3d scan of imperfect globe to a 2d Equirectangular Projection

[Vigilus]

@ Selden

What types of images did you import to Blender?

I imported the .obj file and the UV map as a .jpg file (I didn't know about compression then, wish it was .tiff!)

Or do you intend to produce globes with bumpy, mountainous surfaces?

I'm going for the shading and lighting of the mountains on a physically smooth surface. Cheaper and still pretty to look at, avoiding the annoyance of magnets on bumpy surfaces as well. This is where I think baking would be a good solution.

What do you consider to be "high resolution?"

8192x8192 is what the industry standard is. Although there is a lot of debate, the quality gained for the time to make and find workarounds for programs that don't support high resolutions starts to taper off at that number.

FWIW, the command line program that I use (MMPS project) was able to convert an 8Kx8K orthographic projection to a 16Kx8K equirectangular projection.

I'll have to check that out, thanks for telling me about it.

How are you translating the globe that you created into surface texture images?

The camera at the object's origin, in orthographic view, 6 Renders in Blender along the preset views of: left, right, front, back, top, bottom. 8192x8192, outputting as a .tiff file. I don't know if this is going to work or if these are the correct settings.

Are you using a camera to take pictures of it? If so, what model of camera is it and how many megapixels are involved (since that's how cameras often are rated)?

I originally used an Artec Space Spider to 3d scan the physical globe into the digital medium.

[selden]

I'll have to check that out, thanks for telling me about it.

For more information, see http://www.users.globalnet.co.uk/~arcus/mmps/

A limitation of MMPS is that it was written for Linux. While I found it relatively easy to recompile it for use with Cygwin under Windows 7 (which is what I use), others might be less comfortable doing that. (In principle, at least, I could provide the Windows binaries.) An alternative, if you aren't already using Linux, might be to use one of the Linux environments available for Windows 10.

The camera at the object's origin, in orthographic view, 6 Renders in Blender along the preset views of: left, right, front, back, top, bottom. 8192x8192, outputting as a .tiff file. I don't know if this is going to work or if these are the correct settings.

That procedure seems reasonable to me. Combining the resulting images so their borders aren't noticeable might be challenging.

I originally used an Artec Space Spider to 3d scan the physical globe into the digital medium.

Aha! Not as simplistic as I was thinking.