Introduction

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 an Illustrious model, WAI-illustrious-SDXL v17.0, with SDXL models, namely RealVisXL v5.0 and UltraEpicAi Realism v2.0, with the help of Hako-mikan's SuperMerger; in addition, I merged in (at half strength) some LoRAs trained on a personal dataset of around eight hundred photos. My hope is that the IRE will serve as a foundation for better ones in the future.

Usage

Summary:

• VAE: already baked in

• Prompts: use Booru tags

• Sampler: euler a

• Scheduler: karras

• Steps: ~30

• CFG: 2-5

• Resolution: 1024x1024, 896x1152, and 832x1216

For the best results, you should use Booru tags, placing the following tags at the start of your positive prompt:

masterpiece, best quality, realistic, photorealistic

Including the tags below in the negative prompt may also lead to better quality:

bad quality, worst quality, sketch, flat color, lowres

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.

Credits

IRE is based on block merge of WAI-illustrious-SDXL v17.0, RealVisXL v5.0, and UltraEpicAi Realism v2.0. You are welcome to create new merges using this model without crediting me, but please do not use them for commercial purposes.

Version History

• 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.