Open felixhorger opened 2 years ago
I agree this would improve some things, but I also worry about the case where the first volume might have ~1000 slices:
julia> using ImageView, BenchmarkTools
julia> a = rand(2048, 2048);
julia> @btime ImageView.default_clim($a)
18.510 ms (2 allocations: 128 bytes)
Observable{CLim{Float64}} with 0 listeners. Value:
CLim{Float64}(3.144755991169035e-7, 0.9999997443110233)
Now multiply that by 1000 and you have 18s. Is it really worth that amount of time? Is there something else we could do instead?
Hi Tim, sorry for taking so long to get back to you! I wrote a function which is faster than minimum_finite/maximum_finite from Images.jl by a factor of nearly 5. On my machine the above test then takes only 4ms which would result in four seconds on the huge volume, that's alright I'd say. Another option would be to recompute the colour range for every slice upon selection, however in my experience that confuses the user (some MRI scanners do that -> flickering when scrolling). Let me know what you think. Cheers, Felix
I think we still might want to limit the number of points considered, e.g., if we've checked more than 10^8 points then I think we'd be safe. But otherwise I like this solution.
Done! :) What do you think about using threads to speed up the value span determination?
Hi,
this is a useful tool, thanks for developing it! One minor improvement would be that the value range of grey scale volumes is computed over the whole volume instead of over the first slice.
Cheers, Felix