google
The girl in the green dress always has artifacts near her legs
I reproduced the dataset of the girl in green clothes. There are always some artifacts near girl's legs shown in above image. Could you please share your training config file，I used almost the same config file as the kid-running dataset.
I found that there are some artifacts near the legs in your results, but the situation was better than mine. Is it impossible to completely eliminate or alleviate these artifacts?
Hi I believe this is common artifacts from our system, I believe this is caused by some limitation of static IBRNet for rendering static scene contents at disocculusion with dynamic video. Currently, I don't have an easy trick to fix this issue, and only thing I found is useful for reducing these artifacts is to intentionally adding more source views for static model (since we might need much more further-away frames for feature aggregation for disocculusion contents).
@sjdfkjawe hi there, could you please tell me where do you obtain the dataset from? Is it open to public for research purpose? Would be interesting to know. Thanks in advance!
@sjdfkjawe hi there, could you please tell me where do you obtain the dataset from? Is it open to public for research purpose? Would be interesting to know. Thanks in advance!
Sorry for waiting a long time, you can get this dataset from https://zhmeishi.github.io/dvs/