Calling all filter Gurus.
... such, FF is resolution independent as advertised.
However - internally - many filters are 'hardwired' to a certain maximum detail level for the sake of rendering performance: The high detail that would be possible is not required for low-resolution renderings (most filters are 'optimized' for the filter library resolution of 600x600 px). This internal 'detail restriction' is often done with filters that use noise components, since these can be very slow. Also, filters that use blur-type components may look different at higher resolutions because the blur radius is measured as a percentage of image size.
Still, it would probably help if you posted some examples from those filters that supposedly ...
What is the relation between the resolution size and the FF filter ?
... the list below and have had no issues with them....Tried at 9000x9000,15000x15000 and 20000x20000.....I have the AA settings set to render "All Pixels" 25 samples.....I also turn my ram usage to 40% so I can still do other things while FF renders....
Also I am on a slow machine by todays standards....Only an Intel Dual core OCed at 3gig....
Concrete
Plastastic
Woeful Wood
Decay
Energy
Filtration
Ink Ghost
Lost Letter
Grass
Muscle Fiber
Crackles
Lake Bed
Odd Rock
Rock Salt
Cyber Circuits ...
Beta 2 Gallery
Author: Skybase. "Kraellin wrote:
can it be tightened down a bit more to give more detail?"
Yeah that's true. I'm thinking of redesigning some aspects of the filter itself. 5 mins render is a bit too slow for me, and I wanted to go for something under 2 mins max. the AO is actually the hindrance but also the beauty aspect which I can probably make fake AO maps. Who knows. It's getting there.
Oh and yes, can totally become a magnetic field sort ...
Request: GPU acceleration (CUDA and OpenCL)
... basis similar to the new node setting locks so that in the end we can switch them in/out of background rendering individually, on selected nodes/node groups, and across the whole filter as a switchable and pre-selected default setting. If they were GPU rendered to top if off we'd be all set :)
Also, any sort of pre-allocation on the Ram or HD's that might help speed things up should be considered. In cases with very large images 10k+ on either side, it seems to allocate ram at a slow, trickled pace. That's not an insult nor is it unreasonable that it behaves this way. However, some of the jobs I would like FF to do will require 25GB+ of Ram. That's fine by me. I have 64GB so it's acceptable for FF to grab all that it needs. In ...
NEED FOR SPEED!!! :)
... realistically, I've read the most I can stand to gain on optimized apps is closer to 80% gain, so not quite a doubling of speed, but still worthwhile. But my gladdiction to Filter Forge is certainly a compelling reason to get a quad-core compy!
Really slow filters would be useful with some sort of batch render too, the thing you could let run while you get a cup of coffee. :)
FF3: Slow
... to say, and that after the answers from GMM and knowing that most probably FF inc. will not do anything to make a faster render engine, make it have more ram to make it faster, convert it to 64 bit, have GPU support or whatever that could speed the render and preview time of FF I will not continue requesting and complaining about this topic.
From my point of view now, there is no point in continuing this, as we will not get nothing from FF Inc. and only be a problem for GMM.
So if FF is SLOW, the ONLY REAL AND POSSIBLE solution is to have a faster CPU.
For more information about this please see this thread here (http://filterforge.com/forum/read.php?FID=5&TID=9799)
Well you knew I had 2 be working on it
Author: voldemort. After figuring out an infinite mirror you know I had to be working on how to port a swirl to it ---I had several early attempts that worked but rendered Way to slow
Here is the latest and it Renders Really fast :D
http://img143.imageshack.us/img143/3806/a6xj0.th.jpg (http://img143.imageshack.us/my.php?image=a6xj0.jpg)
http://img143.imageshack.us/img143/9476/a5mx7.th.jpg (http://img143.imageshack.us/my....
Inflatable by Carl
... (physics processor) to computational efforts but not right now theese are dedicated chipsets used for their singular purpose
Rendering such as this is strictly cpu --the bigger question is how well does ff take advantage of multple cors, hyper transport ... ... foccus on maximum utilization of chipset tech to enchnce performance since both photoshop and FF can be hogs and together really slow down a system that is less then top notch
HUGE Resolutions = FAIL
... those on that list and have had no issues with them....Tried at 9000x9000,15000x15000 and 20000x20000.....I have the AA settings set to render "All Pixels" 25 samples.....I also turn my ram usage to 40% so I can still do other things while FF renders....
Also I am on a slow machine by todays standards....Only an Intel Dual core OCed at 3gig...."
"GMM wrote:
Filter Forge, Inc
Posts: 1247
joBox, your anger is useless unless you provide more details. What is your OS, it is 32-bit or 64-bit, do you use ...
Mixing two filters.
... several dozen repeats and layers. The result is a seamless tile with either full or partial coverage.
