Cooprative rendering

[Realmind]

What is cooprative rendering in maxwell and how can it increase rendering speed?

[leoA4D]

Cooperative rendering divides a render job among more than one computer and reunites the pieces in one MWR mxi file. The more computers, the faster the rendering. Assuming NL gets the bulk of the bugs solved, multi-computer setups or render farms are, at this time and into the foreseeable future, the only way to make Maxwell a viable production tool for interior archviz projects.

[cjlopez4]

I'm trying to set up my machine as the render manager and the one other as the slave render. Maxwell doc's states to input a command in the "command shell" Where is the command shell?

[cjlopez4]

I'm trying to set up my machine as the render manager and the one other as the slave render. Maxwell doc's states to input a command in the "command shell" Where is the command shell?

 

 

nevermind, it's the winxp command shell, duh

[Realmind]

Hi.Why is maxwell render so slooooooooooooow!?

I think that it can render very nice but the problem is time!

[Greg McDowell]

it's not slow... it's precise... it just takes time to do all the math...

 

i'm playing around with the demo right now and i'm fascinated by it... it seems that the only real learning curve (considering I already know a thing or two about photography) is in creating new materials (too many years of Max)

 

i think the real key to maxwell is understanding photography... could be wrong but it seems right...

[Devin Johnston]

Everything in Maxwell is easy except for the materials, hopefully the new material database that NL is collecting will fix that problem to.

[AJLynn]

...and it's slow. This is unavoidable.

[Realmind]

it's not slow... it's precise... it just takes time to do all the math...

 

i'm playing around with the demo right now and i'm fascinated by it... it seems that the only real learning curve (considering I already know a thing or two about photography) is in creating new materials (too many years of Max)

 

i think the real key to maxwell is understanding photography... could be wrong but it seems right...

 

Have you ever rendered any clear image by maxwell?

I have done!

It has taken about 90 hour to finish!!!!!!!!!!?what is this?

[Realmind]

Have you ever rendered any clear image by maxwell?How long has it taken to finish?

I have rendered just one time and I think that it's enough for me for ever to put it away!

How can one afford 90 hours for a picture!!!!!!!!?

[Maximus3D]

Ok, so that makes me wonder why you bought it in the first place when you knew it works completly differently from other renderengines. Things do take time with Maxwell and it will remain like that, not much you can do about it when the maths it's based on are of such a complex nature that you cannot get realtime feedback or common GI renderer speed. That's not how it works.

 

/ Max

[manta]

Ok, so that makes me wonder why you bought it in the first place when you knew it works completly differently from other renderengines. Things do take time with Maxwell and it will remain like that, not much you can do about it when the maths it's based on are of such a complex nature that you cannot get realtime feedback or common GI renderer speed. That's not how it works.

 

/ Max

 

What makes you think he bought it ? he's probably using the demo...

[AdamT]

I disagree that everything is easy except the materials. Network rendering is also godawful, and really, at least as a Cinema user, the workflow just sucks. Hopefully we'll get a decent plugin one year.

[Devin Johnston]

Cooperative rendering once it's is set up isn't that bad if you can find a way to deal with all of the built in inadequacies. I find the instability of the plugin and the complexity of creating some materials to be the biggest problems, but cooperative rendering could definitely be made better.

[Adehus]

How can one afford 90 hours for a picture!!!!!!!!?

 

I'm still waiting for a good answer to this question as well. For the Maxwell faithful, it's apparently worth it. And then the other 99.9% of the CG community just thinks it's absurd!

[Greg McDowell]

First, from what I gather from the Maxwell boards if you're taking 90 hours to "clear" a scene something is likely "wrong" with your model... or you're rendering at an unrealistically high resolution.

 

Second, if the largest amount of time spent during a traditional rendering sequence is all of the back-and-forth needed to get a rendering "just so" (and this may be more applicable to those of us who also do other things besides render) were to be removed and you were able to "trust" the output implicitly, how much more time would you have at your disposal? My impression is that you would have much more time to design knowing that when you get in the next morning (or over a typical weekend) that you will have one amazing result.

 

Third, every situation is different and as long as we're not avoiding new technology because it's not what we're used to then the reasoning behind the decision made is justifiable... in other words, if it ain't right for you, don't use it.

[Adehus]

Honestly, Greg, you're saying all of the things that the Maxwell moderators have been saying for quite some time... yeah, setup is easier (or was before the new material system.) But not easy enough to compensate for the render times. Trust me on that. 90+ hour render times are, protests to the contrary, very very real- even on top of the line machines. Maxer did cooperative render tests on 10 machines for 10 hours. Still not noise free. It does depend *somewhat* on the situation, but interiors in particular can take 90 hrs or longer- easily.

 

BTW, the only resolution that's unrealistic is one that's larger than the client requests... just a little reality check.

 

And I can certainly say that I'm not avoiding new technology because of it's unfamiliarity- I was among the first to purchase Maxwell, and had only high hopes for it. Now, a year later, I wish I hadn't spent the money. Maybe one day it'll be better and/or machines will be fast enough to make it practical, but I can absolutely guarantee that that day is quite far off. Right now, it's primarily a hobbyist's toy.

 

One last point- you will not get unbiased information on the Maxwell forums. They've purged a large number of respected users (and me lol) who were vocal about Maxwell's inadequacies in comparison to NL's promises. All who remain are now cowed into submission. Almost anyone who peruses this forum will willingly admit that they were lied to about quite a few significant aspects of the app... during pre-sale, they were marketing based upon the promise that Maxwell would be sped up by a factor of 10. When that didn't happen... well, can you blame people for being unhappy?

