Logo by Pauldelbrot - Contribute your own Logo!

END OF AN ERA, FRACTALFORUMS.COM IS CONTINUED ON FRACTALFORUMS.ORG

it was a great time but no longer maintainable by c.Kleinhuis contact him for any data retrieval,
thanks and see you perhaps in 10 years again

this forum will stay online for reference
News: Visit the official fractalforums.com Youtube Channel
 
*
Welcome, Guest. Please login or register. April 26, 2024, 06:36:04 AM


Login with username, password and session length


The All New FractalForums is now in Public Beta Testing! Visit FractalForums.org and check it out!


Pages: [1]   Go Down
  Print  
Share this topic on DiggShare this topic on FacebookShare this topic on GoogleShare this topic on RedditShare this topic on StumbleUponShare this topic on Twitter
Author Topic: Supersampling: RGSS, flipquad, ??  (Read 3265 times)
0 Members and 1 Guest are viewing this topic.
murkey
Forums Newbie
*
Posts: 5


« on: January 23, 2015, 10:44:31 PM »

I multithreaded my libgmp-based Mandelbrot renderer and got render times down to a couple minutes - not great, but good enough to move on to some smoothing. I'm thinking about using the RGSS or Flipquad supersampling pattern (or maybe quincunx), and was wondering if anyone had experience with these or thoughts as to which would yield the highest quality results.

Discussion of various techniques in these papers:
https://mediatech.aalto.fi/~samuli/publications/hasselgren2005cgf_paper.pdf
http://fileadmin.cs.lth.se/cs/Education/EDA075/lectures/L11a-AA.pdf

I'm also interested in filter-based methods for anti-aliasing, but it seemed like supersampling would yield sharper, more accurate results - maybe?

Thanks!
Logged
hobold
Fractal Bachius
*
Posts: 573


« Reply #1 on: January 24, 2015, 02:47:19 AM »

Supersampling is a general high quality solution. Flipquad and rotated grid, however, are tuned for speed over quality. They were both designed with realtime graphics in mind. Both target the most objectionable aliasing in that context: straight edges that are nearly horizontal or vertical (rotated grid also helps for near diagonal edges).

The advantage of the above sampling patterns is that they require relatively few extra samples per pixel. They won't help smoothing pixel noise in highly detailed regions, but they probably do a good job for the outer, smoothly curved, level set boundaries.
Logged
murkey
Forums Newbie
*
Posts: 5


« Reply #2 on: January 24, 2015, 03:09:51 PM »

That's true, good point. So do people generally just use a square grid for supersampling?
Logged
dom767
Explorer
****
Posts: 40


dom767
WWW
« Reply #3 on: February 26, 2015, 07:26:17 PM »

Dunno if you got anywhere with the supersampling? I just completed a review of a bunch of different 2d sampling algoirthms using antialisaing as the case study, my conclusion is the best sampling algorithm for 2d is a variant of poisson disc sampling I've called relaxing poisson sampling. But there are forms of grid sampling that perform well too.

http://woo4.me/wootracer/2d-samplers/

Note that there's some definite gotchas with grid based supersampling if you're doing an incremental renderer!

Dom
Logged
Pages: [1]   Go Down
  Print  
 
Jump to:  

Related Topics
Subject Started by Replies Views Last post
Simple anti aliasing / supersampling help! Programming richardrosenman 6 2303 Last post April 15, 2011, 02:41:21 AM
by Softology
Mandelbrot: Supersampling vs Multisampling? Programming « 1 2 » Kalles Fraktaler 21 9449 Last post April 05, 2014, 08:03:34 PM
by SeryZone
How do I implement supersampling? Programming « 1 2 » Gore Motel 20 11961 Last post May 26, 2014, 03:58:06 AM
by laser blaster

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Valid XHTML 1.0! Valid CSS! Dilber MC Theme by HarzeM
Page created in 0.146 seconds with 24 queries. (Pretty URLs adds 0.006s, 2q)