Right, we need to back up a little. Generally the ortho from drone data is huge, and high res.
Its too heavy (too many pixels) for most uses, and not necessarily so.
First, I normally do not have to deal with reprojection situations. My CRS in my ortho creation software (Agisoft or DD, Pix4d…whatever) is set to state plane for my area. If I did need to reproject, I do that first in QGIS.
Second issue is resampling the image to say, 1/4 its size which means 50% reduction in pixels on each side. I use the free GIMP program to do that. I save as jpg.
Now the issue of the “world file” comes in, as that is how everyone that does not use geotiff tells progs where images go. I guess GE users use kml, one more format.
What I do is pull in the full image to autocad, using the world file agisoft spit out. Then I pull in my downsampled images and line up with original, then run a too I wrote l to make the world file.
You will not have that so the equivalent is doing that in google earth, making the kmz instead of a world file.
The point is to make smaller images. Pyramiding might work in GE, never tried it. Its a GIS thing and have used it for that. You can skip that though and just go with a smaller image.
That should speed things up a ton.