Array casting from channelview question

Hello,

I am not sure about how to get an Array containing the values resulting from a convolution to an image.

This is what I tried:

using Images
using TestImages
img = testimage("lena");
K =  [[1,1,1]  [0,0,0]  [-1,-1,-1]];
kernel = centered(K)
border_decision = Inner()
result_conv     = imfilter(img, kernel, border_decision)
result_conv_arr = size(Array(channelview(result_conv)))

prints

BoundsError: attempt to access 3×254×254 Array{Float32,3} at index [Base.IdentityUnitRange(1:3), Base.IdentityUnitRange(2:255), Base.IdentityUnitRange(2:255)]

Which I don't quite understand.

Nevertheless, when working with Pad() (or nothing which defaults to Pad) then the API behaves as I would expect:

border_decision = Pad()
result_conv     = imfilter(img, kernel, border_decision)
result_conv_arr = size(Array(channelview(result_conv)))

Prints

(3, 256, 256)

• How can I get an Array in the first case?

• Is there a method to retrieve the Array containing the data that works no matter how the offsetarray is?

Start with the imfilter, channelview, Inner(), and Pad() examples in the issue, and reproduce the BoundsError using the lena test image. Document how the result differs for offset arrays and what users should do to obtain an Array regardless of its offsets; done means both questions have clear guidance.