Maxwell Render | Official public forum

If I have 3 machines, cooperative rendering, and they are calculating the sampling level 3, does this means that the overall image is getting to sampling level 9 (3x3=9)?

During cooperative rendering a jpg image is being saved alternatively from one of the 3 machines. Is this jpg image a merged image, or just the partial result of each of the 3 machines alternatively?

In the cooperative rendering interface I can read a column xRes, then another yRes, and after that there is another, which I cannot read, because it is out of the window range. I wonder what is this last column, and why I cannot see it. I tried in 3 diferent machines monitor resolution at 1280 wide.

In the lower box in the Cooperative rendering window, there are lots of specs refering to the servers in the network Ram Processor etc , but it is all empty. Is this a problem with my systems or this will be added in the future patch?

I do not understand the command -th:N in relationship with hiperthreating processors. I have some HT processors in a network and I do not know if the right comand is -th:0 or -th:1. Any further explanation will be very appreciated!


Thanks in advance

Ernesto

Ernesto wrote:If I have 3 machines, cooperative rendering, and they are calculating the sampling level 3, does this means that the overall image is getting to sampling level 9 (3x3=9)?

This is a good question and only one that the NL team is going to be able to answer.

Ernesto wrote: During cooperative rendering a jpg image is being saved alternatively from one of the 3 machines. Is this jpg image a merged image, or just the partial result of each of the 3 machines alternatively?

Different people are getting different results but in my case this is just a partial result from one of the render nodes that is working on the job. My cooperative rendering setup will also write a MXI file into the same folder once the whole process is done. This is supposed to be a merged MXI from all the nodes but it is also just a partial result from one of the render nodes. In order to get the merged MXI I have to manually combine all of the single MXI from each of the render nodes. Some people are not having this problem, but Mike said today that he is experiencing the same thing on his system and will have NL take a serious look at it.

Ernesto wrote: In the lower box in the Cooperative rendering window, there are lots of specs refering to the servers in the network Ram Processor etc , but it is all empty. Is this a problem with my systems or this will be added in the future patch?

None of those work on my system either so I assume it's something that will be added.

hi .. i have doing a lot of test in cooperative rendering, it "works" but i have seen a lot of bugs

sorry it is in spanish, i will try to translate this later, just feeling a bit angry right now because all the bugs this "final release" has.

http://www.maxwellrender.com/forum/view ... 066#147066

But i can tell:

- It does render in all nodes and merges the image automatically
- Only first image works, subsecuent images are lost
- Temporary files creates a lot of problems
- Some buttons in the monitor panel doesnt work.
- Cooperative rendering does not respect the time specified by user.
- Still testing to see if there is any advantage by doing this cooperative rendering stuff, so far it seems to be that its tha same rendering an imagen with 1 core that with 4 cores.

I am still testing

- Still testing to see if there is any advantage by doing this cooperative rendering stuff, so far it seems to be that its tha same rendering an imagen with 1 core that with 4 cores.

I had this impression...

Hi guys, same problem here.

I tried cooperative render 780 Kb mxs file on 4 machinces successfuly, but whenever the file goes beyond 1 MB cooperarive render stuck (state:receiving scene, and stuck). 2 machine have 1.5 GB and other 2 BG.

Is this bug?

Thank You.

Yadi

right now i am still doing some tests, i am rendering a 53mb mxs file, i could do the cooperative rendering and i am checking now if it makes any difference between 1 core and 4 cores. This thing has a LOT of bugs. Will be posting a couple of imagenes as soon as i finish with this tests. I i ever manage this thing to work.

After hours of testing here are some images:

First image made with cooperative rendering with 4 cores second image just 1 core same setting for both images.

rendered for 30 minutes
averange SL 6 on each core
mxs file 53mb
Num Meshes: 657
Num Triangles: 1479968
Num Vertexes: 772137
Num Normals: 772769





Im sorry but i cant see ANY benefit from using cooperative rendering.

http://www.maxwellrender.com/forum/view ... 3&start=30

I made a list of the bugs a have found so far, i am tired right now so i am not going to waste more of my time with this thing. I would really like to know if they test this things before giving them to their final users, its incredible the kind of bugs you have to deal with in order to get an image out of a "finished software". Not bad for a "v1" Congrats ... keep the professional work Next Limit !!!

If I understand it right, the SL has something to do with the bit/color resolution.

If every pixel of the rendered image is statistically hit once by a photon you get SL 1. If every pixel is hit twice you have SL 2. Then if every pixel is hit 4 times you have SL 3, because the time just doubles to reach next sampling level.

This would mean, that with every new sampling level you get 0.5 bits of better picture to noise ratio, because image is correlated and noise is not, so image gains 1 bit and noise gains 0.5bits and 1-0.5=0.5 ??? ???

So SL 20 means you actually have a resulting color depth of 10 bits ???

Any ideas ?!?

No i have not (well, with the firsts test i made i did that, but i want this thing to do what it is suppused to do) ... as i am not supused to do that, the cooperative renderer has to merge the images itself, imagine having 10 or 15 images rendered in 8 nodes

I would really like to this this working the way backburner does, just sending you rendering from your host applicatin and thats it, no uploading, no manual adding jpbs, no manual merging, no temporary manual files deletion, not having to put IPs for every machine every time one core fails, not having to stop everithing because all the cores are not rendering, not having to check if the netrendering is deleting your files even when they have different names, that sort of things that we got on a "FINAL" version.

Just really tired today with this tests just to see things are not working AGAIN.

see here for the -hd option..

http://www.maxwellrender.com/announcement.htm

it's got a star and isn't ready yet.

* * *

deadalvs

I agree with all of you, this is a terrible showing for a V1, cooperative rendering was supposed to break all time barriers but with all the bugs and manual stuff you have to do it's almost not worth the time. I realize NL ran out of time but this is one of Maxwells most important features if not the most important so I really expected it to work properly. The next release needs to clean all of this mess up, no more manual deletion or mergers, it needs to be able to recover if a node goes down, and it should clean up all of the loose files on its own. I'm very disappointed.

I completely agree with you maxer, for me this is one, if not the most, important thing i need maxwell to do, i cant use maxwell if i have to wait 8, 10 or 15 hours to get just 1 image !!! or having to do averithing by hand because i cant trust in a distributed rendering system that will delete all your work

Thank You all for all your answers (as well as questions).
Yours

Ernesto