All posts related to V2
User avatar
By Fernando Tella
#322424
I've reduced the scene to a slate, a box and some lights; Still odd. As you keep deleting lights the problems keeps decreasing its intensity and sun shows faster, but it keeps strange till the last plane is deleted.

One thing I've noticed is that, when rendering only one plane (emitter), Maxwell console reads: 2 meshes, 2 triangles, 4 vertex,... while a proper plane should be 1 mesh, 2 triangles, 4 vertex,... Maybe it is just due to bad geometry.


(Edit: where the hell is motion blur in studio? :mrgreen: )
Last edited by Fernando Tella on Tue Apr 20, 2010 9:44 am, edited 1 time in total.
User avatar
By Fernando Tella
#322441
Bubbaloo wrote:
Fernando Tella wrote:(Edit: where the hell is motion blur in studio? :mrgreen: )
Only through certain plug ins. I'm only familiar with the Max plug in though. :oops:
Oh, so it's not there. I felt dumb for a while. I guess it makes sense as studio does not make animations. :P
By Peder
#322443
If that is the case then there is something going on with the export to Maxwell through the plugin. All the emitter objects are simple square polygons. Here is the project info for the emitters when I copy and paste them into a new project. As you can see the number of faces and vertexes match the number of objects.
http://idisk.me.com/peder.lindbom/Publi ... 104355.jpg

Perhaps there is something going on with the way surfaces are displayed in FormZ. They are shown as double sided so perhaps they get exported that way?

There is an animation tab in the Maxwell plugin exporter for FormZ that has checkboxes for Motion blur for cameras and motion blur for objects. These are not checked however and since I cannot find any setting inside Studio either it is impossible for me to know if there is some problem with the exporter.
User avatar
By Fernando Tella
#322445
Peder wrote:They are shown as double sided so perhaps they get exported that way?.
The problem could be there as Maxwell translates to triangles; a double sided plane could make a problem. Maybe you could try to export two planes, first one single and then one double (include some boxes to check shadows), and see if there's some difference in the sun and noise.
By big K
#322452
Peder, if they show as double sided within formZ, you should defenetly change them to single sided.
there is a tool in formZ which is named make 1/2 sided.
maybe you have used this accidently before and now your single surfaces have become 2-sided.
or there might be other reasons they are 2-sided, but make sure you use this tool to make them single sided.

(check with interactive view normals within formZ)
User avatar
By Asmithey
#322476
Also, Lets try changing the plane emitters to half cylinder emitters just to see what happens. When I get some time today, I will try this out.

Changing the emitter geometry from planes to half cylinders did not work. I still got the same results as before.
User avatar
By Mihai
#322527
Peder, in your particular scene the sun appears slower because you have a LOT of separate emitters (415), so in this case it's better to merge them into one and increase the wattage with the nr of merged emitters. You wouldn't see much difference between merged/separate emitters if you have something like 30-50 separate emitters, but with 400+ separate emitters, some optimizations in Maxwell won't work very well.
User avatar
By Asmithey
#322551
Hi there,

I tried this. After merging I increased the wattage to 40k, the sun appeared right away, but the emitters did not work right. The areas/ rooms with the emitters would not light up. And when I amped them up to ridiculous amounts say 100 and 200k still no light in the spaces and I get a strange pinkish preview. Miahi, were you able to get the emitters to function properly when merged? Basically the emitters will not work when merged.
By Peder
#322609
Mihai!
Here is the other project I mentioned with only 19 + 6 emitter objects and still the sun is slow to develop (noisy) at SL 9:
http://idisk.me.com/peder.lindbom/Publi ... 041145.jpg

Here is the final render I did. As you can see the sun was still very noisy:
http://idisk.me.com/peder.lindbom/Publi ... 041626.jpg

And here is the testrender again at SL 14.62 you can start to see the warmth of the sunlight develop:
http://idisk.me.com/peder.lindbom/Publi ... 043401.jpg

And if I hide the emitters in Studio I get much more sunlight already at SL 5:
http://idisk.me.com/peder.lindbom/Publi ... 044640.jpg

I am unsure about this whole discsussion about emitter strengths and the issue with creating them as separate objects or joining them. These emitters are set to be 40kW. If that gets multiplied on a per object level that is obviously not right. But to my eyes the intensity of the emitters looks about right as they are -surely not an indoor sun anyway. This was not rendered as a Multilight either.

Here is the MXS file together with the FormZ project file and hopefully all relevant .mxm files both outside and inside the .mxs file for further forensic analysis if you are interested:
http://web.me.com/peder.lindbom/AIX/Max ... test2.html

The above seems -to me- to negate your theory that the reason for the problems lies in the number of emitter objects together with Maxwells optimization scheme.
User avatar
By Fernando Tella
#322617
I believe it's all about bad meshes generated in the export process. The noise generated in this scenes is not usual Maxwell noise.

If I open it in studio and replace the light meshes with other planes imported from Maxwell library it renders completely different.
By Peder
#322618
OK. That sounds like an interesting lead. I did ask Ben that writes the plugin for FormZ to check the export and he did not find anything unusual about exporting 2d planes. Guess I'll have to ask him to look again.
User avatar
By Fernando Tella
#322623
I retract!

Well, after some more test I think I was wrong in my last post.

I reduced your scene to a new cube, a new plane and the lights, the problem is still there.

Reduced to one light, I can see it a bit odd, but of course it is a 40000W light.

Reduced to 1000W (more like it should be for a light of that size ¿around 8m long?); everything looks fine.

Deleted original light and replace with a new plane, assign same light material (1000W); everything looks fine.

Copy light multiple times; sun takes a bit longer to show but it's normal for the amount of lights.

Increase the power to 40000; shadows are weaker as we are setting lights too strong. The amount of noise is the same.

Conclusion:

I join the wagon of the ones that suggest to keep to real values; if you spread 40W over an 8m long plane it will be hardly seen, besides you are trying to see it under sunlight and with ISO 100 and shutter speed 100. Don't multiply the intensity unless the lights are the same mesh. To check interior lights I usually set ISO 400 and SS 50, which will make the image overexpose with daylight.
By Peder
#322667
OK I think I can finally put this one to bed. There is definitely something wrong here. I made a super simple testfile with just one room object and 36 emitters in the ceiling. The scale is realistic -the room being roughly 14x12.5meters and 9 meters high. I first did a test with individual objects for the emitters and set the strength of the emitter to 100W (default efficacy) and let it render to exactly SL6 This is what I got:
http://idisk.me.com/peder.lindbom/Publi ... 231119.jpg

Then I did the same only change being joining the emitters to one object and then increasing the strength of the emitters to 36x100W = 3600W Rendering again to exactly SL6. This is what I got:
http://idisk.me.com/peder.lindbom/Publi ... 230423.jpg

As you can see the problem is identical to the one I have been trying to describe. Also I have asked Ben to check if there is something funky going on with the export but he cannot find any fault there. This problem does not seem to be related to unrealistic lighting levels but rather related to joining planes or using individual geometry for emitters. As you can see from the renders emitter strength is consistent and I did not touch exposure at all so it is set at 100/100.

Finally (again) the project files:
http://web.me.com/peder.lindbom/AIX/Max ... test3.html
Help with swimming pool water

Hi Andreas " I would say the above "fake[…]

render engines and Maxwell

Other rendering engines are evolving day by day, m[…]