Ruicheng Wang

您好！训练使用的深度图是特定格式的16bit png，存储的是缩放到了 1~65534 之间的log(depth)值，同时将 near 和 far 存储到png的元数据中用于恢复线性深度。您可以参考[moge/utils/io.py](moge/utils/io.py)中的 read_depth 和 write_depth 函数。确保得到(H, W)float32线性深度，无效的深度值填np.nan，无穷远深度填np.inf，再用write_depth保存，即可存成对应的格式。 一般内参矩阵的参数都是正的。如果是cx cy 有负数的话，大概只是光轴比较极端，不在图像内了，这种问题不大。fx fy是负数会出问题，这一般不太可能，除非图像坐标的原点没有定义在左上角，或是外参的旋转被混入到了内参中，这种情况可以尝试直接对fx fy取反，再反投影成点云检查一下。

Hi, thanks for raising the valuable question! - **Measurement Method:** We measure the runtime of the .infer() function using `torch.cuda.synchronize(); time.time()`. This timing includes the entire inference pipeline: image preprocessing,...

Hi, I just noticed that the current code **falls back to standard matrix multiplication-based attention** when xformers is not installed. This was caused by an inconsistent historical commit (while the...

Hi. The scale (and shift) of prediction $\\{\hat p_i\\}$ are aligned with the ground truth $\\{p_i\\}$ as follows. * For evaluation of scale-invariant point map, $$ s^*=\\mathop{argmin}_s \\sum_i {1\over z_i}\\|...

Hi! Apologies for the late reply, and thank you for providing this great experiment! Currently, MoGe is trained on data with a limited range of FOV, typically between 30 and...

Hi, thanks for your interest! Inference with user-provided intrinsics is a good idea and can be quite useful. It was discussed in this issue https://github.com/microsoft/MoGe/issues/24#issuecomment-2493184207. Note that the model itself...

Hi, thanks for your question. We used radius=1 in our final evaluation, following Depth Pro. The other two settings were only for our own exploration. As a disclaimer: the evaluation...

Hi. Thanks for your interest and this valuable question. The evaluation results are different because the datasets are processed differently from these works. We have meticulously processed the raw evaluation...

Apart from the differences in data processing, it is also crucial to consider the evaluation configurations, such as 'metric depth', 'scale-invariant depth', 'affine-invariant depth', and 'affine-invariant disparity (inverse depth)'. These...