Jianyuan Wang comments

Results 238 comments of


                                            Jianyuan Wang

The Squashing when images are larger than 518

Hi, I haven’t experimented with this exact case yet, so I’ll need to run it myself to investigate further. Could you specify whether your 3D points come directly from the...

The Squashing when images are larger than 518

Hi @ShunyuanZheng can you try first resizing to 336x336 and then black padding it to 336x518?

Some question about BA

Hi, we mainly used BA for refining the camera poses. Specifically, when applying BA, we select a set of 3D points and construct their corresponding tracks. We then jointly optimize...

Some question about BA

According to the recent feedback, it seems the community does have some interest in this. I'll consider it depending on my bandwidth.

Some question about BA

Closing this issue as the BA code has been uploaded.

Discrepancy in 3D Reconstruction With and Without Background Masking

Hi @engrmusawarali, thanks for your interest! I believe the issue is related to the hyperparameter conf_thres in the Hugging Face demo. Instead of using an absolute threshold (like 1.5 or...

Discrepancy in 3D Reconstruction With and Without Background Masking

Hi, yes, it looks ok with point map head, but it would be much better if you adjust the conf thres parameter a bit. For untextured objects, my current observation...

Dear author, I would like to ask if there is a script for visualizing the re-projection error

Hi we do not support this.

Inquiry about DINOv2 Weights and Frozen DINO Experiment for VGGT Training

Hi, We used dinov2_vitb14_reg, initialized it from the official DINOv2 pretrained weights and then fine-tuned jointly with VGGT. As far as I have observed, it can also work but with...

Question about predictions['images'] in vggt_to_colmap.py

Closing this issue as we have provided a new script demo_colmap.py