3 ms·

But that screen projection will not have effect on calculations performed on the globe. Otherwise also I fail to see why this is a valid argument. The viewpoint

But that screen projection will not have effect on calculations performed on the globe. Otherwise also I fail to see why this is a valid argument. The viewpoint usually is manipulatable to view the globe from any desired point of view.

jacobolus 7y ago
What you are proposing in practice is to always use some kind of perspective or orthographic projection, but make it easy for viewers to change the projection parameters (by rotating/zoomming). But an orthographic projection is strictly inferior to other projections for essentially every purpose by every criterion except “draw a picture of what the earth might look like if viewed from space”. * * * As for "calculations": these should not be done in terms of latitude/longitude if it can be avoided. Personally I think we should switch to using stereographic projection as an internal data format. Here’s my draft writeup https://observablehq.com/d/5e75dd8e56fe255f https://observablehq.com/d/5e75dd8e56fe255f
- billfruit 7y ago
  I am not suggesting using an orthographic projection. What I was suggesting was using a 3d earth model, which will be rendered on the screen using a perspective projection. Does not the earth viewed from space indicate true shapes, distances, and areas?
  - lmz 7y ago
    > Does not the earth viewed from space indicate true shapes, distances, and areas? Surely some areas will be out of sight, or squished to the sides of the globe.
  - jacobolus 7y ago
    > Does not the earth viewed from space indicate true shapes, distances, and areas? Of course not. A perspective projection distorts "true" shapes, sizes, and distances more than any other common choice of map projection, and in addition leaves more than half of the globe out of sight.
    - billfruit 7y ago
      But in an interactive display that is not of concern. It allows us to rotate and move vantage point to see different parts of the globe, as much as a real globe would allow someone to peruse.
zokier 7y ago
I'm not a GIS expert, but I would have assumed that any calculations would be performed on the geoid model (or whatever) instead of on the projected map? Is that really not the case?

jacobolus 7y ago

What you are proposing in practice is to always use some kind of perspective or orthographic projection, but make it easy for viewers to change the projection parameters (by rotating/zoomming). But an orthographic projection is strictly inferior to other projections for essentially every purpose by every criterion except “draw a picture of what the earth might look like if viewed from space”. * * * As for "calculations": these should not be done in terms of latitude/longitude if it can be avoided. Personally I think we should switch to using stereographic projection as an internal data format. Here’s my draft writeup https://observablehq.com/d/5e75dd8e56fe255f https://observablehq.com/d/5e75dd8e56fe255f

billfruit 7y ago
I am not suggesting using an orthographic projection. What I was suggesting was using a 3d earth model, which will be rendered on the screen using a perspective projection. Does not the earth viewed from space indicate true shapes, distances, and areas?
- lmz 7y ago
  > Does not the earth viewed from space indicate true shapes, distances, and areas? Surely some areas will be out of sight, or squished to the sides of the globe.
- jacobolus 7y ago
  > Does not the earth viewed from space indicate true shapes, distances, and areas? Of course not. A perspective projection distorts "true" shapes, sizes, and distances more than any other common choice of map projection, and in addition leaves more than half of the globe out of sight.
  - billfruit 7y ago
    But in an interactive display that is not of concern. It allows us to rotate and move vantage point to see different parts of the globe, as much as a real globe would allow someone to peruse.

zokier 7y ago

I'm not a GIS expert, but I would have assumed that any calculations would be performed on the geoid model (or whatever) instead of on the projected map? Is that really not the case?