Hi,

I have been using vaa3d 3.20 for some time now to trace cell during experiments, using the "right stroke to define  a 3d curve - ver 2b sriab BBoxes method". I have good results for my tracings. 

After the update to version to 3.447 (that fixes several annoying bugs, thanks!) i noticed that the tracings behaved very differently, to the point that i cannot do my reconstructions anymore (I compared the two versions with the same stack). More specifically, I think that the tracing explore bigger areas around the initial handdrawn trajectory. It tends to select dendrites or elements that are sometimes pretty far from the handraw line. It happens mostly if I trace a thin dendrite and there is a thick dendrite around.

Did you change a sort of tolerance factor, or radius for the algorithm since 3.20?  Is there a way to restore the previous tracing behaviour / tune it in v3.447

Best regards

Antoine Valera

We didn't  change any code related to BBoxes drawing. Could you please send us a testing image and point out the difference when using two different versions?

Hi, Sorry for the long unanswered post,

I was still still using version 3.20 during all that time. I tried again with your most recent version (3.544). I am definitely sure that the tracing behaviour is different in the most recent versions. Tracings with recent vaa3d versions explore a much larger region, and often end finding regions that are way out of the initial tracing. Here is a quick example, but it is very easy to reproduce in many of my stacks. Stacks are coming from 2p in vivo or in vitro recordings, and are a bit more blurry and noisy than confocal stacks. However, it has never been an issue in v3.20

I hope you can help with this issue, that prevents me from upgrading to more recent versions.

Thanks

Antoine