<h1 id="pony-finalcut-2lu6xcynh">Pony FinalCut</h1><h3 id="note-this-model-is-intended-to-be-run-at-far-higher-resolutions-then-currently-offered-onsite.-consider-petitioning-or-or-(or-@justin-if-your-feeling-lucky)-to-add-support-for-such-maybe-they-could-&quot;steal&quot;-from-tensorart-for-once.-x6ism561s">Note this model is intended to be run at far higher resolutions then currently offered onsite. Consider petitioning @theally or @chipshajen or @Cosmic_Crafter (Or <a target="_blank" rel="ugc" href="https://civarchive.com/user/JustMaier">@Justin</a> if your feeling lucky) to add support for such - Maybe they could "steal" from TensorArt for once.</h3>Or maybe ask they didn't make <a target="_blank" rel="ugc" href="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/aa60d62d-4cf5-4e55-a00b-3acafcb90eb0/original=true,quality=100/ComfyUI_00157_.jpeg">high rez images</a> look like garbage without clicking them three times. The perfectly sharp and single pixel hair at 3k has been thousands of hours of work.<ul><li><a target="_blank" rel="ugc" href="https://civarchive.com/articles/16814/2048x3072-is-now-possible-finalcut-joy-edition">JoyCLIP Edition is capable of 3k generation</a></li><li>832x1216 &amp; 1280x1536 native generation size even 1280x1792 is possible for highly trained subjects</li></ul><ul><li>The best quality will come from forcing unet in BF16 &amp; VAE/CLIP in FP32 use this <a target="_blank" rel="ugc" href="https://civarchive.com/models/1763931/easy-comfy-ui-launch-tool">tool to launch ComfyUI</a></li></ul><ul><li>Forge/Auto1111: set COMMANDLINE_ARGS= --unet-in-bf16 --vae-in-fp32 --clip-in-fp32 --cuda-malloc</li><li>Comfy: cmd /k python <a target="_blank" rel="ugc" href="http://main.py">main.py</a> --fp32-text-enc --fp32-vae --bf16-unet --use-flash-attention</li></ul>This model took over 120KwH of energy to train. Not including the 50KwH used to train the CLIP models.<hr />Note: Auto1111/Forge has an issue with stored BF16 VAE in V1.1 BF16 if not using the FP32 VAE command

FinalCut (Pony)

finalcutPony_hybridJOY.safetensors

Files with SHA256: a8369c286cf2faa2259e17e6def50992af723671dc4fddfcfea537c45c469f23