No, it's not quite as bad as that - each unit of 1 on the master dial, will dial up the slave lights Intensity by a half a percent.  So the effect of a single 80% light should be just 80% divided by 7 = 11.4 or so.  I suppose I could scale the master dial so that 100 = seven lights at 14.3%, but this makes it messy for anyone who might want to use less than seven lights - I felt it made more sense to leave the master dial on a 1:1 ratio as the slave lights (even though it isn't, really).

I still have problems with the whole intensity setup going nuts anyway, I'm not sure if this will ever work well. 

I've come across another caveat.  The master Red Greeen Blue dials also do not work correctly - since the slave lights are already dialed to 1.0 on each channel, adjusting the master color dial only does half of what it's supposed to.  I can compensate for this on the dial (simpler) or have the lights load in with zeroed color values (more correct result).  This is so flaky I don't really want to give it out.

Finally I've figured out what is making the lights go nuts.  Adjusting the child lights via a pose file causes this.  Once it occurs in an active scene, it won't go away until the document is closed (revert to saved or some such).  While this is "a bug", I doubt any of what I'm doing is especially supported by Smith Micro, so I don't know if it's worthwhile to submit a bug ticket on this.

I give up.  I can get it to work very consistently the first time, but for any subsequent renders, all bets are off.  Probably it would be safer and more predictable to drive light controls via python script but I'm not interested in dropping everything for the next month or so to learn python.

Quote - Yeah I've just found the same thing. 

how did you model this metalic parts? with zbrush or mudbox?

All the modeling for the outfit was done in Cinema 4d.  IMO something like this would actually be harder in a sculpting app unless you just didn't care at all about poly count.  Stuff like the shoes I don't see how you'd do it in Zbrush at all.

 The black spots go away when you raise the shading rate for the object and/or apply a tiny amount of displacement (but be sure to render with displacement on) - its only a solution where you dont have too much detail, like in cars, or when you use uniform shaders with simple colors etc.

uuuuuuuuu thats bad.

The artifacts are real, no doubt. Stefan intends to fix this for SR1. I have a call scheduled to discuss renderer stuff with Uli, Steve, and Stefan Thursday.

If anyone cares to fool around with this sample scene file:
http://cid-b233dcaeefa9709c.skydrive.live.com/self.aspx/Public/Poser8%20samples/Simple%20GI%20Test.zip

Maybe you can get better results.  Constraints:

• No lights with shadows disabled allowed, although you may add additional lights if you feel you must
• No IBL lights allowed, period
• Fill lights are strongly discouraged
• Minimum # of bounces allowed is 3, so that all surfaces "technically" should catch some indirect light
• No deleting any geometry in the scene (there is a 10x10x10 box around the scene, to serve as a "room" environment).

Quote - The artifacts are real, no doubt. Stefan intends to fix this for SR1.

Yeah I think everyone gets that, I just keep hearing that turning down # of bounces is a good workaround.  It isn't, it's a terrible workaround.

Quote - The artifacts are real, no doubt. Stefan intends to fix this for SR1. I have a call scheduled to discuss renderer stuff with Uli, Steve, and Stefan Thursday.

Hum, a stupid thing, but, without hate ... the beta testers are really beta ? I say that cause i think GI is the most feature who most poser users wait for a long time, with render time, and finally it's the most bad !

May be i'm stupid !

piz99 - Thanks for the series and the scene. I'd started a similar series, but, stupidly, with a more complex scene/figure. I'll try a few settings.

Have you tried the IDL Only option on the Dimension3D script? Aside from saving a little time, it may give us a feel of what's going on at the IDL level.

Whichway

Yeah I understand you, I wish I'd been in the beta myself.

Ok, dumb question #1 - where do I find the Backdrop.obj file?

Whichway

Quote -
Have you tried the IDL Only option on the Dimension3D script? Aside from saving a little time, it may give us a feel of what's going on at the IDL level.

The intent of that feature is for those who would like to render in passes, the way rendering shadows or other effects to a separate pass; I never render with that technique so I'm not a good person to guinea pig it, both because I don't know about it and I don't really care enough to learn it.

Quote - Ok, dumb question #1 - where do I find the Backdrop.obj file?

Wherever your default runtime is installed ^^  The only scene contents are Poser 8 native stuff.
edit: oops, no, that's my own backdrop, sorry; I'll fix that and re-upload that file

edit: fixed

Ok, re backdrop. Will wait.....[edit - Got it!]

