Skewed heatmap / image / isosurface...

Hi!

I have scientific data represented in 3D grids. Heatmaps and isosurfaces are great for displaying them. However, in some cases the axes of the grid are not x, y and z, but some other direction in space. Then, I would like to somehow see the “skewed” images. I believe it is not possible in plotly to specify the “direction” of the axes in these kind of traces, is it? I don’t know how hard it would be to implement, but it would be extremely useful.

In case this is not feasible or you are not interested in implementing it, could you suggest a workaround? I’m quite lost on this 😃

PS: Sorry if this belongs to plotlyjs github page, I’m not sure.

[empet]

@pfebrer96 An instance of the class go.Isosurface is defined by plotly.js as a Mesh3d trace, with vertices and faces (triangles) returned by the Marching Cube algorithm. This algorithm acts on 3D grids associated to a parallelepiped with sides parallel to the standard axes, i.e. axes of directions e1=(1, 0, 0), e2=(0,1,0), e3=(0,0,1). @archmoj can give more details on how this algorithm works for go.Isosurface

If go.Isosurface had a transformation attribute, i.e. an array of shape (3,3), and positive determinant, T=np.array([u1, u2, u3]).transpose(), to map vertex coordinates resulted from marching cube triangulation, then we could simply define an isosurface carved within a volume with sides of directions u1, u2, u3. Namely it could be defined as a Mesh3D having as vertices, the T-transformed initial vertices, and the same faces. See a workaround for this procedure in this notebook https://chart-studio.plotly.com/~empet/15738.

Notice that unlike the isosurfaces defined as go.Isosurface(s), the isosurfaces constructed by the method presented in the above Jupyter Notebook can be colored according to vertex intensities (defined by a scalar field on the isosurface).

If you intend to propose a PR, for such isosurfaces, then it must be posted on plotly.js, not here.

[pfebrer]

This is exactly what I needed! Thank you very much 😃

With that, I have all the workarounds that I needed, so you can close this issue if you want.

Cheers and I really appreciate the help!