Possible fixed V2 depth anything model released

[donrikk]

@nagadomi it seems like this new model which comes in base large small and small-transparent fixes the issues that plague depth anything V2 when it comes too blurred objects and transparent object in source videos. I have not tested it yet but thought I should bring it to ur attention since there are checkpoints out on there huggingface page. I do not think they are diffusion based so it may be something u may be interested in looking at, I hope u have a fantastic holiday season thanks again for all the work u do on this program!
https://promptda.github.io/

[loawizard]

This looks great! Would love to see it implementated if possible

[nagadomi]

It is technically impossible regardless of my personal opinion.
Google what LiDAR is.