This is an SDXL LoRA trained on a diverse set of analog photographs.I typically use this LoRA at 0.6-1.0 Strength. In your prompt, use the activation token: <code>analog style</code>Sometimes pushes a green tint too hard, in these instances you can use negative prompting such as <code>green</code> or <code>split-toning</code> to help. Some prompts may struggle to generate consistently in color, in these instances you can use <code>sepia</code> and <code>grayscale</code> in negative to help. Trained on sd_xl_base_1.0_0.9vae.safetensors . All examples are generated on sd_xl_base_1.0_0.9vae.safetensors using this LoRA. No hi-res fix, upscaling, inpainting, etc. done on any samples here.I have tested this LoRA with some popular community Photo SDXL checkpoints and gotten good results, you just may have to play around with the LoRA Strength to achieve your desired result (may need &gt;1). More info @ <a rel="ugc" href="https://huggingface.co/wavymulder/Analog-Diffusion-XL">huggingface</a>

Files with SHA256: 4e984de0fd9ef7d469cea66fb78a16ada746304c9a2ffb86270d9e182068797f