I have a vector of matrices:

```
julia> n = 4
4
julia> sz = (5,6)
(5, 6)
julia> imgs = [i*ones(Int, sz...) for i in 1:n]
4-element Array{Array{Int64,2},1}:
[1 1 … 1 1; 1 1 … 1 1; … ; 1 1 … 1 1; 1 1 … 1 1]
[2 2 … 2 2; 2 2 … 2 2; … ; 2 2 … 2 2; 2 2 … 2 2]
[3 3 … 3 3; 3 3 … 3 3; … ; 3 3 … 3 3; 3 3 … 3 3]
[4 4 … 4 4; 4 4 … 4 4; … ; 4 4 … 4 4; 4 4 … 4 4]
```

I want to keep this form, but I also need another form: a matrix where each cell is a vector containing all the equivalent elements of the matrices from before:

```
julia> noise = [[imgs[j][i] for j in 1:n] for i in CartesianIndices(sz)]
5×6 Array{Array{Int64,1},2}:
[1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4]
[1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4]
[1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4]
[1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4]
[1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4] [1, 2, 3, 4]
julia> noise[2,3] # for example
4-element Array{Int64,1}:
1
2
3
4
```

Because I won’t need every single one of those vector-of-elements I’m using MappedArrays on the `noise`

matrix to save time. I’m looking however into using some `@view`

wizardry or some such for the reshaping step (turning the vector of matrices into a matrix of vectors)… Anyone knows a cool trick I can use?

For context: this is a vector of images (1440x1080 pixels) and can be 5000 images long. I’m accessing ~5000 pixels in this 10^10 soup.