Hey everyone, I just tested Flux Fill OneReward, and it performed much better than the Flux Fill model from Black Forest Lab. I created an outpainting workflow to compare the fp8 versions of both models. Since outpainting is more challenging than inpainting, it's a great way to quickly identify which models are more powerful. If you're interested, you can download the workflow in the attachments. You can also get the fp8 version of the OneReward model here: https://huggingface.co/yichengup/flux.1-fill-dev-OneReward/tree/main