Re IDL Only: I just thought that since we were trying to understand what was going on with IDL, it might be easier to sort things out if we were rendering only the confusing IDL bits.

Whichway

There's a corrected version of that test scene put up, have at it :)

Working on it. Dang, now I probably won't eat dinner. Damn you pjz99!

I'm being realistic and not expecting this to be remedied for the next six months.  If it gets done sooner than that, because you skipped dinner, thanks in advance.

I'm betting 4 weeks.

if anybody tries the scene with d3d's script, please post the results.  stefan also had a script and may have more before long.

Paul,

I'm using the D3D script to control all parameters. I had been working on my own setup kind of like yours, and found some settings that improved things. However, loading the settings I found into your scene, the "room" is a solid wall of red dots - no gaps! The foreground objects are rendering quick, but the buckets with "room" in them are taking forever. Any idea why that would be? Why is the room different?

WTF?

Dinner (and sleep) is highly overrated ;) 

It's simply an 8x8x8 poly cube taken from the Poser 8 Runtime "PropsPrimitivesBox" , so I've no idea why it would be sucking up so many rays (this is what the red dots are I believe, ray hits).  However this is consistent with what I told Stefan earlier, that the artifacting problem appears much worse when you have some environment geometry of any kind, apparently the environment geometry eats up a lot of rays prematurely.  No idea why.

I was about to make the same observation. My little laptop may melt. Bailing again to look for something funny going on.

Whichway

Hiding or removing the "room" geometry is unacceptable under my constraints, for this test ;)  Replacing it with alternate geometry of the same shape is OK though (a higher or lower poly box).

pjz99 - If you're getting through that swarm of red ants in 3 min, you must have one whale of a machine.

Whichway

Oh wait, I got confused. There's a room and a backdrop? It's the backdrop making solid red dots.

Addendum: Never mind, the room does it too, even if I hide the backdrop.

Oh right, turn off Smooth Polygons for the backdrop prop as it looks like crap otherwise.

I think the room caused problems, perhaps because we're inside, with all the polygons facing outward? Replacing with my EnvSphere, where all the polys face inward, and it renders extremely fast.

If normal facing has a real impact on how GI is calculated, that right there is probably not desirable.  I'm not picky about what box gets used, but as my studio is not shaped like a basketball, a sphere is unacceptable for this test.  Must be a box.  If you like I can export a cube with inverted normals and make it a prop.

No need - I can reverse the normal no problem.

Okey doke, although if normal facing has any impact at all here I would consider that a bug that needs to be fixed.

Can dumb ole me reverse the normals in Poser?

Whichway

[Edit: Found it. Maybe]

Select cube
Group Editor
Create group
Add All
Reverse group normals

(actually in the case of Poser's default primitives there is already a group containing all faces so those steps "create group" and "add all" are not necessary)

I agree reversing the normal should not matter, you had "Normals Forward" checked. That's a bug. But easy enough to work around at the moment. The reversed room is rendering well now.

Thanks. I'd mistakenly found the Normals Forward option without reading the manual first. I'll follow your prescription now, Dr.

Whichway

I'm rendering and can't check what I did to screw up the transition between the backdrop front edge and the ground and/or room floor. I'm getting a seam.

Since I have to wait for the render to finish, I thought I'd just ask - is the "floor" supposed to be the ground, the room bottom surface, or the backdrop, or some combination? Did you intend for the backdrop to be above the "floor"?

I realize that the "room" geometry also should have have Smooth Polygons disabled (will make the corners pillow out undesirably).

Revised version available here:

• normals on the "room" geometry are flipped
• smoothing turned off for backdrop and for "room"
  http://cid-b233dcaeefa9709c.skydrive.live.com/self.aspx/Public/Poser8%20samples/Simple%20GI%20Test%202.zip

Quote - Since I have to wait for the render to finish, I thought I'd just ask - is the "floor" supposed to be the ground, the room bottom surface, or the backdrop, or some combination? Did you intend for the backdrop to be above the "floor"?

The room box was moved down one foot to prevent it intersecting the backdrop's floor.  If it's not moved down the renderer becomes unhappy and tries to render both surfaces in the same "space", no good.  Although I suppose some rays will inevitably get wasted in the area between the box and the backdrop I don't particularly care.  Incidentally that's another example of why considering normals in GI is not that wise, imo - the backside/underside of the backdrop prop will behave differently from the front side, very counter intuitive.  Default behavior really ought to be "ignore normals" regardless of any material settings.

Is  it possible to post more than one image per reply? I may have something interesting. Only JPEGs allowed?

Whichway