maxwell archicad plugin - thoughts

[neba77]

I was just wondering on your thoughts about maxwell being integrated in archicad. Lot of architecture studios here already use archicad in their production. What about archviz industry? Will it feel the imapct?

 

Wanted to hear your thoughts, because i feel that i will definetly fell the impact on my business, i think that my services as an architectural visualiser will be noticable less wanted.

 

Thanks.

[mzex]

I was just wondering on your thoughts about maxwell being integrated in archicad. Lot of architecture studios here already use archicad in their production. What about archviz industry? Will it feel the imapct?

Wanted to hear your thoughts, because i feel that i will definetly fell the impact on my business, i think that my services as an architectural visualiser will be noticable less wanted.

Thanks.

 

I haven't seen Maxwell, but I think it is not a Artlantis like software, which is very easy to learn software and still it is very unknown to architects. Not many people are using all the benefits of the software. Example is ArchiCAD, most of my friends know how to use only basic commads in ArchiCAD. Same goes with AutoCAD etc. My point is if Maxwell will be very good and easy to learn only few will really use it.

[Devin Johnston]

ArchiCAD is very versatile and with the addition of the Maxwell plugin it's going to be very easy to do renderings of a very high quality. Right now I import all of my models from ArchiCAD into Max but I'm thinking at least for still renderings that getting rid of Max might be a good way to work. One problem with this whole scenario is how fast will Maxwell be? According to my calculations you will need a small rendering farm of 32 processors just to get out a print rez image in less than 12 hours. So I don't think many Architecture firms will immediately jump into using Maxwell simply because it's to slow and they may not have the resources.

[jntdigital]

I have tried the maxwell link to archicad. it will be awhile before it is suitable for production. the plugin barely works right now. plus the render speed of maxwell is very slow at the moment.

[mzex]

ArchiCAD is very versatile and with the addition of the Maxwell plugin it's going to be very easy to do renderings of a very high quality.

 

But Neba won't feel any impact to his business.People are too lazy to learn Artlantis, which can get you just fine renderings, so I think they won't learn Maxwell.

 

When is Maxwel for AC coming out?

[Devin Johnston]

I assume that everything will be available on April 26th, but that is a guess.

[neba77]

But Neba won't feel any impact to his business.People are too lazy to learn Artlantis, which can get you just fine renderings, so I think they won't learn Maxwell.

When is Maxwel for AC coming out?

 

 

Heh... same thing in croatia then....

 

Thanks guys for your replies....

[ivox3]

Maxer, ......how did you calculate 32 procs. necessary for a 12hour image?

 

Curious.

[Devin Johnston]

I took what I believed to be the time it would take to generate a fairly noiseless image, say 100 hours on a single dual processor dual core machine. Then I did a little simple math to figure that it would take 32 processors of equal speed to give me a result in 12 hours. That's assuming that V1.0 is only marginally faster than the beta is and you get about a 1:1 ratio in performance when you use cooperative rendering. Given that I don't know how fast v1.0 is going to be I assumed that there wouldn’t be much of a speed increase since some of the A-team members have been talking like there won't be. My times might be totally off and V1.0 may be much faster than I'm giving it credit for, in that case I would be pleasantly surprised but I'm going to assume it's going to be very slow until I see evidence that it's not.

[ivox3]

thanks Maxer.

 

but what is a single dual processor dual core ? I think that's a typo.

[Devin Johnston]

I guess I should have said a singe dual processor dual core machine.

[ivox3]

...sorry. it just sounded funny. ....maybe just this: dual processor dual core.

 

Addmittedly, ....all the dual,dual corex2, dual processor x2 etc...terminology is getting a bit confusing. And now,.....we're going to be commonly adding in the whole quad thing.

 

All the same, .....thanks for the reply.

 

We'll know on the 26th. ....at least, hope too!

