<h3 id="re-upload-v2-11.1.-knsquwncg">RE-UPLOAD v2 - 11.1.</h3>at least 12 Steps max 25Its a merge of:DEV_Shuttle <a target="_blank" rel="ugc" href="https://civarchive.com/models/943001?modelVersionId=1058817">https://civarchive.com/models/943001?modelVersionId=1058817</a> and FLUXDEV_FP8 <a target="_blank" rel="ugc" href="https://civarchive.com/models/622579/flux1-dev-fp8?modelVersionId=695998">https://civarchive.com/models/622579/flux1-dev-fp8?modelVersionId=695998</a> with my lora distilled <a target="_blank" rel="ugc" href="https://civarchive.com/models/977247/flux-distilled-lora">https://civarchive.com/models/977247/flux-distilled-lora</a>so you can use CFG ~2 (but you don't have to) Q3_K_M running fully on 12GB VRAM dont go above1200pixfor me it is a start to merge models and loras because i cant train a model, iam only a single one v2(FP8,Q3) - nudity, poses on board, BUT dont expect to much fine details, (add freckles, body freckles), all girls look like next door v1(FP8,Q5) - its a well balance between steps (not to low) and CFG 2 for complex promptsEuler with Beta or SGMUniform or (swarmui)DPM++2Sa with Beta (better skin tone) or dpm++2M with BetaTell me please how it works ... Loras should be work at least the main direction (maybe some details gone)some pictures with my Loras to show how it works, my impression some times bit lower weight than usualHere are some comparisons between Steps and CFG<a target="_blank" rel="ugc" href="https://civarchive.com/images/46982169">https://civarchive.com/images/46982169</a> <a target="_blank" rel="ugc" href="https://civarchive.com/images/46984474">https://civarchive.com/images/46984474</a>

Flux_Sevenof9_12Step_Distilled

fluxSevenof912step_sevenof9ShutDistQ5KM.gguf

Files with SHA256: 7249b08724bb53e2da8e9a7617e543d46273e3453ac1c9c6dcad73882fec4ef7