This model is an IlustriousXL extension of my previous PonyXL models. It retains a high creative elasticity typical of IlustriousXL models but with a high degree of realism. It accepts complex prompts and high weights without creating artifacts or collapsing as long as you don't overdo it. It accepts Pony and SDXL LoRAs well or at least the majority. In any case I recommend not overdoing it with LoRAs to avoid incurring artifacts.recommended parametersSampling method: Euler A ( DPM++ 2M)Schedule: Automatic (Karras)Step: 25-35Skip: 2CFG Scale: 6-7 max 7.5Upscaler: Reamcri oroginalHires steps: 20-35Denoising strength: 0.3-0.4Upscale: Based on the available VRAM on A1111 : (with VRAM 12GB) -&gt; aprox. x1.7.On ComfyUI you can use Upscale x3It is recommended to use the VAE: fixFP16ErrorsSDXLLowerMemoryUse_v10.safetensors (Thank's to Mtigo for sagnalation)

Files with SHA256: 90d89e1627927ab83a8f5a2750a35216b277deae17db93772936d469cb3f7d52