PDA

View Full Version : Alittel 931 vs 95 direct load samples



mis
08-27-2008, 11:58 AM
i did today a render test 931 vs 95
all i did was load the scene made in 931 into 95 as well

and render both 1 by 1 and yes i know 95 tread different on the options abit
but still, check the render times.

a scene for my upcumming showreal

here is first 4 images with rendertimes

Frame 0 931 = 3 minutes 23 sec : 95 = 11 minutes
frame 150 931 = 5 minutes 52 sec : 95 = 16 minutes 16 seconds

to more images that i havent posted, as i can only link 5 is

frame 280 931 = 8 minutes 50 sec : 95 = 25 minutes 4 sec
frame 300 931 = 8 minutes 18 sec : 95 = 22 minutes 32 sec

almost feels like my quad core just became a single core

hope its better if the scenes is fresly build up from grownt in 95


, ill post one more image with the next 2 images

mis

mis
08-27-2008, 12:00 PM
heres the 2 others

rendertime wasin first post

jayroth
08-27-2008, 12:22 PM
i did today a render test 931 vs 95
all i did was load the scene made in 931 into 95 as well

and render both 1 by 1 and yes i know 95 tread different on the options abit
but still, check the render times.

You cannot simply ignore the fact that the options are different. They are DRASTICALLY different, as has been discussed in other threads. If you choose to properly use the new features, your results will be markedly better than 9.3.1. If you do not, then you will get results long the lines of what you have here.

There is no "but still."

9.5 has had some pretty drastic and dramatic improvements build into it. Many of these improvements have to do with radiosity and caching of radiosity, which makes it easy to get great results over a render farm. The Radiosity Guide created by Tom Bosschaert (Exception) does a great job of illustrating proper use and concepts of the radiosity system. There are several threads which reference the guide, and you can find the links in them. Give it a read; it will help you to get the most out of the new features of v9.5.

Great looking images, BTW.

mis
08-27-2008, 02:18 PM
well i will be frank and say that i havent had the time to follow all the beta versions and my mind also tells me that the versions has changed reguarding light and speed and do

, my brain is just wondering now that ime rather used to the 9 3 1 way to set up stuff

but then have to figure out from ground zero again how to set up
to get deacent rendertime.

that ill take some time

i have seen the several pages guide hoever my dyslexia is not treading me well when reading large texts and so.

so when theres no like translation guide or simular
i bet it will take some time to get used to tweaking the new parameters to not get renders that is 4 times as long

any bright clues are very wellcome


mis

Exception
08-27-2008, 03:47 PM
mis: I understand your frustration, but some eggs need to be broken in order to move ahead. One could argue these eggs are used and made into a nice omelette for 9.5, as the only drawbacks are that purely loading an old scene and rendering might take longer.

You might want to check your area light settings. The quality of area lights went up markedly, and you can reduce their quality now.

Also, if you scene uses Adaptive Sampling, you need to turn it off to compare render times. Since the brightness and contrast is different, the AS system will change its rendered samples, so it's hard to compare.
I'd be interested to know if the render times are slower in normal pass rendering, preprocessing, or radiosity preprocessing.

That said, there is still the off-chance you ran into a bug somewhere... see if you can find the parameter that affects the render time the most. I'd suggest starting with GI and Light settings... those would be normal behaviour to be slower or faster. If it has to do with transparent polygons, I'd be very interested as I;m trying to find a possible bug with very slow rendering transparent polygons, but have not been able to isolate the case.

The radiosity guide is on my site which is in my signature. I hope it helps.

mis
08-28-2008, 04:40 AM
hey Exception

the current scene that i was testing and also posting here
and in meny other scenes i use AS set to 0.02

wich ofcause takes time but to smooth out better, expecialy
with the level of dof and motionblur that i have in the scene to remove the grain

however ime not shure if 0.02 in 931 is equal to 0.02 in 95

Reguarding the arealights thats usefull information, i ahve a few of those
wich i also havent changed but some of em mignt not be set to cast shadow at all
as its a mix of arear + gi + some occlusion.

theres not much transparent in the referred scene so i doublt that can cause much damage on it
i did a few tests on another small scene for fun yesterday

i notised that if you load a scene from lw 931 the Angle Tolerence is set to 0
when it in a default scene in 95 is set to 45

i then testet a sample scene with 0 and 45 and it seems
that when on 45 i obtainet better shadows at about half or 2/3 the render speed then with it set to 0

so on that test i made it was defenetly a gain both in speed and quality to set it to 45 (faster render)

mis

mis
08-28-2008, 04:46 AM
And Oh

your guy behind the radiosity guides
Thanks for your nice work though i must say being a dyslexic i have it prety har
getting much understanding out of it, ime more visual / voice like
as the larger texts mess up in my brain

so make somthing more readfriendly or so guidence on it do let me know

would really be most aprecated
mis