What is the principle of generating MipMap using non-power-of-two Texture2D supported by UE5.1? The mentioned ‘however depending on the platform it might not always save memory relative to the next power-of-two size up’ What are the specific platforms and what is the principle of ‘not always save memory’?
For UI case yes it recommended not to use MipMap, since you want a clear image for your UI. When using MipMap it add a +30% overhead if am not wrong, so for a UI you will actually save memory. (there some special case for UI mipmap when using scale ui)
You can test that in unreal, by going through your texture and set the texture group to UI Group.
yes, I know that, but my question is not asking that.
I found it: Using non-power-of-two will not allow streaming. BCN formats need block-of-4 alignment, so most sizes are not going to use BCN format, and also, no streaming will cost more memory.
1 Like
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.