We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
in the backward.cu, why the abs only added for .z and .w gradients, but not for .x and .y?
` // Update gradients w.r.t. 2D mean position of the Gaussian atomicAdd(&dL_dmean2D[global_id].x, dL_dG * dG_ddelx * ddelx_dx); atomicAdd(&dL_dmean2D[global_id].y, dL_dG * dG_ddely * ddely_dy);
// Homodirectional Gradient atomicAdd(&dL_dmean2D[global_id].z, fabs(dL_dG * dG_ddelx * ddelx_dx)); atomicAdd(&dL_dmean2D[global_id].w, fabs(dL_dG * dG_ddely * ddely_dy));
// Update gradients w.r.t. 2D covariance (2x2 matrix, symmetric) atomicAdd(&dL_dconic2D[global_id].x, -0.5f * gdx * d.x * dL_dG); atomicAdd(&dL_dconic2D[global_id].y, -0.5f * gdx * d.y * dL_dG); atomicAdd(&dL_dconic2D[global_id].w, -0.5f * gdy * d.y * dL_dG); `
The text was updated successfully, but these errors were encountered:
No branches or pull requests
in the backward.cu, why the abs only added for .z and .w gradients, but not for .x and .y?
`
// Update gradients w.r.t. 2D mean position of the Gaussian
atomicAdd(&dL_dmean2D[global_id].x, dL_dG * dG_ddelx * ddelx_dx);
atomicAdd(&dL_dmean2D[global_id].y, dL_dG * dG_ddely * ddely_dy);
// Homodirectional Gradient
atomicAdd(&dL_dmean2D[global_id].z, fabs(dL_dG * dG_ddelx * ddelx_dx));
atomicAdd(&dL_dmean2D[global_id].w, fabs(dL_dG * dG_ddely * ddely_dy));
// Update gradients w.r.t. 2D covariance (2x2 matrix, symmetric)
atomicAdd(&dL_dconic2D[global_id].x, -0.5f * gdx * d.x * dL_dG);
atomicAdd(&dL_dconic2D[global_id].y, -0.5f * gdx * d.y * dL_dG);
atomicAdd(&dL_dconic2D[global_id].w, -0.5f * gdy * d.y * dL_dG);
`
The text was updated successfully, but these errors were encountered: