Messages 1 - 45 of 62
First | Prev. | 1 2 | Next | Last |
Vincs
Posts: 13 |
Hi everyone,
I'm a graphic designer and i'm testing filter forge since 3 weeks now. I made some many tests about time rendering at work and wow... How is it possible? All my tests made with a very simple filter : http://www.filterforge.com/filters/9436.html with no antialiasing, in a 2000x2000px resolution. My time render is : Last macbook Air (i5 2x1.8ghz - Ram 4go) : up to 1h MacPro C2D 2,5ghz : up to 2h And the best, in my work's renderfarm (over 88cores at 2ghz - windows 2008 r2) : over 1h30... wow.. Somebody can explain that? I think i have a problem, but i don't find it. Thanks in advance, i'm going crazy... edit : and for information, filter forge in macosX lion isn't stable (over 4,5 crash per hour..). (Sorry for english, i'm french...) |
|||||||||||
Posted: July 10, 2012 12:31 pm | ||||||||||||
SpaceRay
![]() |
Welcome to the MAIN AND GREATEST PROBLEM of Filter Forge that is SLOW!!
This is a self generated filter of FF so is slower than others filters, BUT if you say that it takes 1 HOUR, 2 Hours and 1,5 Hours to make a 2000 x 2000 is TOO slow, and I think that there may be something wrong in the settings or something weird is happening I suposse that you are testing the FF Professional version and you have activated the multicore option setting. If you have a 88 core computer it CAN´T take 1,5 hours to render ![]() ![]() ![]() I will make a test myself and will tell you |
|||||||||||
Posted: July 10, 2012 12:38 pm | ||||||||||||
Vincs
Posts: 13 |
Thanks SpaceRay, you're quick !
I understand FF is slow, but i think i have a problem too. For now, i'm using the free trial but it's full fonctionnal and yes, i activated multithreading, and use a maximum ram usage at 90%. For information, if i use the same components but in a "filter type : simple", time is perfect : down to 50secondes (with AA) on the macbook air. But when it's a "filter type : surface" : bad results. Thanks for your help, i'm interesting by your test, if you can. edit : the 88cores is a cluster, not a single computer. I think FF isn't optimize for that. |
|||||||||||
Posted: July 10, 2012 12:45 pm | ||||||||||||
SpaceRay
![]() |
You are right, you have a problem.
I have just made a test for you, and in my computer it takes 10 MINUTES to render the 2000 x 2000 of the Simple Stone filter |
|||||||||||
Posted: July 10, 2012 12:56 pm | ||||||||||||
Vincs
Posts: 13 |
ok thanks.. I think FF don't like MacOs..
Can you give me your hardware configuration please? |
|||||||||||
Posted: July 10, 2012 12:57 pm | ||||||||||||
Kraellin
![]() |
i just ran a 2000 x 2000 on my i7 pc, with anti-aliasing turned on. the result was 27.26 seconds. so, it's definitely NOT filter itself which is giving you problems.
If wishes were horses... there'd be a whole lot of horse crap to clean up!
Craig |
|||||||||||
Posted: July 10, 2012 1:00 pm | ||||||||||||
Vincs
Posts: 13 |
wow.. ok thanks for your rapidity SpaceRay and Kraellin.
I'm going to test with windows on the macbook, i think Macos is the problem. Thanks guys !! |
|||||||||||
Posted: July 10, 2012 1:02 pm | ||||||||||||
SpaceRay
![]() |
If you change the result you will not get the same result I think, although it may be much faster. Here below I show you my configuration of FF ![]() |
|||||||||||
Posted: July 10, 2012 1:03 pm | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 10, 2012 1:03 pm | ||||||||||||
Vincs
Posts: 13 |
Ok, thanks !
I am currently testing on Windows and I come back here, this could perhaps help someone else .. |
|||||||||||
Posted: July 10, 2012 1:05 pm | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 10, 2012 1:06 pm | ||||||||||||
SpaceRay
![]() |
PLEASE, PLEASE Tell me HOW you have been able to get JUST 27.26 seconds with a intel i7 and I that also have an intel i7 have taken 10 minutes ??? ![]() ![]() I have an intel i7 2600K 3.5Ghz And I have activated the 4 cores + the 4 HT cores ![]() |
|||||||||||
Posted: July 10, 2012 1:09 pm | ||||||||||||
SpaceRay
![]() |
Please, please, anyone else can make a testing on your computer on how much it takes to render this 2000 x 2000 of Simple stones filter ?
I can´t believe that I have taken 10 minutes and Kraellin ONLY 27 SECONDS !!! How is this possible ? I have a wrong and bad configuration ? Something is happening ? ![]() ![]() ![]() ![]() |
|||||||||||
Posted: July 10, 2012 1:16 pm | ||||||||||||
Vincs
Posts: 13 |
Ok, so MacOs is the problem, don't use it for rendering.
With AA on, i rendering in 75secondes (2000x2000). Configuration : Windows Seven MacBook Air 2012 Ivy Bridge i5 1.8ghz 4go ddr3 SSD Thanks for your help, and if i can help you SpaceRay with another test, just tell me. |
|||||||||||
Posted: July 10, 2012 1:24 pm | ||||||||||||
SpaceRay
![]() |
How can you be able to make it in 75 seconds with a i5 1.8 GHz ???? ![]() Please, please, could you take a screenshot of the same settings in the option panel I have shown above to see what is wrong with my own configuration ? Does the 75 seconds is WITH RENDERING AND SAVING the file ? Because in my configuration IF you use the reduced preview you get a 29.81 seconds BUT WHEN YOU WANT TO SAVE IT to use it it takes 10 MINUTES TO RENDER AGAIN the result!!! ![]() ![]() ![]() ![]() ![]() |
|||||||||||
Posted: July 10, 2012 1:44 pm | ||||||||||||
Kraellin
![]() |
an i7 has 8 cores and 8 more virtual cores. i have 6 gigs of ram and an ssd drive (solid state drive) as my C:/ drive. i keep my system fairly trim. i dont allow a lot of startup programs. and yes, it's the simple stone filter by jitspoe and yes, my imaze size is set for 2000 x 2000 (i just checked both to make sure). i also turn off a lot of unnecessary services (go to blackviper.com for info on this). i also monitor my cpu usage, drive usage and network usage with those little gadgets on the desktop. this helps me understand and handle lags, memory leaks, network traffic and so on.
also note, my 27.26 seconds was on the default preset. i dont know if any of the other presets would run the same. as a suggestion, go to www.blackviper.com for help with turning off unneeded windows crap! do not use active virus scanners. put all that stuff on manual, except maybe for the active email scanning. programs that actively scan for problems actually become a problem in that they tend to eat a lot of system resources. i can always tell when filter forge is about to crash due to the memory allocation bug because the ram usage will hit a certain level and if it gets pushed over that, boom! understand that folks that want to sell you anti-malware and anti-anything-bad-happening-to-my-computer are playing on your fear of those things happening. it doesnt mean their programs are any good. i use a minimum of anti-virus and anti-malware junk, but i use reputable stuff. the trend lately in anti-virus has been to give you the anti-virus for free and then sell you the anti-malware stuff. the more stuff you buy, the more it clogs your computer. i use a good firewall, a good, free anti-virus and one other very smart program that keeps crap out of my startup. for browser problems i use hijackthis, for the most part. computers get bogged down. this is especially true if you just allow programs to put anything they want, anywhere they want. i dont. i disallow a lot of stuff into my startup routine. my computer boots up in about 30 seconds, fully to windows with everything loaded that i want loaded. the SSD drive helps a ton here. it's a LOT faster than a normal drive. filter forge is a hefty program. when it first went into beta i seem to recall it being around 20 gigs, maybe a little less, maybe a little more. it's now over 60 gigs. surface filters do generally take longer to render than simple ones. anti-aliasing does tend to take long that not. filters can sometimes be optimized with little loss in function, but not always. i often see my cpu usage climb to near maximum when i run a filter. just remember, ram is the workspace and cpu is the worker. those two combined give you your overall speed. with filter forge currently capped at 2 gigs, the 32 bit maximum, your cpu has to work harder. if you dont have 2 gigs available, then your swap file kicks in and that slows things WAAAAAAAAY down! oh, and keep your swap file fairly large. if you have 4 gigs of ram or more, you'll probably never hit the max with filter forge and your swap file wont kick in, but you shld keep it fairly large anyways. oh, and if you dont know what a swap file is, it's the way windows makes up for the lack of ram. it simply takes a portion of your harddrive and uses it as if it was ram. it allocates a certain amount and simply uses it for ram. you can set this amount manually, if you want, or just let windows manage it. the choice is yours and i dont know if one is better than the other, but if your harddrive light is coming on a lot when rendering, i'd start buying more ram. the reason filter forge can only use 2 gigs of ram is that it's 32 bit and 32 bit basically can only count as high as 2 gigs. it just goes dumb after that and cant use anything larger. 64 bit would remove that particular barrier and we'd love it if FF could go 64 bit, but vladimir has already stated that's not going to happen any time soon, i seem to recall. that was a while back, so i'm not sure how that stands right now. also, if you have any of the occlusions turned on, like ambient occlusion, in the surface filter, this can add a ton of time to rendering. sometimes that doesnt make a bit of difference to the look of the render so just turn it off when you can. so, trim the fat from your system and things shld run more quickly! If wishes were horses... there'd be a whole lot of horse crap to clean up!
