The display color of network weights in iTorch

[Mata Fu]

I loaded pre-trained network and trying to visualize the weights

new_dis = torch.zeros(dis:size(1),dis:size(2),zoom,zoom)
t = {}
for i = 1,dis:size(1) do
    for j = 1,dis:size(2) do
        local img = dis[i][j]:float()
        new_dis[i][j] = image.scale(img,zoom,zoom,'bicubic')
    end
    t[i] = new_dis[i]
    itorch.image(new_dis[i])
end
itorch.image(t)

Then I found the filter display one by one (itorch.image(new_dis[i])) is quite different from display together (itorch.image(t)). Here are the 112 weights from 1st layer and  the weights of last 4 filters. Could someone help me to change the together-display filter to be higher intensive contrast as showing one by one?