confused: from sparse cloud to dense cloud

[renevg]

So, I'm confused
I went from sparse cloud to dense cloud. In sparse cloud I completely deleted a part of the scanned object, but in dense cloud it's back! So where does it come from? I totally deleted those tiepoints. I thougt that dense cloud is generated from sparse cloud. Can anyone enlighten me?
Thanks

[Kiesel]

Hi renevg,

the sparse cloud is the result during the aligning step. You can then use 'Model/Gradual Selection' to edit it for:

• Reprojection error
• Reconstruction uncertainty
• Image count
• Projection accuracy

Don't forget to reoptimize your cameras after that to get the improved camera extrinsics (positions and orientation) and intrinsics (internal camera parameters).
With this new parameters (extrinsics+intrinsics) the dense point cloud will be generated, the sparse cloud isn't used for that at all. Btw. it is the same for the step from the dense cloud to the model, where an edited dense cloud has no impact to the model (you will see the out edited parts in the model again).

So if you want to exclude something in your dense cloud or model you need to mask those parts in the input photos.

Best regards,

Karsten

[renevg]

Thanks for answering Karsten,
I've been working with metashape pro for many years now, but never came across this issue. I've always been using model /gradual selection and optimizing I've always done. But I just don't understand why I didn't see this problem earlier. So hence the confusion. I haven't used combining chunks very often up to now, what I'm using out editing for now. So masks it will be in the future.
btw I couln't reproduce your statement "it is the same for the step from the dense cloud to the model, where an edited dense cloud has no impact to the model (you will see the out edited parts in the model again)."   Once out edited in dense cloud it stayed out edited in mesh. So confusion again, but this time it's the way I want it to be.
Thanks again
René

[Kiesel]

Hi René,

Sorry for the confusion. My sentence:

Btw. it is the same for the step from the dense cloud to the model, where an edited dense cloud has no impact to the model (you will see the out edited parts in the model again).

wasn't precise enough.

It depends on the method you choose for model calculation. When you choose
a) from depth maps, my sentence is right
b) from dense cloud, the out-edited parts in the dense cloud aren't used for model calculation.

Hope it makes more sense now.


Regards.

Karsten

[renevg]

Ah yes that would make sense.
Thank you
René