Blending Lightmaps on GPU?

So,

I’m trying to blend two lightmaps into a third one, which is active in lightmapsetrings.

(I can be wrong)
My current understanding is that active lightmaps are loaded in gpu
That’s why we need to Apply() after we’ve modified the texture data / pixels

Currently, I’m using IParallelJobFor to blend chunks of pixels per frame and then set the output into target

And hit Apply()
That’s where the problem arises, that specific frame drops the rate to about 60% of usual performance.

Is there any way to avoid this path and modify the texture data without needing to hit Apply()

Compute Shader or something?

Keeping in mind that Target is used in lightmapsetrings