rgoer
Posts: 46 |
Hi FF peoples,
Let's say you have a profile gradient such as this: ![]() And let's say you want to quantize that gradient into 8 by 8 blocks like this, with these exact pixel values in the quantized result: ![]() The catch is that you want to quantize it using only an offset node. What would your H and V offset inputs look like? (The reason I'm looking for a way to get this exact result by quantizing an input gradient rather than just creating the quantized result from the start via a step curve is: I want to eventually apply this kind of quantization to non-linear gradients) edit: the exact pixel values in that result are: [(1/64),(2/64),(3/64)..(8/64)] [(9/64)..(16/64)] .. [(57/64)..(64/64)] |
|||
Posted: October 3, 2008 1:10 pm | ||||
ThreeDee
![]() |
I ain't gonna try and do it with one offset, but here it is with four offset steps.
Quantize 8x8.ffxml |
|||
Posted: October 4, 2008 4:37 am | ||||
rgoer
Posts: 46 |
Wow, you rule. On a guess I changed the one percentage from -6.13 to -6.25 (half of one eigth) and the other from -43.88 to -43.75 (the remainder out of 50), and the results still seemed to work, so very cool.
One interesting bug has popped up out of this, though--working with large image sizes (2k*2k or larger) you will see weird border artifacts show up between quantized "cells" in the render preview unless you have the "view actual size preview" option enabled. |
|||
Posted: October 4, 2008 1:45 pm | ||||
ThreeDee
![]() |
Glad you saw that. Math before morning coffee always carries a certain liability... |
|||
Posted: October 4, 2008 2:38 pm |
Filter Forge has a thriving, vibrant, knowledgeable user community. Feel free to join us and have fun!
33,712 Registered Users
+19 new in 30 days!
153,534 Posts
+31 new in 30 days!
15,348 Topics
+72 new in year!
9 unregistered users.