YOUR ACCOUNT

Login or Register to post new topics or replies
Ramlyn
Ramlyn

Posts: 2930
Filters: 691
Derivative Diffusion by Rick Duim
http://www.filterforge.com/filters/12305.html

  Details E-Mail
Ramlyn
Ramlyn

Posts: 2930
Filters: 691
Very nice filter with a lot of presets!! smile:D
  Details E-Mail
EAdams

Posts: 447
Beautiful filter!
  Details E-Mail
ddaydreams
Frank Hawkins
Posts: 412
Filters: 1
I really like this filter, it's a lot faster than the old one.
I like the line size control which at lower settings makes an outline which seems to ad some punch.
I ran across something odd, it looks like barcode artifacts in the background.
I would like to keep the nice outline but loose the barcode.
Any thoughts?

  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
Thanks for noticing this. I had seen the "artifacts" before but only realized how to fix it today.

There is digital clipping in to the last high pass component in the diffusion section, it is set to 100% contrast. I fixed it by changing the contrast to 99.9%. Did a little testing, not quite ready to post the update, but I'll attach the filter with the usual warnings, this is an beta version of the filter, blah, blah, blah. I should have the update out this week.

Derivative Diffusion clipping fix.ffxml
Math meets art meets psychedelia.
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
Odd thing, though, it appears that an image created on my system does not match yours using the same parameters as listed. Here is one generated on my system (with the clipping fix).

Math meets art meets psychedelia.
  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
Very sweet!

Bet if you iterate that entire filter it'll go mad crazy smile:)
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
All I can say is smile:D . It occurred to me that since the first couple of high pass-blurs were really just edge detection... derivatives came to mind. But since you only get one direction with one derivative... use two. It works! Different blending modes with the derivatives adds spice. Not a popular filter, though smile:|
Math meets art meets psychedelia.
  Details E-Mail
ddaydreams
Frank Hawkins
Posts: 412
Filters: 1
Thanks Rick That fixed it for me.
As far what you said
"Odd thing, though, it appears that an image created on my system does not match yours using the same parameters as listed"

Are you sure your variation set 14790 ?
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12298
Filters: 35
Wow! This looks very good and very well, regetably I do not have time to try it but surely will do whenever I can.

Well done and thanks very much for making a filter with this very cool and interesting topic
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
Thanks all. Frank: images are the same, missed the Line Angle Adjust, my bad smile;) I'll probably get the bug fix out later today. Special thanks to SpaceRay and Skybase for the various code snippets & examples that helped make this happen. As for you, Skybase:

Quote
So in our current discussion, I just think the filter needs to be pushed to achieve a more distinct appearance. It just needs more awesome!


Thanks for the critique. It helped push this filter along. Sometimes I need an editor smile8)
Math meets art meets psychedelia.
  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
Could go like at least 20 iterations before showing signs of something more life-like really.

Uberzev's more responsible for starting me up on the whole thing. haha. Would have not bothered to deal with it without seeing some example pictures first.
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
I tried a version on my fast machine with more passes. The original is 4.5 iterative passes. I tried one with 12.5 passes, and one with 16.5. The 4.5 & 12.5 version are posted here. 12.5 is a lot better than the original and took about 3 minutes on the lifesaver on my fast PC. The 16.5 version was not noticeably different in look from the 12.5 version and took 15 minutes (probably take over a day to generate a large file).

4.5 passes:


12.5 passes:
Math meets art meets psychedelia.
  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
Wait how do you have "0.5" of a pass?

The set of operations is 1 iteration, so when you break out it's not an iteration anymore, it's a separate process. So technically you can't have 0.5 or something.
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
Ah, the last pass is only a high pass with high contrast, no blur. I call that half a pass... the DD filter is a hybrid, a couple of the iterative passes are altered by replacing the high pass with derivatives and replacing the blur with noise on different passes. So strictly speaking, I guess we won't call the public version iterative. You can subtract 2.5 from my numbers for the iterative count since 3 passes are altered.
Math meets art meets psychedelia.
  Details E-Mail
Skybase
2D/3D Generalist

Posts: 4025
Filters: 76
lol I'm extremely picky by the way. You'll notice me drill at some of the most trivial things from time to time.
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
You picky, ha smile;) I kinda picked up on this earlier... You gotta admit I had an (overblown) answer. But, I am experimenting with other inputs to the high pass-threshold-blur engine, who knows, I might find something else.
Math meets art meets psychedelia.
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12298
Filters: 35
i still love much this filter and gives a very good result with this reaction difussion patterns based on the source image, well done
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
Funny how damn slow this filter is... anytime I do reaction diffusion i use AppleScript to do it in Photoshop. It was an interesting experiment.
Math meets art meets psychedelia.
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12298
Filters: 35
It may be slow but this makes something that is not possible as far as I know in Photoshop, and is to make the reaction diffusion FOLLOW THE SHAPE and contour of the source image and that is the real beauty and amazing thing of this filter

Quote
Rachel Duim wrote:
Funny how damn slow this filter is... anytime I do reaction diffusion i use AppleScript to do it in Photoshop. It was an interesting experiment.


Sorry that I have no idea about scripting with AppleScript with Photoshop, but can you really make in AppleScript the same as it happens in this filter where the reaction difusión effect Is created from the source shape AND keeps correctly the contour of it ?

Searching for AppleScript effect I found this CNC software reaction diffusion that just found and not tested yet

AWESOME AND AMAZING ANIMATED REACTION DIFFUSION IN ADOBE AFTER EFFECTS

Alive tool for After effects Reaction Difussion tool

And in the behance page you can see even more

Alive tool in Behance
  Details E-Mail
Rachel Duim
So Called Tortured Artist

Posts: 2498
Filters: 188
Yes, it is possible in Photoshop. I wrote actions in Photoshop CS6 that were called from the AppleScript. However I no longer use Adobe products and don't have the actions anymore. The basic loop is Blur > Highpass > Threshold repeated as much as possible. I am working on a more basic version of this filter, it allows changing the parameters to the 3 components in the reaction diffusion loop. Found some unusual results, I've attached one.

The gmic.eu filter set (used in GIMP, Paint.Net etc) has a color reaction diffusion filter in Testing / Lylejk / Lylejk's Stencil

Math meets art meets psychedelia.
  Details E-Mail
SpaceRay
SpaceRay

Posts: 12298
Filters: 35
I still think that this is really an awesome and amazing filter.

Quote
Rachel Duim wrote:
The original is 4.5 iterative passes. I tried one with 12.5 passes, and one with 16.5.


After so much time, I am ashamed that I have missed a very important thing that I have just seen, and is HOW do you count the iterative passes?

I mean that there is not available a setting for the passes to be modified, and inside of the filter there is NO LOOP component, so please, can you be so kind to tell, HOW have you increased the passes to 12.5? It is 3 times the 4 passes plus the final 0.5 pass.

Do you multiply the original 4 passes 3 times so it makes 12? But HOW you do it?

Thanks very much for your possible help and sorry that I did not notice this before
  Details E-Mail

Join Our Community!

Filter Forge has a thriving, vibrant, knowledgeable user community. Feel free to join us and have fun!

33,711 Registered Users
+18 new in 30 days!

153,531 Posts
+39 new in 30 days!

15,347 Topics
+72 new in year!

Create an Account

Online Users Last minute:

22 unregistered users.