Create sensor from e57

[Laserbrain]

Cool!

I had a chance to try it for a minute. It kind of works.

Looks like the 'sensor' position CC is generating isn't getting put in quite the right spot.

The scanner head data listed in the e57 exported from FARO scene:
<translation type="Structure">
<x type="Float">-9.36669992614051372e-01</x>
<y type="Float">1.90614000563962516e+00</y>
<z type="Float">1.19162517055379635e+00</z>
</translation

Hopefully these screenshots help. You can see the scanner head in FARO Scene and CC are in slightly different places. I didn't insert the screenshot twice. The insert attachment is just acting funky.

CC_screenshot_scan000_.jpg (204.04 KiB) Viewed 19102 times

CC_screenshot_scan000_.jpg (204.04 KiB) Viewed 19102 times

[daniel]

When you edit/modify the sensor object (Edit > Sensor > Edit) do you read at least the right 'center'?

[Laserbrain]

I didn't know about that feature. Here are the sensor parameters for that same scan:

CC_screenshot_sensorParameters_scan000_.jpg (39.87 KiB) Viewed 19088 times

[daniel]

Interesting... I'm not sure where the difference comes from. Can you maybe send me your file? (admin [at] cloudcompare.org)

[daniel]

Ah, so I as feared it was a bug (nice catch ;) ).

I fixed it and updated the 2.11.alpha version.

[Laserbrain]

It works!

Terrific, what a useful feature.

Thanks for implementing!

[Laserbrain]

Just wanted to follow up on this feature.

It works as stated and is amazingly useful... but it takes a VERY long time with larger scans.

scans in the <40mb range compute quickly

scans in the 100mb range can take much longer. 30 minutes. An hour. I've left some larger scans to compute overnight and it never completed.

It feels like it takes exponentially longer the larger the e57 is.

I've built a tool in Houdini that brings in these same e57 files. I create the scanner head location (from the structure date) by hand as a point in space and then compute the normals of all of the points to point at that point. But in this case it only takes seconds. Maybe 5-30 seconds for any of the scans I've thrown at it.

In any case, it'd be great if CC could compute these faster. Thanks

[daniel]

In CC we still compute the normal with the given 'radius' parameter. We only use the scan origin to orient the normal. So for large scans you may want to reduce the radius...

[Laserbrain]

Thanks, I'll give that a try.