Small typo

[mgautam98]

Small typo in example

[johnnychen94]

Ha, actually this isn't a typo; the lower-case ones preserve the colorant types.

julia> using TestImages

julia> img = testimage("cameraman");

julia> Float64.(img) |> eltype
Float64

julia> float64.(img) |> eltype
Gray{Float64}

julia> img = testimage("lena");

julia> Float64.(img) |> eltype
ERROR: MethodError: no method matching Float64(::RGB{Normed{UInt8,8}})
Closest candidates are:
  Float64(::Int8) at float.jl:60
  Float64(::Int16) at float.jl:60
  Float64(::Int32) at float.jl:60
  ...
Stacktrace:
 [1] _broadcast_getindex_evalf at ./broadcast.jl:631 [inlined]
 [2] _broadcast_getindex at ./broadcast.jl:604 [inlined]
 [3] getindex at ./broadcast.jl:564 [inlined]
 [4] copy at ./broadcast.jl:854 [inlined]
 [5] ...
 [6] top-level scope at REPL[10]:1

julia> float64.(img) |> eltype
RGB{Float64}

There's also float, float32, n0f8 and etc defined in ImageCore.

Also, float(RGB{N0f8}) == RGB{Float32}, and floattype(RGB{N0f8}} == RGB{Float32}. The difference between floattype and float

It might be a little confusing at the beginning, here's the most frequent usage of them:

# MappedArrays.of_eltype
of_eltype(float(eltype(img)), img) # lazy type promotion

# this is more generic than `Gray{Float32}.(img)`
float.(img)

cc: @jiviteshjain as you might be unaware of these small details, either.

[mgautam98]

Oops! Thanks for detailed information.

[timholy]

Just to make sure the logic is clear, since Float64 is a type, Float64(x) is a constructor call that you should expect to return an object of type Float64. (There's nothing built-in to the language to enforce that, although a proposal to add it and other things like it is being actively discussed, see JuliaLang/julia#36463 which comes at the problem from the standpoint of improving inference but in Julia 2.0 it would also enforce consistency). In contrast, float64 "just" a function with no expected relationship to a particular type, so is under no such constraints.