[mzex]

yes I am waiting that 26th. Maybe my Archicad will finally become a complete software :-)

[josephus]

Hey Maxer, your 100 hrs using a quad-core seems really steep based on the posts I've been following on the NL site the last three months (I know that you're a "regular" there too. I've done some Maxwell interior rendering tests w/RC1 and RC5, but I've never let it run to clear out the noise, although I was rather expecting with my P4 3.2GHz single processor that it would have to run for about 50 hours. A LOT depends on your emitters, I've seen a lot of Maxwell arch renders using only skydome that looked really good in less than 20 hrs "cooking", but the time dramatically increased with the number and type of emitters. I've put Maxwell on hold until V1.0 (as many have), at which time I'll jump in again....those adjustable emitter controls are fantastic, they will save a LOT of my time. I've actually been using Maxwell with the FormZ plug-in which worked very well and was actually REALLY easy to use. I'm somewhat under the impression that the Archicad plug-in was not working as well. We also purchased Archicad a few months ago, so all this will be very interesting to follow as I don't care for the Lightworks rendering engine inside Archicad although the Lightworks rendering engine inside FormZ produces decent images...but takes a lot of tweaking of the lights.

[AdamT]

but I've never let it run to clear out the noise

That explains it then. The problem, as you know, is that the time between levels increases geometrically, so even if you have a *pretty* clear image after 10 hours, it could take another 100 hours before it's *really* clear.

[josephus]

That explains it then. The problem, as you know, is that the time between levels increases geometrically, so even if you have a *pretty* clear image after 10 hours, it could take another 100 hours before it's *really* clear.

 

Yeah, I'm aware of the arithmetic progression in time required between levels I've been working on a P4 3.2GHz laptop and was afraid after "cooking" with the fan running non-stop for 24 hrs that I was going to fry the cpu, so I stopped. What I was basing my estimate on was more what I was seeing in the many posts when time, SL, and machine config were posted. I really hope Maxer isn't right on about this (although being "right-on" is good), because the four Athlon X2's I just purchased to run MW (since we all got a freebie license ) are only equivalent (maybe not quite, but close) to two of Maxer's quads...hence at least 50 hours render time

[Devin Johnston]

josephus I hope your right about me being wrong but were not going to know till 1.0 comes out. You’re also right about emitters slowing down the process; however what's the point of having them if you can't use them? The way I like to test software is to give it the hardest possible scene I can come up with and see how it performs. Yes the times might be slow but it won't give you any misconceptions on how the software performs under difficult circumstances. All of my tests so far have concentrated on interiors since exteriors render reasonably fast at print rez, let's say less than 30 hours. Also most of the images you see on the Maxwell forum are screen rez not print rez which is really what I'm interested in since most of my clients want printed images.

[josephus]

All of my tests so far have concentrated on interiors since exteriors render reasonably fast at print rez, let's say less than 30 hours. Also most of the images you see on the Maxwell forum are screen rez not print rez which is really what I'm interested in since most of my clients want printed images.

 

Maxer, thx for your reply. You obviously have tested MW a lot more than I have, so I guess I'd better start thinking about looooooong render times. And true, print res is what matters, hence the huge time increase. I agree fully with your expectations/assessment regarding v1 speed vs beta...not much better. I also agree with being able to use a lot of emitters, although my for my initial tests apparently the emitters had way too many faces.

 

You said 100 hrs on one quad machine (right)? How fast is your quad (Mac)?

[Devin Johnston]

It's an dual core dual processor AMD Opteron 275 which is 4x 2.2Ghz. The last test I did was an interior with sunlight and emitters along with blurry reflections, basically everything that makes a scene render slow. I rendered it at 2500x2500 pixels for 64 hours and by the end I still had quite a bit of noise that even neat image couldn’t get rid of. 100 hours is just a guess as to how long it would take to get a mostly clear image, in reality it could take much longer on my single machine.

[josephus]

It's an dual core dual processor AMD Opteron 275 which is 4x 2.2Ghz.

 

I was afraid you might say something like that....they did some performance tests on the FormZ forum a while back, and the AMD Opteron 275 clocked the best times, outperformed the Mac Quads (not by much though).

 

I could probably work very acceptably with a smaller image for prints than the 2500x2500, like 2000x1200....looking for ways to cut the time without sacrificing what makes MW so fantastic.

[Devin Johnston]

agreed, good luck!