Craig |
|||||||||||
Posted: July 10, 2012 1:48 pm | ||||||||||||
Vincs
Posts: 13 |
I can take a screenshot SpaceRay, but i use the Command line batch with a default xml for rendering.
I'm not a hardware specialist, but Ivy Bridge is really good, i don't know what is the generation of your processor, but it's important. My brother have one of the first i5, and time render in 3d soft are best in my computer. |
|||||||||||
Posted: July 10, 2012 1:58 pm | ||||||||||||
Kraellin
![]() |
oops! an i7 has 4 cores and 4 virtual.
also, i was using the smaller preview for the render. i'm rendering in full view now. hang on... it's definitely a bit longer but shldnt be 10 minutes. this will be with ambient occlusion turned on and anti-aliasing set at 5 samples and 'actual' size turned on. all cores are working their butts off, from about 94% to 100% and ram is hanging around right at 2174MB. hehehe, ok, my apologies, SR. with the actual size turned on this is running quite a bit slower. lol. my bad. ok, 8 minutes, 40 seconds for the full render in 'actual' size. If wishes were horses... there'd be a whole lot of horse crap to clean up!
Craig |
|||||||||||
Posted: July 10, 2012 1:59 pm | ||||||||||||
Kraellin
![]() |
with ambient occlusion turned off, 1 minute, 10 seconds.
If wishes were horses... there'd be a whole lot of horse crap to clean up!
Craig |
|||||||||||
Posted: July 10, 2012 2:01 pm | ||||||||||||
Kraellin
![]() |
with ambient occlusion and anti-aliasing turned off, 16.74 seconds.
If wishes were horses... there'd be a whole lot of horse crap to clean up!
Craig |
|||||||||||
Posted: July 10, 2012 2:04 pm | ||||||||||||
Vincs
Posts: 13 |
Interesting Kraellin.. Thanks for your posts.
I recheck my xml file and 75 secondes is with AA OFF. Sorry about that ! With AA on, rendering time is to 521 secondes, so 8,5 minutes. Don't have a heart attack SpaceRay, sorry man ! ![]() |
|||||||||||
Posted: July 10, 2012 2:17 pm | ||||||||||||
Morgantao
![]() |
Vincs, in the lighting tab of the filter, click outside the shiny orb (which is the enviroment setings) and you get the ambience options.
Turn shadowing off to remove any ambient rendering. |
|||||||||||
Posted: July 10, 2012 2:37 pm | ||||||||||||
Vincs
Posts: 13 |
yes, i see that, thanks Morgantao !
![]() |
|||||||||||
Posted: July 10, 2012 2:44 pm | ||||||||||||
SpaceRay
![]() |
Thanks very much for all the information
One VERY IMPORTANT THING if you use the Reduced view, the render time is NOT REAL and is only for the 600 x 600 reduced size preview if you really want to know THE REAL time it takes to render and save the result you have to turn on the "ACTUAL SIZE" view and then you will be able to know the real and true render speed. |
|||||||||||
Posted: July 10, 2012 2:49 pm | ||||||||||||
Morgantao
![]() |
LOL, by the time I finished typing you already got it, and had time to edit your post... I'm so sloooow ![]() |
|||||||||||
Posted: July 10, 2012 2:53 pm | ||||||||||||
SpaceRay
![]() |
WHich of the Antialias have you made this test with of the ones available in the list ![]() I mean if it is 5 samples, 9 samples or... and also if it is "edges only" or "all pixels" |
|||||||||||
Posted: July 10, 2012 3:01 pm | ||||||||||||
SpaceRay
![]() |
I have some interesting results I have been making in some tests right now
The first 3 ones is of NO use for the real render time when saving the render result to use it somewhere, the real true time IS when you have the Actual size active 2000 x 2000 Reduced Mode (600 x 600) Antialias 5 Samples, Edge Only Ambient Occlusion ON 29.91 seconds 2000 x 2000 Reduced Mode (600 x 600) Antialias Off Ambient Occlusion ON 11,15 seconds 2000 x 2000 Reduced Mode (600 x 600) Antialias Off Ambient Occlusion OFF 6,90 seconds 2000 x 2000 Actual Size Mode Antialias 5 Samples, Edge Only Ambient Occlusion ON 8.53 minutes 2000 x 2000 Actual Size Mode Antialias Off Ambient Occlusion ON 1,50 minutes 2000 x 2000 Actual Size Mode Antialias Off Ambient Occlusion OFF 12,72 seconds The ambient occlusion setting used here are the default ones that came with the filter as shown below here ![]() |
|||||||||||
Posted: July 10, 2012 3:10 pm | ||||||||||||
SpaceRay
![]() |
I think I have solved the mystery behind the great difference in rendering times.
