CRIMP
Timeline
Not logged in

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

14 check-ins using file demos/wavy.tcl version 78cedb7e90

2010-07-10
23:19
Added another table generator, which wraps on over- and underflow, instead of clamping check-in: a5062dc2b6 user: andreask tags: ak-experimental
23:18
Added plot widget to the main package check-in: 4e71c8d3d1 user: andreask tags: ak-experimental
23:17
Added plot widget to display the various 256x256 functions we have.
This is still a bit buggy.
When a change is made the old series is still shown in the display.


check-in: 13eb3abf7b user: andreask tags: ak-experimental
2010-07-09
21:56
Fixed bucket access in histogram operator, and extended to grey16. 64K possible values still allow allocation of full counter array check-in: 45b5a80050 user: andreask tags: ak-experimental
21:42
Added operator returning image histograms check-in: 121647be3a user: andreask tags: ak-experimental
15:34
Fixed another doh, this time in the (de)gamma tables.

I failed to scale the pixel values into the range where gamma
operates, i.e. [0..1] before doing the (de)gamma. Now that I do this,
and scale back after the application the operation behaves as expected
and proper.


check-in: 2809295fef user: andreask tags: ak-experimental
04:02
Doh. When implementing the blending for HSV to understand the weird
colors generated by the RGB I noticed the reason why. I am a person of
no brain at all. Forgot to initialize the 'ralpha' value, i.e. instead
of mixing alpha versus 1-alpha it was alpha versus some random value.

This is now fixed.

All blenders have now also been converted to the MIX macro and ralpha.
Two demos now, RGB and HSV blending. Both work and look ok. They are
different however in the path they take through the color space while
interpolating, with HSV looking a bit more unnatural.

Added a second over demo as well, reversing the roles of fore- and
background images.


check-in: 1ea8c1d2af user: andreask tags: ak-experimental
2010-07-08
21:40
New operators:
* blend (2 images per user chosen alpha value)
* over (blend two images per the alpha of the foreground)
* setalpha (replace/add alpha channel in image from 2nd image)
* blank (generate image of uniform color/grey level)

Additional demos for these ops.
Additional images for use by the demos.

The blend demo currently does a single image against uniform
black. Blending two independent natural images has psychedelic
results, doesn't look good. This in RGB space. Needs more thought.
Should maybe convert both to HSV and result back ?


check-in: 8b81d87cd6 user: andreask tags: ak-experimental
21:35
Added rgba -> rgb conversion = remove alpha channel check-in: 142d30800e user: andreask tags: ak-experimental
06:49
Added notes about the CIE XYZ/L*a*b* color spaces and transforms check-in: fcc23a1350 user: andreask tags: ak-experimental
2010-07-07
23:13
Changed the naming schemes of join, convert, write commands. The input-image type-dependent part goes last check-in: 441e2a2219 user: andreask tags: ak-experimental
20:05
Final name for the new demo framework check-in: 8528700710 user: andreask tags: ak-experimental
20:04
Dropped old demos (mine and andy's) check-in: ecaf4070d8 user: andreask tags: ak-experimental
20:03
Converted the remainder of akdemo into plugins for the new xdemo. check-in: a4d17c5fe4 user: andreask tags: ak-experimental