I'm currently experimenting with ReconstructMe's ability to first record depth and rgb to disk and later reconstruct 3D data with the replay option.
My focus is larger volumes, so I adjusted the z range in my cfg file to eg. 500-3500mm. Surprisingly, using the replay option I get geometry up to a "depth" of maybe 2000mm only, even if there is clearly recognizable geometry beyond that range.
I cross-checked the same cfg file with the live scan option, and with that setting the geometry beyond that range gets reconstructed successfully.
> I'm currently experimenting with ReconstructMe's ability to first record > depth and rgb to disk and later reconstruct 3D data with the replay option.
> My focus is larger volumes, so I adjusted the z range in my cfg file to > eg. 500-3500mm. Surprisingly, using the replay option I get geometry up to > a "depth" of maybe 2000mm only, even if there is clearly recognizable > geometry beyond that range.
> I cross-checked the same cfg file with the live scan option, and with that > setting the geometry beyond that range gets reconstructed successfully.
> Any ideas?
Yes, it might be that for compression reasons we limited the valid range to not extend beyond 2000mm. Will check.
On Tuesday, October 23, 2012 7:39:32 AM UTC+1, Christoph Heindl wrote: > Am Montag, 22. Oktober 2012 21:10:29 UTC+2 schrieb Arayn: > Hi,
> I'm currently experimenting with ReconstructMe's ability to first record depth and rgb to disk and later reconstruct 3D data with the replay option.
> My focus is larger volumes, so I adjusted the z range in my cfg file to eg. 500-3500mm. Surprisingly, using the replay option I get geometry up to a "depth" of maybe 2000mm only, even if there is clearly recognizable geometry beyond that range.
> I cross-checked the same cfg file with the live scan option, and with that setting the geometry beyond that range gets reconstructed successfully.
> Any ideas?
> Yes, it might be that for compression reasons we limited the valid range to not extend beyond 2000mm. Will check.
Do the config options work with the record and/or processing modes? a
Again coming back to running as a dumb device and then post-processing the data on a more capable machine; being able to record everything the sensors pick up, then tailoring the config options during processing would be a boon.
Am Mittwoch, 31. Oktober 2012 08:59:49 UTC+1 schrieb dbro...@gmail.com:
> > Yes, it might be that for compression reasons we limited the valid range > to not extend beyond 2000mm. Will check.
> Do the config options work with the record and/or processing modes? a
> Again coming back to running as a dumb device and then post-processing the > data on a more capable machine; being able to record everything the sensors > pick up, then tailoring the config options during processing would be a > boon.
Hi all, i've now checked our code basis. When using ReconstructMe Console, the depth range is limited min:100mm and max 2000mm. These values are currently not changable. You could use ReconstructMe SDK and the recorder example that comes with it to circumvent these regulations. However note that the config files produced by ReconstructMe SDK are newer than the ones by ReconstructMe Console. You should replay them using ReconstructMeQt.
