Maxwell Render | Official public forum

Dan, I remember this coming up before, I am searching the forum to find an answer. Hopefully someone here that has more experience can give you the answer.

Dan,

You are on Mac right?

try this link:

http://www.maxwellrender.com/forum/view ... nger+icons

dann_stubbs wrote:tyrone - how do i see the SL level when using the network rendering?

As far as I know you can't. Seems to be disabled. Each node puts its current image in whatever location you specify in the render options (don't make this a network drive or each node is constantly overwriting the last one). So you can see a current preview of each particular node by opening that file, but you can't preview the merged file. I'm not sure if this is something that can be changed in future versions. It takes a long time to send all those MXIs across the network and merge them (at least 15 minutes on my network, sometimes an hour), so I assume it would be impractical to do that in real time during the render.

dann_stubbs wrote:i did find another thread here and others have the same question as to seeing the SL level - and how to figure out the SL level - i.e. do all network render computers render to the set SL level THEN combine to get a higher SL?

or do they all go to a lower SL when then combined makes the SL level that is entered.

The way it currently works is that if you do a network rendering and specify SL 15, all machines will render to SL 15 and then combine to form a higher sampling level.

As far as predicting, I honestly believe there will be a way to do this in the short future. I need to directly lobby Kabe for access to the equations he used to create his calculator, but it seems there should be a straight forward way to do this. If two 5 minute renderings done on the same machine combine to create a 10 minute rendering, I'm sure we can figure out a way to combine time and benchmark to give a value of something, combine all those values for different machines and back out a sampling level. Easy right?

Hi Dann,

this worked in v1, but doesnt in v1.1. seems to be a minor bug.
Do you offer maxwellrender on your farm?

cheers
stefan

Hi dann,

i have to test more with v1.1.
i am on the process to change from 1.0 to v1.1(still have severe problems with the new v11 c4d plugin though)

sorry i cant help with netrender yet, but later when i have tested more this week.

i think the project folder is save to have on a network drive, at least i have it so, but the software itself should be located at each PC's - C: drive.
but double check it. i am not 100% sure.

cheers
stefan

dann_stubbs wrote:

lllab wrote: i do have another concern - i read a post somewhere on this forum that said NOT to use a network drive for the files as each client will overwrite the last?

That's only true for the in progress image that each node saves out every couple minutes. MXIs are not saved until you stop the job, and the individual node MXIs are always saved on the local node in the Maxwell directory.

I have had great results with net rendering so far, but the lack of feedback is a problem, as is crash recovery.

dann_stubbs wrote:i do have another concern - i read a post somewhere on this forum that said NOT to use a network drive for the files as each client will overwrite the last?

so anybody know if that is true or not? that is crazy if so - the instructions on setting up maxwell are fine - but from there it is just guessing... that is why i am not sure if i am seeing a combined image or am i just seeing the result of whatever last single computer wrote the file...

Dann,

The person who wrote that was referring only to the output image file, such as the JPEG. If you tell the network rendering to put the output image on a network drive, all the render nodes will overwrite each other's output files each time they update on the network with a new JPEG. The MXI they are each working on is stored on each render node's C drive in the Maxwell folder. Something called "cooperative.mxi." That's why when the job is over you see "Sending MXIs" and "Merging MXIs."

What the person in that thread was trying to say was that if you specify a spot on the C drive as a location for the output file, you can go back afterwards to each render node and see what image it created by looking on its C drive in the spot you specified. But as far as the MXI goes, if you specify, say, "L:\Maxwell\test.mxi" for a location, it'll wait until the end, send all the MXIs back to each other and combine them as they go, and then it'll write the final combined MXI to the network location.

And, just a little fluke I noticed, the last computer to do the combining (not always the manager) that writes the MXI to the network, there will be a copy of the final MXI on that computer also, named either "cooperative" or "merge."

Hope that clears everything up for you.

Chris

dann_stubbs wrote: so then where is the final combined image? technically is there one?

The combined MXI will be in the folder you specified unless there is a problem when the computer merges all the MXI files together. You will know there was a problem because the job won't say "success" it will probably still say "rendering".

dann_stubbs wrote: or do i then have to open that final saved mxi file use that to generate one? (is that also a file i can use to resume the rendering?)

Sometimes you do, I've had instances where the final image file created from the MXI's was messed up. I'd say that until cooperative rendering is working flawlessly you’re probably going to have to do this.

dann_stubbs wrote: so to stop a job (new question) do you press stop job on the top line or on each render client one at a time? one time i clicked the top and then wasn't sure since it didnt' respond right away (i think it would have if i waited longer) so i clicked a couple of the render clients themesleves and they did also say stopping - but the result image was corrupted looking (all jumbled image) so i'm not sure if that messed with the "right" way to stop the job.

To stop a job you highlight the main branch of the job and press stop, depending on the speed of the computers it could take several seconds before you see a response. If the job doesn’t stop that means that one of the slave machines had an error, sometime restarting the service on the machine will fix it but most of the time it's an unrecoverable error and you won't be able to finish automatically merging the MXI's together. You will in this case have to merge them together manually which could require a lot of time depending on how many machines you are using.