Recent

Author Topic: How to Optimeze use of BGRAColorQuantizer??  (Read 15296 times)

circular

  • Hero Member
  • *****
  • Posts: 4224
    • Personal webpage
Re: How to Optimeze use of BGRAColorQuantizer??
« Reply #30 on: September 04, 2016, 05:12:52 pm »
For one leaf, the largest dimension is determined, and then pixels are sorted according to this dimension and half of it goes into one branch and the rest into the other.

GetDimensionValue returns a value along an axis, that can be used to sort values / determine min-max.

There is indeed a sort of averaging for the leaf.
Conscience is the debugger of the mind

Sanem

  • Full Member
  • ***
  • Posts: 173
Re: How to Optimeze use of BGRAColorQuantizer??
« Reply #31 on: September 05, 2016, 01:37:36 pm »
For one leaf, the largest dimension is determined, and then pixels are sorted according to this dimension and half of it goes into one branch and the rest into the other.
There is indeed a sort of averaging for the leaf.
It is exactly what median cut means, but I want to know a bit more details where does this part of comparing and branching the colors executes??

GetDimensionValue returns a value along an axis, that can be used to sort values / determine min-max.
2- what sort of values?? what is the purpose of shl and shr values in its code??

Thanks
« Last Edit: September 06, 2016, 08:39:27 am by simin_sh »

Sanem

  • Full Member
  • ***
  • Posts: 173
Re: How to Optimeze use of BGRAColorQuantizer??
« Reply #32 on: September 06, 2016, 09:27:43 am »
Is it the part where decides to cluster colors in which dimension??
« Last Edit: September 06, 2016, 09:31:54 am by simin_sh »

Sanem

  • Full Member
  • ***
  • Posts: 173
Re: How to Optimeze use of BGRAColorQuantizer??
« Reply #33 on: September 06, 2016, 09:43:17 am »
In gaining the colors of ColorBox class, does this part means that it separates colors that are transparent?? And why?? Does it means that they wont be considered in calculating the pallet colors and clusters that does it purpose??

 

TinyPortal © 2005-2018