[JCAddy]

I can't see any of the images?

what the heck?

[Maxer]

Sorry about the problem with the images our bandwidth isn't up to the challenge.

[tom]

One thing I didn't post was that I let the same scene render over the weekend on a dual core dual processor machine for 63 hours, it was basically the same as the 40 node rendering.

This is quite perceptional, in fact the equivalent is 40x13 h when the rendernodes are identical.

[daros]

in every rendering engine the difference between 1 sample/pixel and 5 samples/pixel is more visible as the difference between 32samples/pixel and 256samples/pixel.
I my opinion Maxer examples shows that the cooperative system performance is efficient and almost linear.
The quality difference between 30 nodes and 40 nodes is exactly how i think it should be.

[Frances]

One thing I didn't post was that I let the same scene render over the weekend on a dual core dual processor machine for 63 hours, it was basically the same as the 40 node rendering.

This is quite perceptional, in fact the equivalent is 40x13 h when the rendernodes are identical.

Math weenie here. I don't think the 10-node render @ 13 hours is comparable to a 1-node render at 130 hours, if a 1-node render @ 63 hours looked closer to the 40 node rendering.

Devin, can you post the 63 hour render?

[daros]

I don't think the 63 hours rendering on a single machine (even on a dual 285)
looks close to the 40 node x 10 hours rendering.

that is a 20 hours rendering at 1500x750 on a dual 265.
http://www.animax.it/nextlimit/test3a.jpg

[Maxer]

This is the 63 hour rendering:

ftp://kirkseyviz@ftp.kirksey.com/maxwell/63hr.jpg

[Frances]

This is the 63 hour rendering:

ftp://kirkseyviz@ftp.kirksey.com/maxwell/63hr.jpg

I was hoping they'd be the same camera and materials so I could scrub between them. Ah well.

But the 63-hour render seems to have less noise on the ceiling than the (40x13) one.

[Mihai]

Too many users......why do you post them via ftp? There's always imageshack.....

[Maxer]

I'm in the process of re-hosting the images now, I'll let you know when I'm done.

[Maxer]

Ok I've changed the hosting of the images, hope that solves the problems.

[pluMmet]

I was going to make a scene and render out an objectID.jpg but if possible this would do well. I don't know Maxer if you have Neat Image (I didn't know it cost money and i don't have it) but if you do you could do a 4 or 5 min render (just enought to get it going) and it will make the objectID.jpg then you could run neat image on just certin areas. It seems to me that none of the sharpness would be lost if it was done this way

just an idea if not I'll continue making my scene for a test. I'll have to pass it on to someone who has Neat Image

[Maxer]

Ok, say that again pluMmet I don't follow?

[Ernesto]

This test is great! to understand that we are still far from the desired results in Maxwell Render. I know it is the BEST rendering engine, but I would love getting MORE control, instead of failing my deadlines forever.

What I see as a solution (from outside NL) is a light spot diameter control.
Now Maxwell calculates each lightray, as one pixel, but this causes some scenes run forever. What I propose is a wider raylight, which could produce a lightspot higher than a single pixel. In this way we could find out our deadline rendering time.

Ernesto

[jo]

One thing I didn't post was that I let the same scene render over the weekend on a dual core dual processor machine for 63 hours, it was basically the same as the 40 node rendering.

Really?


I think that merging is not so efficient (I have done some tests)
And also that is not the same than resuming!!!
I'm waiting for new realase of Maxwell!!!
When for a 1.0.1?

Ciao, Jo