photoscanも結構豊富に作れて早いのだが、ここは細かな点群を調整する意味でcloudcompareを使う
これも手順が必要
まず、
メッシュ化に前に不要な点群を間引く
photoscan also fast to it and make quite rich, but here use the cloudcompare in the sense to adjust the fine point cloud
This also requires a procedure
First,
Decimating unnecessary point cloud before the meshing

Edit >> Subsamples
不整形三角網の平均化だねきっと。
三角網の平均化で同じような近傍点で細かな三角網はいらないって事かな。
Edit >> Subsamples
It is the averaging of the irregular triangle net surely.
I wonder if a triangular mesh is averaged and you do not need a fine triangle at similar neighbor points.

planeが最適
plane is optimal

quadric

Trianglation

autoをクリック
最適な大きさを導いてくれます。
Click auto
It leads to the optimum size.

この点群は、画素数から換算すると0.03の点群なのでまあまあって事で

後は、そのままでOK
i7-3930
64GB
GTX670
5年前の機器でも十分なスピードですね。
で３分程度
photoscanはこの手順抜きで早いけど・・
Since this point cloud is a point group of 0.03 when converted from the number of pixels, it is so so to be so

After that, OK as it is
i7-3930
64 GB
GTX 670
It is enough speed even with equipment five years ago.
In about 3 minutes
Photoscan is early though this step, but …

こんな感じ
Like this