<h3 id="introduction">Introduction</h3>The Illustrious Realism Experiment (IRE) is an attempt at a realistic Illustrious model that does not sacrifice its character knowledge, prompt adherence capabilities, and compatibility with Illustrious-based LoRAs. This was mostly accomplished by block merging <a target="_blank" rel="ugc" href="https://civarchive.com/models/827184/wai-illustrious-sdxl?modelVersionId=2883731">WAI-illustrious-SDXL v17.0</a> with a mix of <a target="_blank" rel="ugc" href="https://civarchive.com/models/139562/realvisxl-v50?modelVersionId=789646">RealVisXL v5.0</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/1721070/ultraepicai-realism">UltraEpicAi Realism v2.0</a>, and <a target="_blank" rel="ugc" href="https://civarchive.com/models/1125067/cyberrealistic-cyberillustrious">CyberRealistic CyberIllustrious v11.0</a> using <a target="_blank" rel="ugc" href="https://github.com/hako-mikan/sd-webui-supermerger">Hako-mikan's SuperMerger</a>; in addition, I merged in a custom LoRA trained on a personal dataset of high quality photos.<h3 id="usage-guidelines">Usage Guidelines</h3>Summary:<ul><li>VAE: already baked in</li><li>Prompting: use Booru tags (see below for quality tags!)</li><li>Sampler: euler a</li><li>Scheduler: karras</li><li>Steps: ~30</li><li>CFG: 2-5</li><li>Resolution: 1024x1024, 896x1152, and 832x1216</li></ul>For the best results, you should use Booru tags and place the following tags at the start of your prompt:<pre><code>masterpiece, best quality, realistic, photorealistic</code></pre>Including the tags below in the negative prompt may also lead to better quality:<pre><code>bad quality, worst quality, sketch, flat color, lowres</code></pre>The preview images used the euler a sampler with the karras scheduler at 30 steps, but feel free to experiment with other options (the DPM++ 2M SDE sampler, for example, can produce sharper images). The recommended CFG scale is 2-5, with a lower CFG tending toward more realism, in my opinion. Good resolutions include 1024x1024, 896x1152, and 832x1216.While the model should be compatible with most Illustrious-based LoRAs, results will vary depending on the LoRA, and you may need to lower the LoRA's strength and/or the CFG scale.The model has the default SDXL VAE baked into it.<h3 id="credits">Credits</h3>IRE is based on block merge of <a target="_blank" rel="ugc" href="https://civarchive.com/models/827184/wai-illustrious-sdxl?modelVersionId=2883731">WAI-illustrious-SDXL v17.0</a> with <a target="_blank" rel="ugc" href="https://civarchive.com/models/139562/realvisxl-v50?modelVersionId=789646">RealVisXL v5.0</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/1721070/ultraepicai-realism">UltraEpicAi Realism v2.0</a>, and <a target="_blank" rel="ugc" href="https://civarchive.com/models/1125067/cyberrealistic-cyberillustrious">CyberRealistic CyberIllustrious v11.0</a>. Free free to create new merges using this model, but please do not use them for commercial purposes.<h3 id="version-history">Version History</h3><ul><li>v4.1: Improved LoRA compatibility by making some slight changes to the merge formula.</li><li>v4.0: Made significant changes to the block merging formula in an attempt to improve visual fidelity and realism; incorporates a custom LoRA trained on a personal dataset of around two hundred photos.</li><li>v3.0: Reduced graininess by integrating CyberRealistic CyberIlustrious v11.0 into the merging process; incorporates a custom LoRA trained on a personal dataset of around two hundred photos.</li><li>v2.0: Some improvements to visual fidelity by incorporating RealVisXL V5.0 and training on a higher quality dataset for the LoRA merged into the model; incorporates custom LoRAs trained on a personal dataset of around eight hundred photos.</li><li>v1.0: Initial version based on a block merge of WAI-illustrious-SDXL v17.0 with UltraEpicAi Realism v2.0; incorporates custom LoRAs trained on a personal dataset of around six hundred photos.</li></ul>

Illustrious Realism Experiment (Compatible with Illustrious LoRAs)

Improved LoRA compatibility by making some slight changes to the merge formula.

Files with SHA256: 48c5c36886104fa655720add70d2eb0582f227bc6d976aea313ef08681bf7fdf