This the real render time when saving the file with antialias and Ambient oclusion
This is from a Reduced mode preview that IS NOT the real time and as shown above I have been to get 29.91 seconds in in this same mode.
This must be also a Reduced mode preview with the 600 x 600 time
YES, this is better than 27 seconds ![]() ![]()
Also this is REAL time in Actual Size |
|||||||||||
Posted: July 10, 2012 4:43 pm | ||||||||||||
SpaceRay
![]() |
SO it seems that the Antialias and Ambient Occlusion makes the filter MUCH SLOWER with a great difference
Between 8.53 minutes with Antialias 5 samples and 1,50 minutes without Antialias is really a GREAT difference. And Here I show the real size of the render to see that there is no real difference in quality (at least for me) that could justify waiting for the 9 minutes instead of 2 minutes without. Here Below is WITH antialias 5 samples ![]() |
|||||||||||
Posted: July 10, 2012 4:57 pm | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 10, 2012 4:58 pm | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 10, 2012 4:59 pm | ||||||||||||
Vincs
Posts: 13 |
5 samples / Edges Only. I think you are a render-time master now ! ![]() |
|||||||||||
Posted: July 10, 2012 5:02 pm | ||||||||||||
Vincs
Posts: 13 |
Yes, sorry for editing ! ![]() |
|||||||||||
Posted: July 10, 2012 5:10 pm | ||||||||||||
Morgantao
![]() |
SpaceRay, I totaly agree than in this filter the 8 minute difference between AA5 and no AA is a waste of time... I can hardly see the difference.
There are filters that the AA is a must, but there are filters that it's just a wate of time. I guess it's worth trying to turn off AA if a filter is sluggish. ![]() |
|||||||||||
Posted: July 10, 2012 5:55 pm | ||||||||||||
SpaceRay
![]() |
Do you mean that you can know when Filter Forge is going to crash by looking at the memory level in the Task Manager ? Which is that certain level when it goes Boom! ? ![]() |
|||||||||||
Posted: July 10, 2012 5:57 pm | ||||||||||||
SpaceRay
![]() |
I am making a 6000 x 6000 test
YES, is true this is very important, thanks |
|||||||||||
Posted: July 10, 2012 6:00 pm | ||||||||||||
SpaceRay
![]() |
I have made another test with 6000 x 6000 that shows much better the difference in time between one mode and another
6000 x 6000 Reduced Mode (600 x 600) Antialias Off Ambient Occlusion OFF 3,90 seconds 6000 x 6000 Actual Size mode Antialias Off Ambient Occlusion OFF 11.56 minutes 6000 x 6000 Actual Size mode Antialias Off Ambient Occlusion ON 1 Hour 18 Minutes 43 seconds Activating the Ambient Oclussion is 1 HOUR AND 16 MINUTES MORE !!!! I do not dare to make a 6000 x 60000 Antialias ON and Ambient Occlusion ON test ![]() ![]() |
|||||||||||
Posted: July 10, 2012 7:07 pm | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 10, 2012 7:12 pm | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 10, 2012 7:17 pm | ||||||||||||
SpaceRay
![]() |
AA5 = Antialias 5 Samples
AND I totally agree with you that is not worth the 8 minute difference for getting the same or nearly the same result. Is true that this is IN THIS specific case, and on others filter the AA and also Ambient Occlusion may be important or really needed, BUT as you say, is really worth trying to see to turn them off first and see if the quality is much worse than the filter preview or preset. |
|||||||||||
Posted: July 10, 2012 7:23 pm | ||||||||||||
Skybase
![]() |
... this calls for a video tutorial on how to speed things up just a notch!
|
|||||||||||
Posted: July 10, 2012 9:55 pm | ||||||||||||
SpaceRay
![]() |
Does this means that you are going to make a video tutorial to explain about this important and interesting topic? Thanks very much |
|||||||||||
Posted: July 11, 2012 1:32 am | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 11, 2012 1:38 am | ||||||||||||
SpaceRay
![]() |
||||||||||||
Posted: July 11, 2012 1:40 am | ||||||||||||
SpaceRay
![]() |
By the way the Mystery of why Vincs got this very slow render times is STILL unsolved so please GMM or any other, help him be able to solve this problem, thanks
|
|||||||||||
Posted: July 11, 2012 3:40 am |
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!
24 unregistered users.