360 Spherical vs Cubic vs optimisations

Started by paulcurtis, December 03, 2021, 07:09:45 AM

Previous topic - Next topic


Hi All,

Again thanks in advance for any pointers. One day when i know enough i hope to be able to contribute! 

I've searched the forums and found a reasonable amount of info but not quite what i am looking for

Ultimately i am looking for a 360 degree spherical in very high resolution - think 16k to 32k. But actually i only need 270 degrees of it and no floor.

If i render a 360 spherical i assume i am spending a lot of rendering time rendering the poles unless terragen optimises the renders so they are lower quality already? But i assume not.

So one question is that it doesn't appear possible to either render less than 360, or change the vertical fov? Ideally i could render a low res  full version and then a high res slice for the horizontal, where all the resolution is needed. But the UI shows no ability to change horizontal or vertical fov - so is there any way this could be done?

If no, the an alternative would be to render cubic, with a higher res on front, left and right and lower res top. This would most likely cut down render times a lot.

So could i set up cameras in terragen and animate a parent node, or would i need to construct it outside of terragen and import? Does anyone have any examples of this?

Many thanks!


Hi Paul,
as far as I understood your question, you're looking for a way to render only the part you need, so that the rendertimes are OK.
What about a crop render? Do you know, how to do that?


Hi Hannes,

I did not know about the crop, that could be the way - i'm going to have a look into that.

So yes, i am really looking for how to cut down render times by either not rendering parts that would not be seen and/or rendering parts of less importance in less resolution.

So crop could be the way, thank you. I will try and see if it works for spherical



So just for completeness in case anyone searches for this

Then yes, the crop options work wonderfully. The only thing i'd say is that they are done as a fraction of the image from 0 to 1 and therefore working out pixel values is a bit painful.

So on a spherical i can render a 4K version of the whole scene and then overlay a 16k version of the central portion and that seems to work nicely.

I just need more rendering cores! 

GPU support would be wonderful!