I have a bit of a specialized problem to solve.

I have a mesh model, created from laser scanning data. Also from that data are 10 different equirectangular images of the scene being scanned. Here's an example of one: enter image description here

I've erased some tripods and added a transparent buffer at the bottom to make the image properly equirectangular. I also already know the UV maps to align my scanner mesh with all of the different images.

The trouble is that when I try to add all of these images to the material for my mesh, Blender slows down and eventually crashes well before I add all 10. I think it has to do with the fact that each one is a very large image texture (8168x4085 pixels). Also, probably due to the many different UV maps, the render of the mesh looks pretty awful. Here's an example using 4 image textures: enter image description here

Clearly, there's something wrong, but I haven't any idea what it would be.

I think a good solution to these problems may be to bake a new texture that contains all the colour information of my ten textures, but I don't know how I would do that without loading all 10 textures at once. I'd need to do it sequentially, one after the other. I'd also like to consider the alpha values in each texture, to accommodate for the fact that some portions of the textures are transparent.

Any advice on this would be helpful, I'm really stuck here.

  • $\begingroup$ Map the textures, then bake. $\endgroup$ May 30, 2016 at 21:58
  • $\begingroup$ I don't know if there is a method in Blender that let you achieve all at once, but it should be possible to bake one texture at time and then mix them in an external editor. They'll be surely more easy to handle. $\endgroup$
    – Carlo
    May 30, 2016 at 23:47
  • $\begingroup$ If blender is crashing these textures may be too big for it to handle or beyond the memory available on your machine. I have very little experience in baking, but have you considered maybe using a lower resolution version of the images instead? Make a copy, scale them down in some Image editing software and then try re-rendering. You probably wont notice the quality loss unless you are doing some serious closeups, and the good news is scaling them down should reduce memory required exponentially, so you may not need to reduce them all that much. Powers-of-two are usually memory friendly $\endgroup$ May 30, 2016 at 23:47

1 Answer 1


I would create low res versions of the image textures, then bring them all into Blender and apply them to the object material with the various UV maps. (If you bake out the combined low res image you'll have a starting point for after the next step when you have all your high res texture bakes.)

Then bring in the high res image textures (one at a time) and bake out the high res UV texture as applied to the image.

Once you've baked out all the applied high res textures, you should be able to combine them all in an external image editor (alphas included if baked out properly). You can use the previous combined low res bake as a reference.

Then you'll have a single high res image texture. If Blender still cannot handle working with this single high res texture image, you'll have to lower the resolution until Blender can work with it. It should be fine on shots that are not too close to the object in your scene.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .