Yasaman Haghighi

Results 9 comments of Yasaman Haghighi

Hello, thanks for your interesting work. Is it possible to upload the annotations to google drive as well?

> Hi! Could you provide some images to illustrate this issue? Hello, ![new](https://user-images.githubusercontent.com/71403412/206190347-d837b1cd-b20f-43aa-a6da-10ab4282c29e.png) ![old](https://user-images.githubusercontent.com/71403412/206190354-afb27a5f-27fb-4609-8a86-76f261767f01.png) The one on top is what I get with the new code and the one below...

I'm using the following code for retrieving color and the [old version](215274e2f99d2481ba00f3345fe3f3dd1b9c3d74) is Automatic calculation of render_step_size. ``` b,_ = mesh['v_pos'].shape sdf, sdf_grad, feature = self.geometry(mesh['v_pos'], with_grad=True, with_feature=True) normal =...

I re-trained it.

Yes, the values are between [0,1]. The only thing that I changed in the config file is the num_samples_per_ray. I tested both 1024 and 2048 values but the color issue...

I can get correct results with the old version and changing view direction doesn’t affect the output.

Thank you for your quick replies. I changed the marching cube threshold to -0.5 and the colors are correct now. Is there a reason for this?

Thank you but I am using the same config file during training and testing.

I don’t have that in my config.