Jump to content

  • Log In with Google      Sign In   
  • Create Account

mahendragr

Member Since 06 Jan 2010
Offline Last Active Sep 08 2011 02:27 AM

Posts I've Made

In Topic: Scale values from one range to another, but seems tricky

08 September 2011 - 12:43 AM

Awesome, thanks for the reply, i solved it!

You can see that 256*256 = 65536. So to scale your values, first multiply the index by 256, to get the new locations for the actual values you are certain about.

So your value at position 3 in the first array ends at position 768 in the second array. This leaves rather large gaps between the values, however. How to fill them depends on how much computational power you want to spend, and what appearance you want the output to have.

I'll give an example with smaller sets below:

first array (size 4) -> resulting array (size 8)
We multiply the indices by 2

array[0]= 0.5 -> 0*2 -> newarray[0]= 0.5

array[1]= 0.3 -> 1*2 -> newarray[2]= 0.3

array[2]= 0.7 -> 2*2 -> newarray[4]= 0.7

array[3]= 0.1 -> 3*2 -> newarray[6]= 0.1

We know the positions of the original values, but we have gaps for the values newarray[1,3,5,7]. The quickest method would be to use the value of the previous element in the array. So newarray[1]=0.5, newarray[3]=0.3 and in general newarray[x]=newarray[x-1], when x is odd.

You can also do linear interpolation, where the value at each point is the midpoint of the values at each side. So for x=odd, newarray[x]=(newarray[x-1]+newarray[x+1])/2.

Depending on how smooth you want the output to look, you can go with quadratic or higher order interpolations. You should look up interpolation of you're not sure how to do it.


In Topic: Regarding my thesis

13 February 2011 - 09:47 AM

Hey, thanks for the link tachikoma, looks interesting.

In Topic: Regarding my thesis

12 February 2011 - 02:53 AM

Hello guys,

Thanks for your suggestions, i'm actually reading some papers and found out global illumination in the voulmetric lighting to be interesting. Is there any work done on out of core rendering of large medical data sets, say like 8gb or more size??, and is there any research on compression/de compression algorithms for these huge data sets??, it would be really nice if someone with volume rendering experience replies.

In Topic: BRDF implementation problem

16 May 2010 - 07:32 AM

@cignox1

well actually i want to have a distribution (whose variance is known) around the view direction, this is what i wanted to know, how to center this distribution around my view direction

In Topic: How to generate rays in a simple ray tracer/pin hole camera

05 May 2010 - 06:05 AM

@superVGA

Yes i got you. The reason why i dont want to generate all the rays at once is that i'm trying to write a function which is then called inside a loop in some other function.

PARTNERS