new on FLUX so this LORA might be a little bit experimental.Guide on using this:<ol><li>Use SD WebUI Forge (not reForge) or ComfyUI to use FLUX model.</li><li>Set the sampler to Euler ONLY and the schedule type must be set to Simple ONLY.</li><li>Distilled CFG should be at around 3.5 while CFG must be at 1 only and steps should be at around 30.</li><li>Prompts must be written in natural language form and it should be like this, the first part are the trigger words:<pre><code>super long hair. a woman with silky red extremely long hair styled with bangs and wears casual clothing and sitting on a bench on the street</code></pre></li></ol>Also, please note that FLUX model generation requires an Nvidia GPU with at least 12GB VRAM and FP8 mode must be enabled.PS: You will need to learn on how to construct the natural language prompts as simple and good as possible without using AI...

Files with SHA256: e35ca5cfdbc3dddb065fcb0a482c8e7fd14107d77b12d0e5e0208197a7fc1d9a