[aitraaz]

Is there anything else we should complain about besides the preview images? Was the font OK in the email?

Well, now that you mention it i do find it irritating getting messages with verdana family typefaces, and 12 point characters is hell on my eyes...

[JDHill]

Ernest,

Whereas kirkt's test averaged randomly generated noise, in Maxwell, combining the knowledge contained across multiple independent .mxi of the same scene represents the sum total, not average, of their solutions. Does this make more sense?

~JD

[ricardo]

Does anyone know if rendering the same image multiple times on the same machine produce a random seed for the noise in the image or is it machine based?

No. The multinode pseudo advantage comes from some machine specific random seed.

Some time ago I was bothered by the fact that the same mxs file would result in different SL values for machines using the same OS/Porcessor combination. It's not about SL/time, but SL steps. One would pass the 12 benchmark with 12.34, the other would have 12.51. I repeated the renderporcess in the same machine to make sure it repeated the result.

Asked a dozen times, never got an answer. The result was if you compared images from the two systems they would have noticeable differences in pixel values.

By Maya's example (on THAT email) rendering across different systems would have wasted of 15% of the full rendering time. What this means: if you render in 20h00m with one machine, you would need a combined 22h30m spread across nine sytems to reach the same result.

Bottom line is: if you have plenty of images to deliver, do not spread every single frame across the network, have each node doing a single image. This will save you 15% of the total time.

There is no magic.

Ricardo

[b-kandor]

Bottom line is: if you have plenty of images to deliver, do not spread every single frame across the network, have each node doing a single image. This will save you 15% of the total time.

Not exactly as the speed of the machines used is different.

So: (2.2ghz x 10) * 2 = 44 ghz hours
and: (2ghz x 2.5) * 9 = 45 ghz hours

So 44/45 = 97.7% not quite 15% of the total time. But approx. 2.3%.

So it is true - there is no magic.

[Frances]

Because test scenes are not ment to be sexy, but to test as many features as possible, in certain conditions.

Is there anything else we should complain about besides the preview images? Was the font OK in the email? Was the logo too big/not big enough?

Just in case you were referring to my crit of the image. That's just what it was - a criticism. Why TF should I complain about an image? Are they incapable of producing better images to showcase their product's features? Or just not interested? Or [your reason here]?

[ivox3]

Is there anything else we should complain about besides the preview images? Was the font OK in the email? Was the logo too big/not big enough?

.....nope .....I"m good.


....shoot, I'm happy to get any email from Maya. ......even if it is a bit tardy .....

[WestCAD]

No email here, but hey I am still rendering occasionally with beta when I can afford to dedicate 20-30 hours to my 1 cpu for limited results. Could be sailing into unchartered waters.

[WestCAD]

I would be happy if they just fixed the emitter values and maybe if not asking too much sunlight and glass.

[Kabe]

Nope, no volumetrics in RC6 afaik. I've seen that great cloud photo before.

@ Mihai:
Cállate. It's ok to criticise an image if the first thing we hear from NL for
10 days is a questionable image with a questionable announcement.

Kabe

[pluMmet]

Sucks to be in the same boat as Mihai but the email image was fine for me. I feel bad for whoever made it looking at this thread and seeing everyone bash it. It's clearly an image desgined to show features and not completly lacking in artistic value. Of course it could have looked better but it's set up that way for a reason and whoever did it did a good job for the purpose it was designed for.

I hope I don't need to agree with Mihai again

[Fernando Tella]

I think that a technical test image has to be that, just technical and there's no need to be sexy or to showcase all the possible features of MWR, just the one you are showcasing. It has what is needed: soft light, multibounced, clear reflexions and refractions of dielectrics in indirect light.
If you mix other features in a test, the test might be corrupted by that other feature.

Apart from this: I would like to know how mxi storages the light info. It's clear it stores the effect of each light separatedly so I guess the mixing in the coop render doesn't have to do with pixel or noise averaging but with light rays adding. But how!? Or maybe is by adding energy information per pixel if mxi makes the translation between energy and RGB. But how is that the calculations are not duplicated when rendering the same images and the results can be added? That's the mysterious part and the magic of the mxi cause the cooperative render is not controled by any overall process which tells every CPU which part calculate.
I guess that there is a sampling level from which the coop render stops being useful to the final image (when the mixing reaches the final light solution).

Oscar? Juan? Hello?

[Fernando Tella]

Yeah but those are images made to show the product; to sell it, not to test it. That's a great difference.

[Fernando Tella]

I guess they choose scenes that are hard to solve. That makes it easier to see the differences. Would be very clarifying to see the 9 mixed images too.

[WestCAD]

Just received email from Maya this evening Melb time. I have a question hopefully someone may answer.

I have a total of 3 computers...pentium 3.4, AMD 2800 and 3.2 Notebook.
Could I utilise all 3 of them to co-operative render and still do cad work on one. (I only have 1 Maxwell License)

[ricardo]

Hi,

This is what I got from Maya:

What’s interesting is the fact that it has taken 10 hours to reach the
level of image (a), while on the other hand, it has taken only 2 hours and
30 minutes to get to the same point with image (b).

The images where missing and I didn't have any info on the machine specs. Is there another email? I'd like to see it.

Thanks.

Ricardo