Circular Angle Gradient 2 by uberzev

Circular Angle Gradient 2 by uberzev
http://www.filterforge.com/filters/671.html

New in version 3.0:

-Nearly 300% faster render time
-Totally new and simplified internals

Here's two simple mutations...

Excellent! Now slam in a switch quickly!

Cool. Thought about doing it myself but I'm still only a padawan when it comes to the Ancient Art of Gradients!

Quote
Crapadilla wrote: Cool. Thought about doing it myself but I'm still only a padawan when it comes to the Ancient Art of Gradients!

If you want to give it a try you have my blessing.

There seems to be a slight discrepancy between the old & new versions. I noticed this as I was about to update my Clockwork Factory filter with the new circular gradient. Check the attached file...

Circular Gradient Old t New.ffxml

Bump!

This one gets the "StevieJ's Slickest Thing Since Sliced Bread" award.... .....What??? You haven't heard of that award??? You must be new to Filter Forge LOL....

Thanks Zephos, I'm honored even though you called me Kraellin.

Nice one, very usefull

Heck yeah! This filter is the leading flagship of the circular gradient and multi-spoke effect on Filter Forge! Every time a user cranks up the latest version of Speedliner Lv2, they're using ubersev's Circular Angle Gradient (Not to mention Any Angle Gradient and Proportional Gradient as well) and I'm sure there are a lot of other filters that use it as well ^_^ Snippets are undoubtedly the unsung heroes of Filter Forge, especially ubersev's ^_-

Indeed it is! Nice observation..
I noticed some constant factors that can be taken advantage of (see attachment). The results are not identical though.

Circular Angle Gradient v4.0 - Sphinx.ffxml

very clever how you worked it out

Right, thats also what I did, but there seem to be a very small difference (I exported the four through a checker comp., saved as 16bpc png and entered the rgb values from there). I think perhaps some decimals were lost in that process.

uber, any reason you only simplified the filter tree for one of the profile gradients in v4.1?

here you go:

<ProfileGradient id="profilegradient 02">
<Color1 red="0.2141" green="0.8543" blue="0.0056" alpha="1"/>
<Color2 red="0.2141" green="0.0056" blue="0.854" alpha="1"/>
<Offset value="0"/>
<Rotate value="0"/>
<Repeat value="1"/>
<Profile value="0"/>
<Mirror value="false"/>
</ProfileGradient>
<ProfileGradient id="profilegradient 03">
<Color1 red="1" green="0.0509" blue="0.0509" alpha="1"/>
<Color2 red="0" green="0.5225" blue="0.5225" alpha="1"/>
<Offset value="0"/>
<Rotate value="1"/>
<Repeat value="1"/>
<Profile value="0"/>
<Mirror value="false"/>
</ProfileGradient>

I saved the colors made via the curves in 4.0 via some checker outputs, as 16 bit png. Opened in PS, converted to 32 bit (to get floating point values) and used the color dialog to get the correct values. The values were then entered in the filter XML source (since FF uses 0..255 range and not 0..1 in the UI).
Note that the intensity of the colors are not similar to what you see in the original, however the relational hue difference is the same, and thus the correct hexagonal hue wheel is created (the hue channel is not influenced by intensity).
You can validate by looking at the saturation channel (it should form some sort of hexagonal gradient for this type of hexagonal quantized angular gradient).

Alright - used the 32bit per channel exr format to figure out the constants - see the attached filter. Don't be fooled by the gradients looking slightly different - this is only an overall intensity difference due to a different color handling model in PS - the hues are still the same.

Angular Gradient.ffxml

Quote
rgoer wrote: Can you explain why the (hue 90°)..(hue 270°) profile gradient needs that ever-so-slightly funky profile? I don't get it.

Honesty I came up with this through a lot of trial and error so I'm not sure I can give you a proper explanation. However if you zoom out a bit in the filter editor you'll see the old version which shows more clearly why the wave curve is used.