FBP scaling issue

[ozanoktem]

When running ODL FBP example, like the example for 2D parallell beam acquisition, the resulting FBP reconstruction is off by a (large) constant factor. In the example, phantom values range from 0 to 1 whereas FBP reconstruction values range from 0 to 0.002.

FBP should not behave like this, and especially not in 2D parallell beam acquisition setting where it is theoretically exact.

[Emvlt]

Can you please provide the code you tested and the odl version you are using? I cannot reproduce your error using the script that odl provides

[leftaroundabout]

This is probably related to #1666. Please check if #1668 fixes it.