Filter Forge - Minimum/Median/Maximum

Active Topics Search

Carl

c r v a

Posts: 7289
Filters: 82

Great new components, opens up excellent new art filters possibilities smile:)

If they had green inputs they would be even better, percentage, roughness, contrast.

A new edge detection ways

visit http://www.carlrolfeart.com/

Posted: October 22, 2009 2:19 am

Details E-Mail

Carl

c r v a

Posts: 7289
Filters: 82

new metals

visit http://www.carlrolfeart.com/

Posted: October 22, 2009 2:20 am

Details E-Mail

Carl

c r v a

Posts: 7289
Filters: 82

Art filter foundations

visit http://www.carlrolfeart.com/

Posted: October 22, 2009 2:21 am

Details E-Mail

Vladimir Golovin
Administrator

Posts: 3446
Filters: 55

A couple of suggestions:

1. Turn High Precision on for metals (and generally for anything where you're using Median or Percentile as a bump/height map). This will reduce quantization artifacts -- when High Precision is on, 65536 histogram levels are used instead of 256.

2. For art filters, you might want to turn Anti-Aliasing on. It will be slower, but the edges will be much less jaggy. Also, it looks like your example uses Maximum. I'd advise to replace it with Percentile set to 95% -- this will produce a similar look but without the ugly squares.

Posted: October 22, 2009 5:41 am

Details E-Mail

Kraellin

Kraellin

Posts: 12749
Filters: 99

good stuff, carl smile:)

If wishes were horses... there'd be a whole lot of horse crap to clean up!

Craig

Posted: October 22, 2009 1:33 pm

Details E-Mail

uberzev

not lyftzev

Posts: 1890
Filters: 36

I like it. But I wish it didn't have a "square" bias.

Posted: October 22, 2009 5:16 pm

Details E-Mail

Carl

c r v a

Posts: 7289
Filters: 82

here are the same examples with the High precision / Anti -Alaising .......

visit http://www.carlrolfeart.com/

Posted: October 23, 2009 4:33 am

Details E-Mail

Carl

c r v a

Posts: 7289
Filters: 82

visit http://www.carlrolfeart.com/

Posted: October 23, 2009 4:34 am

Details E-Mail

Carl

c r v a

Posts: 7289
Filters: 82

visit http://www.carlrolfeart.com/

Posted: October 23, 2009 4:34 am

Details E-Mail

Carl

c r v a

Posts: 7289
Filters: 82

this is a pearl effect smile:)

again a simple construction

visit http://www.carlrolfeart.com/

Posted: October 23, 2009 4:40 am

Details E-Mail

Carl

c r v a

Posts: 7289
Filters: 82

Quote
uberzev wrote: "square" bias

I originally thought these components could possibly be manipulated to substitute blurs, in certain situations to speed them up.
Thanks Craig but they are really just test snippets smile:blush:

visit http://www.carlrolfeart.com/

Posted: October 23, 2009 4:49 am

Details E-Mail

Vladimir Golovin
Administrator

Posts: 3446
Filters: 55

Quote
uberzev wrote: But I wish it didn't have a "square" bias.

Yep, that's a very desirable feature, and algorithmically it's not that complex. However, the current square-kernel version lends itself to optimizations much better, while the circular-kernel version is limited to a non-optimized brute-force approach.

Posted: October 23, 2009 6:17 am

Details E-Mail

Sphinx.

Filter Optimizer

Posts: 1750
Filters: 39

Those new min/median/max components are quite nice. Here's a simple tracing filter that, besides tracing, shows how to get both min and max filtering from one min or max component only (this is possible to do with 2 pairs of grayscale sources).

Minmax Trace.ffxml

Njyldgarkn init/prepare process x no. of render blocks x no. of passes

Posted: October 23, 2009 6:24 am

Details E-Mail

Sphinx.

Filter Optimizer

Posts: 1750
Filters: 39

Here is an example