On Fri, May 19, 2023 at 02:20:11PM +0200, luis vazquez wrote: > I am working with the surface reconstruction tool of Gwyddion. From one > image, I set a model tip (a cone with 70 degrees in slope and 8 nm of > radius) and then I apply the reconstruction tool. What I do not understand > is that the reconstructed image has a lower roughness than the measured one. > In my case, this is more evident when I use 25 nm as the radius of the tip). > In principle, the convolution of the real surface with the tip geometry > should lead to a smoother or less rough image. Accordingly, the > reconstructed one should be rougher than the measured one. Perhaps, I am not > applying properly the tool. > > I note that when I do the opposite, using the dilation tool, I do get an > image with less roughness than the measured one as expected.
First, always be at least a bit distrustful about the reconstruction. It needs to make up information which often just is not present in the data. Tip convolution decreases roughness (σ) only sometimes. For a Gaussian-like rough surface the first effect of tip convolution is in fact an increase of measured correlation length T, whereas the effect on σ is second order. So you cannot really see any change until the tip radius becomes large. For a spiky surface (positive skewness γ₃) tip convolution *increases* measured σ, usually not much but for quite a large range of tip radii. Dilation, unreliable as it is, still kind of reverses the effect. So, it depends. Yeti _______________________________________________ Gwyddion-users mailing list Gwyddion-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/gwyddion-users
