have you try to apply it in real world dataset?

[lagrange10]

Thanks for your excellent work!

I want to test it in real world but at first I must write a LLFF dataloader by myself. Do you think it will work well in some place like textured market meets Manhattan assumptions? Although it is a simple geometry structure, how many details can be presented when it comes to new view synthesis?

Best Regards, Tao

[flamehaze1115]

Hello, thanks for your interest in our work. Since our model is a learning based method, if your testing data is very different from the training DTU dataset , I think the performance will not be very good. Moreover, since SDF is good for closed surfaces, so our method works well for objects. It will be challenging to handle large scenes.

[lagrange10]

thanks for your response. I understand this method may be challenging to a large scenes, thanks. If there is a pretrained model in the target data domain(e.g. indoor scenes with simple geometry but with texture; about 5-10m scale; like the figure below), will it perform not badly?