Path: utzoo!attcan!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!nthropy!andy
From: andy@nthropy.UUCP (Andy Lowe)
Newsgroups: comp.graphics
Subject: bug in hls_to_rgb from Foley, van Dam, Feiner, Hughes
Message-ID: <9009301931.AA14664@nth.com>
Date: 30 Sep 90 19:31:24 GMT
Lines: 59
Posted-Date: Sun, 30 Sep 90 14:31:24 CDT
To: comp-graphics@cs.utexas.edu


Greetings. 

I had occasion to implement an hls to rgb converter for a research 
project, and turned to the new Foley, vanDam, Feiner, Hughes book for
the algorithm.

On p. 596 is a procedure that purports to do the conversion.  While
studying it, I couldn't see how it worked.  In fact, it violated the
stated bounds on r, g and b!  (try it with h = 0.)  Following the
references, they cited Metrick in the Status Report of the GSPC from
1979.  Having once implemented a subset of the Core proposed
standard, I just happened to have this volume handy.  Aside from
differences in convention (the GSPC put blue at 0 degrees, while
Foley et al put red at 0 degrees), I think I found three bugs in this
procedure!

Can someone tell me I'm wrong? (please.)

-------------------- begin quote --------------------

procedure HLS_To_RGB (var r, g, b: real; h, l, s: real);
...
function Value (n1, n2, hue)
begin
if hue > 60 then	{ should be "if hue < 60 then" !? }
...
else if hue < 240 then
	Value := n1 + (n2 - 1) * (240 - hue) / 60

	{ should be "Value := n1 + (n2 - n1) * (240 - hue) / 60" }
...
end {Value}

begin { HLS_To_RGB }
if l <= 0.5 then
...
else
	m2 := l + s - l * s;	{ should be "m2 := l + s + l * s" ?! }
	m1 := 2 * l - m2	{ style nit.  indentation indicates
				  this is part of the else.
				  syntactically ok, though. }   
...
end { HLS_To_RGB }

-------------------- end quote --------------------

thanks, and regards.

Andy Lowe
andy@nth.com

p.s. -- please Refer to 

Computer Graphics -- Principles and Practice
Foley, van Dam, Feiner, Hughes
Addison-Wesley, 1990

for the complete procedure.