Render time

Hello,

i have a little question about the new version of VRay (1.47.02):
The seem file render with the old vesion of VRay (1.09.03r) is much faster then the new version. For the new version, i have just open the max file and render.

The renders :
1.09.03r

1.47.02

Why ??? :shock:

Vince

(sorry for my english :sweat_smile: )

Well as you can see (I hope) there is a lot more noise in the 1.09 version, basically 1.47 takes more samples at the same settings, you can reduce the settings in 1.47 to better match the time and quality of the 1.09 render.

and dont foget that when you just open and render your rendering with QMC samples in the secondary bounce as 50. they should be 8. lower to 8 and im sure there will be quite a difference

Thanks guys for your fast replays :smile:

I have made many tests but my rendertimes is very long (over 3min)

The file of my test http://www.architecture-design.ch/vincent/testrapidite.max

Thanks for your help :wink:

Vince

the only noise I can see is that of jpg compression… I don’t see a difference between the two versions.

The new versions are simply slower on simple scenes. Fill that scene with trees and displaced grass, and it will be faster in the new builds.

the only noise I can see is that of jpg compression… I don’t see a difference between the two versions.

Load the images up in PS as layers and switch between them, the difference is quite obvious… The difference mode is too convoluted by the compression to be of any use. Vlado has stated many times that the newer versions take more samples at the same settings as before.

If I still have a copy of 1.09.03 at home I will try for myself :slight_smile:

Well a little bit slower :slight_smile: maybe with a little more tweaking…
I don’t know what your gamma is at so mine are without gamma correction.
Mostly just played with the QMC, also turning off ā€œmultipassā€ shaved off a few seconds.
Could probably shave a few more with the GI thresholds but I just
left it on high. Also a few seconds seem to be wasted while the last bucket
finishes it’s GI calc and the other 3 threads wait to start the final render.

maybe this would help to understand…
http://www.chaoticdimension.com/forums/phpBB2/viewtopic.php?t=11692&highlight=grain+grain+grain

best regards
themaxxer

You do not agree that the noise levels in my tests are very close with only a slight increase in time? 10% instead of 53% … :slight_smile:

The only thing I see in both of your tests is that the detail of GI is always better in the 1.09 version. The new build looks very blurred.

To me the GI at the bases of the boxes look very similar, please take the time to show me where you see this huge difference.

http://users.pandora.be/stor1/vray/compare.htm

Mouse over and out of the image.

Note that in 1.09 the corners are darker, in the 1.47 version everything gets blurred. (The 1.09 had a coarser jpg compression… don’t watch that ugly noise from it)

Yes it is a little more apparent here on my monitor at work, still it should be noted that speed increases can be had by adjusted the QMC with little noticeable difference between the 1.47 at 2:11(the unadjusted from 1.09) and the 1.47 at 1:13.

Thanks RErender for yours tests :smile:

Good week-end

Vince

:lol: :lol: :lol: The only thing I see is that you have saved the both images using a diferent JPG compression ratio… nothing here related to GI.

I used the images of rerender, they were already saved in different jpg quality…

Look at the bottom of the boxes, GI is darker in corners as it should be. In the 1.47 example the bottom of the box is dark and the flaar where it touches teh box is light, seems very wrong to me, and this is due to undetailed or blurred GI. This is also the case in the 1.09 version, but less pronounced, so better GI there.

The compression difference was bad science, I have re-upped the 1.09 image with the same compression, my computer was a second faster this morning it must be in a good mood :wink:

Yeah actually the difference in not that great. The GI differs between the two, but one is not better than the other, it’s just ā€˜different’.