Image#remap produces unreasonable result about rmagick HOT 5 OPEN

Hopefully, this is caused by the other issue on which you commented (http://github.com/rmagick/rmagick/issues/issue/4) and has already been fixed in source. I'll try to reproduce it using my various versions and let you know if this is a separate bug.

from rmagick.

Unfortunately it appears to be caused by something else. Here's the output:

This is RMagick 0.0.0 ($Date: 2009/12/20 02:33:33 $) Copyright (C) 2009 by Timothy P. Hunter
Built with ImageMagick 6.6.2-3 2010-06-08 Q8 http://www.imagemagick.org
Built for ruby 1.9.1
Web page: http://rmagick.rubyforge.org
Email: [email protected]
red=254, green=254, blue=254, opacity=0
red=255, green=255, blue=255, opacity=0
red=255, green=227, blue=0, opacity=0

when running:

require 'RMagick'
puts Magick::Long_version
img = Magick::Image.read('input.png').first
puts img.pixel_color(0,0)
colors = Magick::Image.read('colors.png').first
puts colors.pixel_color(413,0)
img.remap(colors, Magick::NoDitherMethod)
puts img.pixel_color(0,0)
img.write('output.png')

from rmagick.

Hi,

What is your OS and architecture?

I can't seem to reproduce this problem. Here's my output using your images/program:

This is RMagick 0.0.0 ($Date: 2009/12/20 02:33:33 $) Copyright (C) 2009 by Timothy P. Hunter
Built with ImageMagick 6.6.2-4 2010-06-10 Q16 http://www.imagemagick.org
Built for ruby 1.9.1
Web page: http://rmagick.rubyforge.org
Email: [email protected]
red=65278, green=65278, blue=65278, opacity=0
red=65535, green=65535, blue=65535, opacity=0
red=65535, green=65535, blue=65535, opacity=0

And, the result is the following image:
http://tmp.omer.bar-or.org/bad_remap_out.png

I haven't yet looked into the difference between our two rgb values, and I've asked Benjamin to see if he can reproduce the problem.

I've tried RMagick 1.8.6, 1.8.7, and 1.9.1
I've tried ImageMagick 6.6.1-0, 6.6.2-3, and 6.6.2-4

from rmagick.

I'm running Ubuntu 10.04 LTS, "Lucid Lynx", 64 bit.

It's Q8 vs Q16. I recompiled with Q16 and get the same results as you.

Q16 however is not ideal for memory and speed reasons, so I'd much prefer being able to use Q8.

Does this imply that the bug is in ImageMagick? Sounds like it's similar to what's discussed here: http://www.imagemagick.org/discourse-server/viewtopic.php?f=3&t=13021

(you'll need to edit out the amp; from that URL as the ampersand gets sanitized)

from rmagick.

Woops. You're right. I hadn't considered Q8 vs. Q16.

Yes, your issue does sound the same as the comment to which you linked, and the fact that it depends on Q8 vs. Q16 rather than on the ImageMagick version at least suggests that it is a bug in ImageMagick rather than RMagick. I'll play around and maybe try to talk to the ImageMagick folks, but based on that link, it unfortunately sounds like a known issue without a known solution for now.

from rmagick.