try to fill the picture more with the object, you are covering more background instead of the object itself
hide the hole background, try to make it black or green, the color doesnt matter, the software detects to many features i think.
i think the biggest problem here is the background, on each cycle the background is more prominent than the objeckt. try to hide it with back fabric, so there are no features to detect for the software.
Edit: i tried also a turntable setup one time, had the same issue, so the only problem was the background, i coverd it all with white fabric, so the software only “sees” the object, not any detail of the background.
I like your “truntable” like setup very much, well done
By the looks you do not adjust the reconstruction region to only cover the area of interest before you start the reconstruction. My advice is to do that after alignment.
That way the back wall becomes a non-issue as any features detected there are discarded.
Also, the statue or bust itself is pretty much featureless which can cause problems in itself.
When the alignment is initially broken, setting reconstruction area does not help i think. The aligned must be correct in order to get a correct reconstruction. I would also suggest to use the forum search and type “turntable”, there are many post like this one here
you are right the software needs a backround, but you make one mistake here, you have to rotate the hole background with your object.
so far i have understood, you make a pass with your cam on your lego train, then you rotate your object. in the next pass you are capturing the rotated object, but the feature rich background is at each pass the same, thats the problem.
Did you try to import only one pass of images, without rotating the object? i think then the images should align correctly.
And the software needs features to detect, the software makes no decision what is a background or an object you want to capture. So no background, no features, no problems
your rotating the object and leaving the background stationary so all the alignment is to the background, Add a plate to the base of the object so that rotates the same as the object and remove the background to be black
They wrote:
“To explain the masking in RealityCapture - even we do not have a tool for editing masks build in our app, the app itself supports masks to some extent. For alignment step all you need to do is to modify image color channel and paint it with a solid color. This will cause that natural features will not be detected there and thus will not influence camera alignment. For meshing we support alpha-channel masks. Adding alpha channel and masking only important parts actually speeds up whole computation. Another benefit is that you would not need to use reconstruction region to filter parts which are not important.”
So just fill the background with a black color. As i mentioned before, its the same like taking the photos on a solid colored background.
