Update - 28/02/24 - There is a new mix model, instead of the Pony model. I have updated links.Going to keep this brief because I'm too lazy to make a full writeup. This model and training method far surpasses anything we've seen with NovelAI from 2023. It trains at 1024x1024 based on SDXL.You can refer to <a target="_blank" rel="ugc" href="https://civarchive.com/models/62125/lora-lazy-training-guide">https://civarchive.com/models/62125/lora-lazy-training-guide</a> for tagging etc.Refer to <a target="_blank" rel="ugc" href="https://github.com/bmaltais/kohya_ss">https://github.com/bmaltais/kohya_ss</a> for the kohya link.<ol><li>Download <a target="_blank" rel="ugc" href="https://civarchive.com/models/288584/autismmix-sdxl">https://civarchive.com/models/288584/autismmix-sdxl</a></li><li>Download <a target="_blank" rel="ugc" href="https://files.catbox.moe/dvx8me.json">https://files.catbox.moe/dvx8me.json</a> or the .json from the attachments - When opening Kohya, use this as your configuration (click on config at the top on the LoRA tab, open, select this json).</li><li>Point the training model to the Pony model.</li><li>Change your save location to wherever you have your LoRAs.</li><li>Tag all of your images with score_9, source_anime. Set your folder repeats to 1.</li><li>Train.</li><li>When generating on the above pony model, put scores 6 through 4 into your negative prompts, and add 'score_9, score_8, score_7, source_anime' to your prompt.</li><li><s>I </s><s>STRONGLY</s><s> recommend using an artist LoRA. Try </s><a target="_blank" rel="ugc" href="https://rentry.org/ponyxl_loras_n_stuff"><s>https://rentry.org/ponyxl_loras_n_stuff</s></a><s> for LoRAs made specifically for the pony model. </s></li></ol>NOTE - You no longer NEED to use an artist LORa with the fintune of the 'autismmix' model for Pony. It generates anime out the gate.Generate. It should look good. The results speak for themselves. The pony model is the best we have for local anime at the moment. It blows base NovelAI out of the water.

Files with SHA256: a52998bc095c7e17b6563c2d6f56a5b01c3a0dc99641d4e1ad05ba8210b276af