<h2 id="anima-finalcut">Anima FinalCut</h2>INT 8 Version is sourced from FP32 Training and is high quality/high speed<ul><li>100k Iterations on SFW high complexity image/prompt pairs LLM length</li><li>Use of <a target="_blank" rel="ugc" href="https://civarchive.com/models/2673444/anima-llm-layerwise-fp8-fp32">FP32 QWEN</a> and <a target="_blank" rel="ugc" href="https://civarchive.com/models/2718533/hdr-vae-anima-qwen-image">HDR VAE</a> is recommended but not required.</li></ul><h2 id="turbo">Turbo</h2><ul><li>Works at 8-12 steps</li><li>CFG 1-2 ONLY</li><li>Compatible with most samplers</li></ul><h3 id="base-version-(30-steps-ancestral)">Base version (30 Steps Ancestral)</h3><ul><li>Use ONLY with ancestral (a) samplers such as simple/euler a or sgm_uniform/dpm2_a</li><li>FP8 version is available however the quality is reduced greatly (Image quality not accuracy)</li></ul>FinalCut for Anima brings realisim and FLUX style prompting to Anima<hr /><h3></h3>

Anima FinalCut (New INT8 Convrot)

animaFinalcutNewINT8_baseDiffusionA_full_bf16.safetensors

animaFinalcutNewINT8_baseDiffusionA_full_fp8.safetensors

Files with SHA256: ce6abff274d9f76edbb7da809f0ab95a7d6d59cfbdd1ebf3517160517d5d247a