 

And... geez... talk to the resellers sometime... (!)

[Devin Johnston]

Adehus speaks the truth; Maxwell is the slowest engine out there. Exterior scenes can be rendered to relative clarity in less than 24 hours, but virtually any interior scene could easily take 90+ hours to render regardless of how you set it up. I'm talking about a print rez image that is bigger than a postage stamp, most of the render times you see on the forum are for screen rez images and even those can be over 24 hours. In my opinion Maxwell is virtually useless to most people doing arch viz unless they only need screen sized images or have a big render farm.

[tecton3d]

since the topic of this thread is coop rendering... what would be an ideal setup for making a small rendering cluser/farm for maxwell?

 

I currently have a 4 system (and perhaps expanding to 7) dual xeon farm, each @ 3.2, a dual opteron workstation, and am still wondering if MW is worth a try.

 

cheers

[Greg McDowell]

I think the thing that's making the difference to me is that I'm not a professional renderer (or should I say Artist here?)... I'm an Architect who spends most of his time working on other things and while I know that, with experience, you can generate results that nearly as good (possibly better if you factor in the "noise") I don't have the time or energy for that right now.

 

We do have some junior architects fresh out of college who are doing a lot of our renderings but the quality is almost cartoonish and they spend most of their time in PhotoShop "painting" in textures, reflections, entourage, etc. I know that if they were to put more effort into the rendering application that we would, in the long run have better output as a result. But as it stands I would be embarrassed to place their work along side the work I've seen in my short time browsing this forum.

 

We've even outsourced some of this work in the past to a local company just down the street and, while the results were nice, the experience working with them, the process was terrible... especially for the price.

 

So where does that leave us? It leaves me wanting a package that allows me to be the dumb Architect assigning materials (hopefully that someone else made... that editor looks a little complicated just now), placing "real" light fixtures, clicking the "Render" button and getting consistently good results. And if I have to plan ahead to save the week it may take to render the images we need then I think I can handle that... provided I can trust the software to do what is needed.

 

I've been thinking about an analogy (I'm big on these for some reason) using Excel. At first you might double check the math in your spreadsheet because you don't trust it... because that's the way you've always done it... because "computers make mistakes" (or so my old boss told me about this very issue). But after a time you begin to trust that the results you get are going to be correct because that's just the way it works... I'm thinking that M-R might not be too dissimilar.

 

So, since I don't have time to learn how to be an Artist with these tools (it's hard enough to pretend to be one as an Architect) the prospect of getting the results I see in M-R with little to no effort on my part (we're using Revit to create the model) is very appealing. And, as I theorized before, if we can spend more of our time solving design problems and less of our time tweaking the rendering and still get darn near photo-real results I'm a happy camper!

 

And, FWIW, there's a free application I found through the M-R forum that estimates the amount of time you might spend on a rendering at a given Benchmark, for a given resolution and sampling level and, so far at least, has proven to be fairly accurate. That should allow us to plan ahead for the time we need to get what we need output.

[manta]

Well as long as you're happy, that's all that matters, I personally find it anything but easy, I always get errors and have no idea what I'm doing wrong, sometimes it renders sometimes it doesn't, I'll give it a try again in a couple years...

[Devin Johnston]

I think your going to find that there is no easy way to get good looking images no matter how simple the interface is. Maxwell at one time was relatively simple to operate when compared to programs like Vray and Final Render, but that time has definitely passed since the new material editor hit the scene. That's not to say that it was a one button render engine, you still needed to know what you were doing to make it work. My suggestion would be to take one of those interns or hire someone who want's to do this stuff full time. Your not going to get to where you want to be by sticking your big toe in the water, your going to have to take the plunge for it to make a difference.

[Guest imanobody]

Greg, not trying to discourage you or anything like that, but most of those fancy images in the maxwell galary were created by vary talented artist that could do just as good images in other renderer apps. If you really want to save your time just spend the money and hire a talented renderer - I'm sure they could do in a few hours what would take Maxwell days to do. And want happens when you need an animation? Maxwell is totally useless for that, so you'll have to use a different renderer and then people will be asking why the two don't look the same. Just my opinion.

[Adehus]

Greg-

 

Same thoughts as the previous two posters... Maxwell is no panacea and I wouldn't look at it as if it's a beginners tool or a shortcut to good imagery. You still have to have a great model and excellent materials and texture map creation/mapping to get high quality imagery out of it. All Maxwell does is help with the lighting (and complicate the material creation). I can't imagine how Maxwell would cover for the deficiencies you describe in your 3D help.

 

In an ideal world, it sounds like you need a dedicated 3D guy. And while you've had bad experiences in the past with outsourcing, I can promise you that there are some seriously capable people on this forum who could far exceed your highest expectations. 3d guys are a dime a dozen, but good ones who fully understand the needs of architects are harder to find. This would be a great place to look for them. There's gotta be someone around here who lives in the Phoenix area, if that's a consideration.

[jackb602]

We've even outsourced some of this work in the past to a local company just down the street and, while the results were nice, the experience working with them, the process was terrible... especially for the price.

 

As a renderer myself, I'm curious about what made your experience terrible? Was it meeting deadlines, responsiveness to your needs, or something else? I've learned a lot over the last few years about what clients are looking for, but there's always more to learn. What would make a great experience for you?

 

Jack