[daros]

x Mihai.
Because we don't use the MXI files. We split the renderings in regions and merge simlply RGB data.
This is not the ideal solution for Maxwell since we have problems with emitters. But we can split a rendering across 200 cpu's wit our system and merge the information in less as 5 seconds. We don't use Maxwell Server.
Now we are implementing bradcasting and multicasting technology to distribute geometry and texture data to the nodes so we are able to distribute 4.000.000 polygons on 100 nodes in less as one minute instead of 3 minutes.

David Rossmann

[Hervé]

well well... pretty impressive...

[Mihai]

I suppose then all the renderings you do through your system will have to go to 18 sl, no more, no less?

What if I have a 3000x3000 image that isn't clean after 18 levels and I've already spent the credits on it? How can i resume such a render through your service?

3000x3000= 28 credits= 336 euros.

Or the opposite, if I have a render which is clean after 12 sl?

[daros]

Hi Mihai,
yes i understand wat you mean.
At moment we simply arrive to 18 samples and it's no way to stop the rendering before or to continue the rendering after sample 18.
But with version 1.0 of maxwell we will implement a progressive update on RemoteMatador so you will be able to stop the render at any point and pay only for the used cpu time.
It would be easy to implement it already now but we want to wait version 1.0 of Maxwell.

[Mihai]

Excellent, thanks for the replies daros. I think your system will be of great use for Maxwell users!

[andronikos916]

Very promissing stuff Daros!

cy,
Andronikos