Asixa
RFGen will continue without RFLoRA？
Hello, why did you guys decide to cancel RFLoRA? In my opinion, RFLoRA and PIR are the core of this model. They make the model versatile, and that's incredibly meaningful. As mentioned in the paper, there's no need to limit oneself to a specific diffusion model.
I'm currently a master's student, and I'm eager to learn about your RFLoRA so that I can generate point cloud images with complex backgrounds. Additionally, I would like to explore fine-tuning methods for the diffusion model.
Thank you. If possible, could you provide a version that can be used in an environment with complex backgrounds?
Hi! Sorry for the long wait — the RFLoRA model is now released. Feel free to give it a try!
Generalization was a known issue in the previous LoRA which delayed the public release, but I’ve retrained the LoRA on a larger scale to address this. The RFLoRA model and sample code are available in version 1.1.0.
