- #1
spx2
- 2
- 0
Hi,
I'm trying to write a score function.
The score function is applied to each cell of a grid
and because the grid has many cells ( 400x600 or 800x600 )
If I want to experiment with different score functions to
see which one is best I'd have to see some kind of visual results.
I thought of plotting that score function to the screen
(because viewing just numbers isn't enough any more)
and all it's values will be colors of pixels on a grid.
The problem is choosing the colors and the transition between low-values to
big values of the score function so that they show up as beeing relevant
and smooth to the eye.
Let's take euclidean distance as the score function for example.
If we have 2 objects,one fixed at the bottom right of the screen and one
starting in the upper left and moving towards upper right,then a transition of colors should
be made for the area where the fixed object is , and the signification of this will be
that the distance between these objects is getting smaller and smaller.
I have tried getting the euclidean distance normalized to fit in the color spectrum and then
plotting but that turns out to not be so good either.
I'm having difficulties finding the best mapping between colors->distance.
Has anyone tried this ?
Thank you
I'm trying to write a score function.
The score function is applied to each cell of a grid
and because the grid has many cells ( 400x600 or 800x600 )
If I want to experiment with different score functions to
see which one is best I'd have to see some kind of visual results.
I thought of plotting that score function to the screen
(because viewing just numbers isn't enough any more)
and all it's values will be colors of pixels on a grid.
The problem is choosing the colors and the transition between low-values to
big values of the score function so that they show up as beeing relevant
and smooth to the eye.
Let's take euclidean distance as the score function for example.
If we have 2 objects,one fixed at the bottom right of the screen and one
starting in the upper left and moving towards upper right,then a transition of colors should
be made for the area where the fixed object is , and the signification of this will be
that the distance between these objects is getting smaller and smaller.
I have tried getting the euclidean distance normalized to fit in the color spectrum and then
plotting but that turns out to not be so good either.
I'm having difficulties finding the best mapping between colors->distance.
Has anyone tried this ?
Thank you