Stable Diffusion

OpenSolera α3

model_id: opensolera3

<p>Internal name: OpenSolera-a3-bless09-rf7-1899-0041-3929.safetensors</p><p>Yes, this model has gone through Model Toolkit pruning already. "0041" is WD1.4-bless09 VAE, "3929" is the CLIP from the LoRA merge step.</p><p><strong>Changelog:</strong></p><ul><li><p>Additional models added: Furryrock v_7.0 (<a target="_blank" rel="ugc" href="https://civitai.com/models/71432?modelVersionId=203021">CivitAI</a>) and Ken Sugimori style LoRA (<a target="_blank" rel="ugc" href="https://civitai.com/models/5115/pokemon-lora-ken-sugimori-style-for-fakemon-and-characters">CivitAI</a>)</p></li><li><p>Experimental CLIP replacing (see <a target="_blank" rel="ugc" href="https://civitai.com/articles/7030/personal-experiment-what-happens-if-you-change-a-models-clip-with-a-different-one">this article of mine</a> for more info)</p></li><li><p>Better multi-person prompts (faces are slightly less distorted at further distances, but Hi.res fix is still recommended)</p></li><li><p>Maintained decent hand quality (at least the fuck-ups aren't <strong><em>as</em></strong> bad in my experience)</p></li><li><p>Prompt adherence is... a bit worse? Negative prompts are more required than before, along with detailed prompts. </p></li><li><p>Flatter base style than Alpha 2. For those looking to mimic Alpha 2's base style, use "flat color" in negative prompts (see example image).</p></li><li><p>Better furry generation (mostly due to Furryrock). It's still not great, but it's getting better... I guess.</p></li></ul><p>As usual, for people wanting to check out the model's merge recipe, download the "Training Data" zip file.</p><p><strong><em>Important side note: </em></strong>I recommend using CLIP Skip 2 for this model, as it's the closest to what I intend the base style to be like (with "flat color" in negatives).</p>