Machines rendering differently?

[vlado]

I have an opteron and core 2 duo here both on 64-bit so I will be testing this in a moment.

Best regards,
Vlado

[vlado]

I know it's not much of a help right now, but this seems fine for newer builds:

This is sp1, and there is a clear difference between the Intel/AMD machine:



This is our latest internal version (the images are identical):



Best regards,
Vlado

[dlparisi]

Thank you, thank you . At least I know it's not just my machine that's wiggin out!! BTW, your images match mine exactly (opteron-->opteron and Xeon-->C2d) so it seems fairly consistent (there's a very minor color variation but I think that's just some jpg compression).

Do you have any idea what's causing it and when it may have started? I ask because if necessary I can revert back to regular 1.5 (not SP1) if that fixes it as well. Other than that, when do you think a fix will be available for this?

Also, all of my reinstalling and rechecking seems to have fixed my DR problems (in both Max 9 & 2008 ) from this post (http://www.chaosgroup.com/forums/vbu...ad.php?t=40792) but I'm not sure how that resoloved itself.

[SnipeyX]

Snipey: Did you ever find a solution for this? I'm experiencing similar quirks...

Sorry for the delayed response, we just had the IT department rebuild the machine and it was fine afterwards, so I don't know, looks like you guys have taken the research farther anyhow... good luck!

[victor.nsy]

Hardware differences making bad noise

I read some stuff about hardware difference causing different noise pattern ...
basically ... all machine in a farm should use the same type of CPU if you have
different classes and different brand of CPU ... you will experience subtle
differences in noise pattern.

seems the new build is more stable ... gee

[pinpan71]

i had similar problems of "jumping" reflections and changing brightness of
reflections in animations. all on max9/64+vray 1.5SP1
i ended up using one computer per camera to render the images.
then everything look perfect.
nevertheless a fix for this would be crucial.

[dlparisi]

I read some stuff about hardware difference causing different noise pattern ...
basically ... all machine in a farm should use the same type of CPU if you have
different classes and different brand of CPU ... you will experience subtle
differences in noise pattern.

seems the new build is more stable ... gee

I think the key word here is "subtle". I could have overlooked minor difference in the noise pattern but in this case it was the reflections, lighting and quality that was off. Honestly I may have even not pushed it if I hadn't noticed similar problems in an animation I recently completed. I spent hours then trying to figure out how to fix it and eventually gave up. To fix those frames I sent them all over the network again via BB but avoided rendering them on my local workstation. I was unable to reproduce the errors in that scene until I tried it on the benchmark and they showed up again, but since the bmark is small file I was able to fairly quickly run a lot of tests and narrow down the results.

[tarn]

I was wondering if there is any update on this topic. Vlado said this has been fixed in the newest builds, and I am currently having problems running an animation on our network which has a mix of CPU types. I am also finding that frames render slightly differently on different machines, and this causes noise in the animation. Are any of these newer builds available yet, or will they be soon?
Thanks,
Tim

[vlado]

There might be other reasons for differences between frames, not necessarily related to the CPU types or machine configuration at all. Can you show a few frames from your animation, you can either post them here or email them to vray@chaosgroup.com?

Best regards,
Vlado

[dlparisi]

I haven't seen any problems since updating to SP2.

...what Vlado said above.

[tarn]

We have 3 different computer types in our office which I am using with Backburner to render frames of an animation:

(1) Intel Core 2 Duo E8500 @ 3.16GHz Family/Model/Stepping: 6.7.6
(2) Intel Core 2 Duo E6750 @ 2.66GHz Family/Model/Stepping: 6.F.B
(3) Intel Core 2 Duo E6600 @ 2.40GHz Family/Model/Stepping: 6.F.6

I am running an aerial fly over animation with 1500 frames, using a precalculated IR-Map. If I send a single test frame to either computer (1) or (2) above, the results are identical. If I use the computer type (3) above, there are slight diffences in the noise. Also, if I send the same frame again to computer (3), it will also be slightly different from the same frame rendered previously on that computer. The differences between frames appear to be very small, but when run as an animation, they cause pixel sized flickering.

I have uploaded an example here:

http://members.lycos.co.uk/tarn878/D...reneces_02.pdf

If you click between the two images in the PDF, you should see the difference. I have blown the images up 200% in this sample, to make it easier to see the difference.

All the computers have VRay 1.50SP2 installed and VIZ 2007. Also, I am using VrayToon, in case that makes a difference.

Thanks,
Tim

[burtonr]

sorry if somone has already addressed this (didnt read all of the threads) but we have had some different frames render out if there are machines runing a different version or SP of Vray.

[tarn]

I have done some more tests. Now I have tried rendering the same frame on the same machine twice, but the results are not identical. I have turned off VrayToon and set AA to Fixed 1 with no filtering. Is this to be expected, that the same frame, rendered twice on the same machine is not identical. The differences between the two images are very small, but it is enough to cause some pixel sized flimmering when seen as an animation.
Tim

[RErender]

Could be AA?

[tarn]

Could be AA?

Even if I had AA set to Fixed, 1 sample with filtering off?
Tim