<h2 id="turbo-lora-distillation-of-kirazuri-(anima)-v4.0">Turbo LoRA distillation of <a target="_blank" rel="ugc" href="https://civarchive.com/models/2495369/kirazuri-anima?modelVersionId=3112659">Kirazuri (Anima) v4.0</a> </h2>Important: This LoRA is not expected to work with other base models. Generation Settings:4 or 8 steps (see model version names), CFG 1sampler/scheduler:<code>res_2s</code> + <code>bong_tangent</code><code>res_multistep</code> + <code>simple</code><code>dpmpp_2m</code> + <code>sgm_uniform</code>Resolution: 1536^2 recommended for best resultsOtherwise, please refer to recommendations on the <a target="_blank" rel="ugc" href="https://civarchive.com/models/2495369/kirazuri-anima?modelVersionId=3112659">Kirazuri (Anima) </a>base model card.Training details (Latest - v4.0):Created with DP-DMD diversity-preserved few-step distillation.Training dataset of 959 images generated by the teacher model Kirazuri (Anima) v4.0 with fully synthetic and partially synthetic images generated with img2img.Generated at 1536^2 with a diverse set of prompts intended to increase the style coverage and diversity of the base model.Selected only generations without faults for training (i.e. anatomical or compositional errors).Thanks to sorryhyun for the trainer: <a target="_blank" rel="ugc" href="https://github.com/sorryhyun/anima_lora">https://github.com/sorryhyun/anima_lora</a>

Kirazuri (Anima) Turbo

Same as v4.0 dataset/parameters, using an 8 step target instead of 4

kirazuri-anima-turbo-v4-8step.safetensors

Files with SHA256: 26886aec1f5e5105f96d8cc2e351c788110c3339ddf4abaabae95677111ecb1a