4.3 Conversion and Construction

The return type is imprecise because Typed Racket does not support the object system well yet. As a typed function, this is most useful in DrRacket’s REPL to visualize flomaps; any other typed use is difficult.

A zero-size fm is padded by one point in any zero direction before conversion. For example, if fm is size 0×1, the result of (flomap->bitmap fm) is size 1×1.

Values are clamped to between 0.0 and 1.0 before conversion.

If unscaled? is true, the flomap is converted from the actual bitmap backing bm rather than a scaled version. See the #:unscaled? keyword parameter of get-argb-pixels for more information.

The argument type is imprecise because Typed Racket does not support the object system well yet.

To create flomaps filled with a solid color, use make-flomap*.

The function f receives three arguments k x y: the color component and two positional coordinates.

To build a flomap using a function that returns vectors, see build-flomap*.

Analogous to build-vector.

The draw function should accept a dc<%> instance and use its drawing methods to draw on an underlying bitmap. The bitmap is converted to a flomap using bitmap->flomap and returned. See Floating-Point Bitmaps for an example.

This function is very difficult to use in Typed Racket, requiring occurrence checks for, and use of, experimental types. However, as Typed Racket grows to handle Racket’s object system, the types will be made more precise.

In other words, flomap-multiply-alpha converts non-alpha-multiplied flomaps into alpha-multiplied flomaps, and flomap-divide-alpha converts them back.

You should not generally have to use these functions, because bitmap->flomap returns an alpha-multiplied flomap and every alpha-aware flomap function assumes its arguments are alpha-multiplied and produces alpha-multiplied flomaps.

See Opacity (Alpha Components) for a discussion of opacity (alpha) representation.

There are three differences between the function f passed to build-flomap and the f passed to inline-build-flomap. First, the f passed to inline-build-flomap can be a macro. Second, it receives arguments k x y i, where i is a precalculated index into the result’s values. Third, it must return a Float.

Using inline-build-flomap instead of build-flomap may ensure that f is inlined, and therefore floats remain unboxed. Many library functions use inline-build-flomap internally for speed, notably fm+ and the other pointwise arithmetic operators.

This is not available in untyped Racket.

There are three differences between the function f passed to build-flomap* and the f passed to inline-build-flomap*. First, the f passed to inline-build-flomap* can be a macro. Second, it receives arguments x y i, where i is a precalculated index into the result’s values. Third, it must return a FlVector.

This is not available in untyped Racket.