The filter intentionally does not do any resizing or rotating in order to maintain size relationships between objects that are man made and to avoid slowing the rendering down too much. As with the other tiles I publish, we have standardized our tiles at a size of 3600 x 3600 ... so rendering time is a big issue.
I started in working with the same approach in the Genetica 3 beta which will allow you to work ...
PS Plugin Hangs After FF Update
Author: RobinH. I just updated to 10.007. Now when I try to run FF from PS, the interface comes up, but after I choose my filter and press Apply, I get the spinning beach ball. I've tried waiting a few minutes to see if it was just slow for some reason, but even 15 minutes later, and FF still hasn't started rendering. I've also tried rebooting my computer. I didn't have this problem before the update. Any suggestions?
Sharing Noise
... comes with a price: speed. This is one of those things where either we balance looks over efficiency. This is a difficult task, and a good eye for quality really helps establish a relatively fast filter, but even still a minute and 25 seconds for a render at 512x512 is a bit slow in my view.
While this demo presents no parameters there are a couple places where you can easily change the render time while sacrificing quality of the appearance. In the noise distortion node, we can set the "distortion" parameter ...
Storing node results in memory?
... stuff later in the tree it only rebuilds what it needs to. Is this a possibility? This would be my most wanted feature. Making complicated filters is such a pain otherwise.
I've been making a high quality rock texture that as a whole is incredibly slow (on quad core no less). However, when I split it into multiple filters I feel like the process of rendering each step, saving the result, processing the result into the next step, saving that result, and doing it again for a third time is faster than rendering it all in one filter. Why is this? Maybe I can try a test to see if I'm just imagining it ...
Variable-Size Tessalation
Author: SpaceRay. "I am using many blur's"
From my tests and experiments, having many blur components slows down the filter very much, and then so increases much the render time, maybe you should find a possible alternative instead of using too many blurs, unless they are essential and can not be remplaces, or have as few as possible
"I have done some tweaking and optimizing so I think I have it fixed now ...
Filter Forge 3.0 and multiple image support
... thumbnails I just saw...they haven't changed, but I still have to wait every single time I open a filter for that repeated and wasted rendering effort, every single time, all in order to do something as simple as connect a single tile.
If those thumbnails were ... ... all blank.
EVERY filter you re-edit must be repopulated. It make working casually, a little bit at a time to be really slow and a lot frustrating.
Nine Men's Morris (playable)
... set up. I may change things around a bit later.
It needs some more optimization to really be playable at default resolution.
I've gotten it down to 12 - 18 seconds so far, but I think I can do better.
At 300x300 it takes from 2 - 6 seconds to render a frame on my system, which is a little slow, but playable.
Usually I'm looking at the controls during most of that time anyway.
That's on an overclocked i7 920 though, so results will vary.
If rendering time is too long you can render the board and men as seperate layers and play in a paint ...
Ramlyn's Filter Library
Author: Rachel Duim. Interesting filter. The speed improvements help quite a bit (2 or 3 times faster). Larger files probably need a fast 4 core or more CPU with at least 4GB of memory. I did some testing on my 2 machines. Here are the render time results:
old MacBook Pro (2 core CPU): 600x600 1 min 50 sec, 3200x2400 3 hours 7 min
fast Windows 7 (4 core, 8 thread CPU) 600x600 27 sec, 3200x2400 22 min
This is somewhat slow, but not unreasonable. I've attached a test file, a gazania flower I use often for testing.
Tile Flow by Sphinxmorpher
... but the edges are inverted on 1 or 2 axis
there is a solution utilizing your method --by simply remastering the section where you have the image tiled into 9 pieces you can correct the offset errors Ive done this on a couple filters but it does slow down render time
Could also FF 6.0 be more faster in some way now that is 64 bit?
... obtained from changing to 64bit is very low and just a little bit faster according to this thread here below
Some FF6.0 Impressions and Speed Tests
(https://filterforge.com/forum/read.php?FID=17&TID=14155)
I have already suggested that if the main render engine could not be updated, at least maybe could be made faster the preview, as on many filters it is slow
Please, make the 600x600 preview view faster for FF 6.0 and 300x300 (https://filterforge.com/forum/read.php?FID=9&TID=14154)
As you already know it has been requested since long time ago
LIST of threads about FF Render Speed - Requests, ...
Deco Warp by Artsy
... whether there was a whole lot of intent regarding the construction. I know the attached filter has weird controls but I believe it should be significantly faster.
There were two blurs set at 100% in the filter... for no visible reason. This would slow things down a lot when you start rendering at higher values.
Give the attached filter a try and see if it alleviates the problem, if it's faster you might want to have the author take another look at the guts of the original.
