[SOLVED] C2C/C2M : different results if front or back view is selected

[OMG20Centz]

Hi,

I'm pretty sure this subject has already been treated but i can't find any info on the question.
My concern is about the display of cloud comparison. Here is an example :

On the left the front view of a C2M comparison of a flat object, on the right the back view (with a horizontal mirror correction).
The deformations are not really the same on both views and that can be problematic.
My workaround is to subsample the cloud until the results are similar from both view, but it can doesn't lead to the same visual quality (and the problem comes back if i select an important point size), and it is a very stochastic approach as the subsampling implies :


Is there any better/other way to correct that problem ?
I guess the software is not written to handle that problem through the 3D engine (which is totally understandable), but maybe it could be an option (a checkbox for flat objects ?) to integrate in the render to file tool ?

[daniel]

This effect happens when you have a high density and you display it on a small resolution screen: several points are displayed at the same pixel position on the screen. In this case you'll only see the closest point (relatively to your point of view). If the local variability is huge, you might miss some important information indeed.

Subsampling is an option but the selection of points /values is not controlled.

One way to cope with this is to 'render' the screen (Display > Render to file) with a zoom (X2, X3, X4, etc.) and tell CC not to scale features (points, etc.) so that the points are still displayed with 1 pixel only but with a higher resolution.

Another option if your cloud is relatively flat is to use the 'Rasterize' tool. At least you'll be able to choose how to mix the points and scalar field values in each 'pixel' (raster cell): min, average or max value.

A last option would be to locally 'smooth' the scalar field values (if it makes sense) with 'Edit > Scalar fields > Gaussian filter'.

[OMG20Centz]

I think the Gaussian filter on the scalar field provides the best results (with a sigma value at 3 times the initial sigma).
The rendering option seems to create some artefacts (kind of moiré, that remains even when zomming in), while the rasterize option (with the average option) does not seem to actually average the scalar fields as the results are very similar from the front view (i might have done a mistake in my experiments though, or it's been corrected in the following versions as i was working with 2.9 alpha).



Thank you Daniel, you've been of great help as always.