[brodie_geers]

For some reason that caused crashes on both machines within about 60 frames. I'm going to try again. In the meantime I'm transferring all 60+ gb of MXS files to the respective hard drives to completely avoid any network issues.

-Brodie

[brodie_geers]

I keep having issues which have caused me to restart the progressive animation script several times. As a result I have several copies of the MXI's at lower frames while the script hasn't been able to render long enough to get me a rendering of the last 300 frames or so.

Is there a way to tell the script to start at a certain MXS file (ie. WillitsAnim1129.mxs) rather than always starting at the beginning (ie. WillitsAnim0000.mxs)?

-Brodie

[Mihai]

I guess easiest would be to move the rendered mxs files out of the folder....you could add an if statement in the render function to only start if file number is higher than a certain frame, but why not keep it simple? What are the issues you are getting? Is Maxwell crashing after a number of rendered frames? Can you check if the memory consumption of Maxwell keeps increasing with every rendered frame?

[brodie_geers]

My issues are everywhere I turn. I mentioned the network hangup above which has turned me to trying to render out the frames on each computer individual without using the network. Now I seem to have some sort of licensing issue so far as I can tell. Whenever I start the progressive animation script on one computer, Maxwell crashes on the other computer. Both computers have the same license.txt file which contains 2 licenses and 1 node and both machines recognize the 2 licenses +1 node if I check the Help info in Maxwell Render. However it seems to be acting like there's only 1 license. At least that's my best guess right now. If you have another idea, I'm all ears.

As for moving the mxs files around, I'd probably have to move around either 20gb or 40gb of files which would take some time. Plus, ideally I'd be able to let it start at, say frame 1100, let it run to the end and then go back to frame 0000. I guess I could start it and then paste the 40gb of mxs files back into the folder but with all the issues I'm having I'm afraid something would go wrong and it would jump back down to 0000 rather than continuing through the last frame or perhaps something else unforeseen.

-Brodie

[Mihai]

Probably time contraints don't give you too many options but I'd try first running say 100 frames as a test in network and set the network to debug verbosity (From File>preferences, in node, monitor, manager). It would give us a bit more clues why the Monitor disconnected in the image you posted. Maybe some other app interferes? Maybe Windows goes into hibernation mode or some other power settings? Maybe it disconnects after a certain number of files have been rendered, you could try and find that out by rendering the same sequence but set a higher initial SL, so it can render through the night, a smaller number of files than before.
But if there is no time for that now, I would try to render this sequence in smaller batches with the progressive animation script. I'm not sure how you are doing it but I guess you render say 500 frames on one machine, and 500 on the other? They aren't both rendering the same mxs at the same time, just on different machines?

[brodie_geers]

Well my current plan and setup is this...

I have all the MXS files copied to the local drive of both machines for the machines to pull them from. Likewise the output goes to the local drive so they're saving MXI's and .png's. I'm trying to get each machine to render the entire animation to the highest SL possible. Then I plan to put all the MXI's on a network location and use the Merge MXI Sequence option in Maxwell Render to merge everything together into one MXI sequence.

From there I'm not sure how best to get image files from those MXI's. If I give myself enough time, my plan was to maybe do a batch render to get them all to the same SL and export an image at the same time. That may be a bit idealist though as I'm not sure I'll have the time to do that at the end. Is there another way to just get an image sequence from a group of MXI's without having to do any rendering?

Does all that sound like it should work?

-Brodie

[Mihai]

Not sure why you want to go through the extra step of later merging all the MXI files from both machines, why not just render half the frames on one and half on the other?
You can save images from a bunch of mxi files using File>MXI Batch Processing from Maxwell.

[brodie_geers]

Thanks, just found the batch processing option. I don't really want to split up the rendering like that for a couple reasons.

1. The other computer is a server 5 floors down guarded by our IT department. Their nice folks but having them periodically copy 700 .png's through our network so I can update my after effects file periodically may stir them up and I need them very friendly.

2. More importantly, the server machine is a bit slower than my machine which means splitting it up becomes problematic. Do I try and determine how many frames go to each machine? And do I let them each render as long as possible which means at some point in the animation the quality goes from say SL 10 to SL 8? Or do I fret over finding a compromise SL? It's all a lot of effort and brain work that I don't have much time for.

-Brodie

[Mihai]

I'm going to check if anybody has heard of this problem of one machine crashing soon after another one starts but just wondering, do you have full control over that machine? Can you restart it, access it to restart a node?

About the SL....I'd just send one frame to the server and see how long it takes to reach an acceptable SL for that frame, say SL 17-18. Then you have an idea how many frames you can send to that machine for a given amount of time, just let them render each one until it reaches that SL. As long as the lighting complexity doesn't increase dramatically (say camera moving from outdoor to indoor)....
Render all frames to the same SL.

[brodie_geers]

That's part of the issue. The determining factor isn't really what SL I think is best, it's more a matter of what's the highest SL I can get to given my Tuesday deadline. I knew up front that I wouldn't be looking at SL17-18. I'd be content with 10ish which is still pretty noisy but for this project it'll be ok. This is just for a client to see what the building looks like rather than some sort of marketing animation and at any rate it's a step up from the animations they've gotten from me in the past which are simple SketchUp animations which have been acceptable. I'm hoping to clear up some of these bugs so I can build on this experience in the future on more crucial projects. Even had I been able to determine that I have X hours to render and divided the frames up according to the speed of each computer, I still would have been missing the bug factor which has significantly cut down the time each machine has been up and rendering and either machine has been down for different times because of my issues.

As for this machine, I don't have full control over it. It's a server located in the IT department. Fortunately it has very few functions but they spent a lot of money on it which makes it perfect as a render node...or rather it would be perfect if I never ran into any bugs/crashes/network issues/etc. Although I don't have full access, nor do I even have remote login capability, the IT guys are pretty accommodating and restart the node all the time for me and have been really good throughout this whole animation ordeal. The biggest issue with not having access is well illustrated by what happened tonight. I left work at 5pm, did some stuff and just got in at 10pm and remotely logged into my work computer. The progressive animation script is still running fine however, the rendering I was running on the server via Monitor had stopped working again. In fact it stopped at 5:06pm, six minutes after I'd left work. /cry And I can't restart the node and no one is there to restart it for me.

I found a batch script that is supposed to be able to restart a program that has crashed or hung though. I'm going to see if they can combine that with some sort of scheduler to run the script every 10 minutes or so. If it works how I hope it will, it won't fix the crashes, but at least there will be much less downtime. My rendering today would have started back up by 5:16pm instead of sitting there doing nothing until 9am tomorrow when IT gets around to restarting the node.

-Brodie