Bharat Tak

Hey, yes, I plan to get back to this. We can also close the PR and keep the branch until then.

You mean try to catkinize it with the python script ?

Oh cool! Let me give building it a shot! It'll be amazing if it builds. Yep! You are right about the sim3 implementations in gtsam, i spent some time and...

Cool! It builds fine! But I am running into the same Eigen issue (with commainitializer) that I did when I tried to link gtsam as a shared lib in original...

Oh yes! I made a little hack to modify the Eigen include path to the one i wanted and it works.

Hey! I am not sure how familiar you are with lsd_slam(hopefully very!) could you guess how I could calculate the variance for Sim3 transitions for "right-composition" rather than "left-composition" as...

Hey! Thanks for the reply. I agree that right/left composition to make increments in SE3/Sim3 space should both work, and my concern is regarding the 'consistency' of using the obtained...

It does catch the right scale and orientation, but that is probably because gtsam trust GPS factors too much right now(the relative weighting of the gps and Sim3 factor noise...

Ah! I got what you mean. I leave everything as is, sim3 tracker would still compose on the left. I can just calculate the covariance matrix for the right composition...

Hey @JakobEngel ! I was just looking at the formula you mentioned to transform the variance to the correct tangent space and comparing it against the one you've used for...