[hyltom]

Try to create a dielectric material using the wizards in both 1.7 and 2.0 and render those for X amount of time and see what looks better and cleaner. Then you will know.

/ Max

I did it and...well i'm far from being impressed!



Have I forgot to check the option "give me a juicy render in 1/10 of the time of v1.7" ?

[Richard]

Hmmm? Looks like the reflections clean up better but the floor is still not as clean! Given in architectural exteriors you can't even use dialectrics still then I cant imagine a great time saving.

[Bubbaloo]

I would have bet an upgrade that the simball would be the currently most optimised V2 scene out there!

Question? From all this talk about optimising scenes, I thought we already had to do that with V1.7 - AGS, single plane emitters, lots of non-physically correct light panels to flood a scene? Can we assume the speed gains of most V2 scenes come from further non-physically correct trickery to gain the speed?

This really doesn't make sense to me. AGS is used to eliminate unwanted refraction calculations to speed up rendering. Caustic illumination (light through dielectrics) still requires heavy calculations and therefore render times. It's really a judgment call whether to use AGS or not. Personally, for exterior renderings I use real glass with solid window geometry, and render to very high S.L. For interiors, where light is coming in through an exterior window, AGS is great.

Non-physically correct light panels? Huh? Ask any photographer what they think about light panels and extra "trickery".

[micheloupatrick]

Where can I find these "test balls"?

[Richard]

I would have bet an upgrade that the simball would be the currently most optimised V2 scene out there!

Question? From all this talk about optimising scenes, I thought we already had to do that with V1.7 - AGS, single plane emitters, lots of non-physically correct light panels to flood a scene? Can we assume the speed gains of most V2 scenes come from further non-physically correct trickery to gain the speed?

This really doesn't make sense to me. AGS is used to eliminate unwanted refraction calculations to speed up rendering. Caustic illumination (light through dielectrics) still requires heavy calculations and therefore render times. It's really a judgment call whether to use AGS or not. Personally, for exterior renderings I use real glass with solid window geometry, and render to very high S.L. For interiors, where light is coming in through an exterior window, AGS is great.

Non-physically correct light panels? Huh? Ask any photographer what they think about light panels and extra "trickery".

Brian, mate don't get me wrong AGS and me are really great friends by now! I think it is a solution I would likely continue to use even if render times dropped significantly.

And again the light panels are understandably a big part of lighting a scene as a photographer would - I've just done a catalogue for a home builder and rather than paying out for internal renders they took photo's of their existing interiors with a compact camera with just a flash - only 2 of the 30 images they gave me could be used and only with lots of processing.

I guess my point is that all this talk of optimisations of scenes I thought would have been something almost redundant with the speed increases touted prior to the current release. As Hylton proves here that much is not as it seems! I guess I'm a bit disappointed that V2 isn't as feature pack as I feel it could have been for a new release.

[JDHill]

micheloupatrick: look in your Maxwell folder under library/scenes/simball.


On the topic of the thread, here is a quick comparison (run on a Q6700):


1.7.1 | 10m | sl 8.91


2.0 | 10m | sl 8.34


The mxs was written with a 1.7.1 plugin and rendered directly in both versions. Here is a version with some material tweaks:

2.0 | 10m | sl 9.26

[knurrebusk]

I do´nt care if it´s faster!

Is it correct, and is it reflecting the beauty of life?
I can understand both Juan/ and old Revit/founder, I´m close friend to same persons.

I told one of my best friends who worked for etc, he is shit full of money, you still need to love life.
He has in him the brightness of Nicolas Tessla, and feel comfort in laying down in the sweet love of IQ.

I will preach against the choir if i ever return with architecture on this board, the only person who nagged you persistantly is Korean.
You all know this persistant great person, needed an asian powerhouse to present my case.

So will Juan accept my humble respect, and challenge from fellow minds from Korea.

I will front MR if you listen to the argument of compromise/style go hand in hand.


She/he, dont care just ask a simple question?
Little light from sun into small opening, make great bright light from nothing

Capiche?

[Richard]

Capiche?

Hmmmm, almost!

[hyltom]

micheloupatrick: look in your Maxwell folder under library/scenes/simball.


On the topic of the thread, here is a quick comparison (run on a Q6700)...

This is very interesting as it's very near to the studio setup i use. Unfortunately, i have done the same kind of test, use some old scene exported with 1.7.1 and render with 2.0. But the result is worst, nothing near with what you are showing here.

I will do more test and try to post them here today.

[Bubbaloo]

Dude!

[mashium123]

yep.... back again!

[Neeper]

micheloupatrick: look in your Maxwell folder under library/scenes/simball.


On the topic of the thread, here is a quick comparison (run on a Q6700)...

This is very interesting as it's very near to the studio setup i use. Unfortunately, i have done the same kind of test, use some old scene exported with 1.7.1 and render with 2.0. But the result is worst, nothing near with what you are showing here.

I will do more test and try to post them here today.

I haven't tried MW 2.0 yet, but are looking forward to it.
It doesn't seems like you are the only one experiencing 2.0 not to be faster than 1.7.

Can this be a setup problem, like maybe your 2.0 isn't setup to use all CPU cores or something?

Kim

[JDHill]

This is very interesting as it's very near to the studio setup i use. Unfortunately, i have done the same kind of test, use some old scene exported with 1.7.1 and render with 2.0. But the result is worst, nothing near with what you are showing here.

I can't speculate on what you're seeing there. Focusing too much on the simball scene may be a problem - when originally created, it went though a whole lot of testing an optimization; this time around, I could not say how much work was put into it, but I don't think it comes close to what was originally done. All I can say in this case is that until this afternoon, I hadn't rendered the above scene since I originally created it some months ago, so I just tried it out here today to see what would happen...I did not know whether it would be greatly improved in 2.0 or not, and obviously, it was.

In general, whether just doing test renders, making the think! plugin tutorials, demo-ing at siggraph, or doing a few production projects, it is my position that I would never go back to using 1.7. As we see above, though it is generally possible to directly render a 1.7 scene in 2.0, I would not personally recommend doing so - the changes to the material system are not minor ones, and there is a good amount of gain, both in optimization and in overall material quality, to be had by getting familiar with it.

[hyltom]

Just render this scene (exported from rhino plug-in for MR 1.7.1)

V1.7


V2.0


For me there is not much difference (worst in some area, better in others) excepted in the transparent plastic part that definitely looks better with 2.0 but is that not due to the darker color?

I'm still looking for a scene that will show me a real time improve. Actually, it will be good to share some of the scene that A-Team used to compare 1.7 and 2.0.

[hyltom]

Another one

V1.7


V2.0


What ever i tried, i can't get better with V2.0 without redoing all the setup and material if i understand well. It's insane to have to redo all the materials every time there is new version. During the past few years i have build a large material library and now they appear to be useless.