Filter Forge - Grass by Sphinxmorpher

Active Topics Search

Crapadilla lvl 52 Filter Weaver and Official "Filter Forge Seer" Posts: 4365 Filters: 65	Grass by Sphinxmorpher http://www.filterforge.com/filters/4739.html --- Crapadilla says: "Damn you, stupid redundant feature requests!" ;)
	Posted: November 19, 2007 6:41 am	Details E-Mail
Crapadilla lvl 52 Filter Weaver and Official "Filter Forge Seer" Posts: 4365 Filters: 65	Nice one, but way too slow, IMHO. It renders over 2 minutes on my Dual QuadCore! You should find ways to speed it up. A start would be replacing the 4 worley noises with perlin noises. Since you're motion blurring the small cell dots anyway, you could as well use perlin dots. Talking about motion blur: Just stretching the perlin could have a similar effect, and in the case of perlin noise, you can even rotate it freely. Hence the motion blurs can be removed completely! Then you're using 5 high-passes where 1 would suffice: Just use one highpass AFTER blending all the noises. Just a few thoughts... --- Crapadilla says: "Damn you, stupid redundant feature requests!" ;)
	Posted: November 19, 2007 6:46 am	Details E-Mail
Sphinx. Filter Optimizer Posts: 1750 Filters: 39	Hmm.. I tried your optimizations, but I'm not sure about the result, looks cool and looks like vegetation of some sort, but grass? Njyldgarkn init/prepare process x no. of render blocks x no. of passes
	Posted: November 19, 2007 12:29 pm	Details E-Mail
Sphinx. Filter Optimizer Posts: 1750 Filters: 39	While the worleys can be replaced alright (but its not really worth it unless the motion blurs go too), I'm not sure about the rest... it seems like a crude approximation of what I tried to achieve: the motion blurs where there to stop the "leafs", the many highpass were there to create a better entanglement combining via lighten Njyldgarkn init/prepare process x no. of render blocks x no. of passes
	Posted: November 19, 2007 12:32 pm	Details E-Mail
Sphinx. Filter Optimizer Posts: 1750 Filters: 39	Here's another one with less stretch - looks very nice, but its more like a hedge or something: Njyldgarkn init/prepare process x no. of render blocks x no. of passes
	Posted: November 19, 2007 12:40 pm	Details E-Mail
Kraellin Kraellin Posts: 12749 Filters: 99	the last one looks good, like more closely cropped grass on a putting green perhaps. you might also try mixing some of the perlins or even the worleys into a refraction. just make sure you add any high passes after and not before the refraction. and an elevation gradient is always good for mixing noises as well. If wishes were horses... there'd be a whole lot of horse crap to clean up! Craig
	Posted: November 19, 2007 1:14 pm	Details E-Mail
Crapadilla lvl 52 Filter Weaver and Official "Filter Forge Seer" Posts: 4365 Filters: 65	High-Passes are painfully slow, even more so on worleys, let alone motion-blurred worleys or motion-blurred worleys on a height tree (which are three times slower even). Therefore one should always strive to simplify the filter trees leading up to highpass components as much as possible. In case the highpasses prove irreplaceable, one can at least optimise their rendering times. Now I don't know whether one could achieve the original look without highpasses, but I have the feeling this should be possible with some serious experimenting and tweaking. What one CAN do is approximate the worley/mblur combination quite sufficiently with a perlin noise (see the unconnected filter tree arm inside the attached filter), IMHO. Oh, and I agree, the second image you posted is definitely very nice. Grass blatantly crapified.ffxml --- Crapadilla says: "Damn you, stupid redundant feature requests!" ;)
	Posted: November 19, 2007 1:22 pm	Details E-Mail
Crapadilla lvl 52 Filter Weaver and Official "Filter Forge Seer" Posts: 4365 Filters: 65	I hope you haven't given up on this, Sphinx! Btw, you could also submit your 'fast grass' as an independent filter. --- Crapadilla says: "Damn you, stupid redundant feature requests!" ;)
	Posted: November 25, 2007 8:52 am	Details E-Mail
Sphinx. Filter Optimizer Posts: 1750 Filters: 39	Don't worry - I'm definitely not done with this one - its slow as hell - way beyond whats reasonable, I'm just taking a break from the forging atm. The performance of the filter depends very much on the hardware - a dual/multicore is actually not an advantage here (which is quite strange) it seems. My theory is that the high amount of noise and bitmap based modules actually thrashes the CPU caches and/or memory streaming on dual/multicores. On my single core Athlon XP it performs ok (the machine I designed the filter on), but on my DAW dualcore badass machine its slower.. go figure! Njyldgarkn init/prepare process x no. of render blocks x no. of passes
	Posted: November 25, 2007 10:04 am	Details E-Mail
Sphinx. Filter Optimizer Posts: 1750 Filters: 39	Ok, new version in. The performance of the first version were unacceptable. This new version might not look 100% the same as the first version, but the new one is alot faster, and it has more depth to it I think. If you want the old mode avail. let me know - I'll check in an update with a "Legacy Mode" checker. Njyldgarkn init/prepare process x no. of render blocks x no. of passes
	Posted: December 10, 2007 2:21 am	Details E-Mail
Crapadilla lvl 52 Filter Weaver and Official "Filter Forge Seer" Posts: 4365 Filters: 65	I still like this variation best... --- Crapadilla says: "Damn you, stupid redundant feature requests!" ;)
	Posted: December 11, 2007 7:38 am	Details E-Mail
StevieJ Designer/Artist Posts: 11264 Filters: 163	Nice one!!! I like the last sample best too..... Steve "Buzzards gotta eat...same as worms..." - Clint :)
	Posted: December 11, 2007 10:43 am	Details E-Mail
Sphinx. Filter Optimizer Posts: 1750 Filters: 39	Yeah it looks like soft moss or well trimmed grass.. perhaps it should be the lead preset? ... well I should really tweak out some new ones as the current ones more or less are from the first version (which I'm considering adding as a "Legacy Option") Njyldgarkn init/prepare process x no. of render blocks x no. of passes
	Posted: December 11, 2007 10:47 am	Details E-Mail