Quote - shading rate 0 ????
pixel samples 12????
raytrace bounce 6? 

holly s..... those settings are insane.

The 12 pixel samples (actually, its 12x12=144) are wasted when using just a 1 pixel box filter. If you care about good antialiasing, use a large sinc filter instead - try it. Typically, 5 pixel samples should be sufficient for most scenes - only in cases with complex thin geometry (dynamic hair for example) more samples will make a difference.

i use 3 and it looks good enough IMO.

5 would be MAX for me.

In my discussion with piz99 we talked about to get rid of the GI artifacts when high raytrace bounces in use. To get rid of this artifacts you must raise the setting. My tests maybe insane but was necessary to show differences in setup. :rolleyes:

Mazak

Lowering Diffuse_Value did not help me with the corner-y artifacts I've been getting, although it does seem to help for large smooth surfaces, and I agree that it should probably be kept lower than 1 for all surfaces, I'll do that for everything GI-related in the future.

I still get some garbage around the border of the image but that's not quite so bad, I can just render a few pixels larger and crop the outside to get rid of it.  There is still a little trash around the thigh and boot.  I think if I add another light I could get rid of it.  Render time wasn't too bad either.  Antialasing needs a little work - fairly sure I left post filter at default, I'll change that.

Ignore that shit in the bottom right, that is coming from the "environment cube" I replaced the environment sphere with, to simplify things, but I didn't remember to drop it below the backdrop.

Mental note to set up a "light rack" because this method of having four lights all point at the same target isn't really ideal (many rays converge on a single point rather than acting like a cheapie area light as desired)

I'd say that looks considerably better.
I've been trying to fake area lighting in Poser for a good while now, and yeah, a whole shitload of spotlights seems to be the closest I've been able to get.
I'm still finding it hard to believe though that poser 8 doesn't have area lights. To me it seems GI is only halfway there if all you have to work with is straightforward raytraced point or distant lights.

in re: area light, using a glowing box or sheet as light source (no point-source lights in scene to cast sharp shadows) should give softer shadows but will also probably increase the splotch artifacts.  I think they may be ironed out once all the details of the new variables are explained.

Render time was something like 15 minutes? on my intel qx6700 2.66GHz.  I didn't really time it.  It was much faster than the approach of scaling up 10,000% though, many times faster.

I still think the shadows are wrong for GI. That's what you get with Poser's type of lights though, I guess.
Looks better though and the shadows are acceptable, considering your room isn't all blown out, and sharper shadows are more likely.

I have a feeling though we're soon going to see a lot of self-proclaimed "photorealistic" renders of people in bright rooms with impossibly unrealistic shadows. ;-)

Isn't there a python script out there which we can run to change all the diffuse values on something to .8 instead of doing it by hand?

Keep in mind that the shadows on the floor you're seeing are mostly perpendicular to the camera view and the blur is getting squished.

The realism gallery here is almost 100% Poser renders, often with nostril glow or long hair hovering parallel to the ground or what have you.

so since it says 8 bounces does this mean that the light bounces 8 or 6 times? 
can you show the same render with only 2 bounces? thanks

This is another angle that shows the shadows more clearly - the four key spots are spread out more than they should be, since I've not put any real effort into setting up the "light rack" approach, so it's a little too obvious what's going on.  The GI light is not as "spready" as I would like it to be, but this isn't too bad.  There are a couple of GI blotches on the leg where the boot is against the character's skin, and these do not go away even at max quality settings, but it's a lot better than I was getting with only one key light.  I have not touched tone mapping or attempted any gamma correction setup in any of these pics yet.  That may help the GI light to appear more "spready", hopefully.

The documentation on tone mapping settings is remarkably thin (less than a page) but I'll try the earlier full-body pic with default settings and post that shortly.

I'm guessing and have not done extensive tests to prove any of my guesses to even myself. Having said that...

I think that Exponential is nothing more than the final gamma correction step. I.e. the "Exposure" is an exponent. Each of R, G, and B of the final output are raised to the power 1/exposure. In other words, at 2.2, this is identical to GC at 2.2. But ! It does not do the pre-compensating anti-gamma on the incoming material the way Poser Pro does. Using only the final gamma correction, without matching incoming anti-gamma, you will lose saturation. I figure this mode is pretty much useless.

I think that HSV Exponential is similar, but instead of raising all three components to the same power, it only does it to the component that is brightest. For example, if R is the red component of a pixel coming out of the shader, and that R value is brighter than G or B, then R' = R ^ (1/exposure). Then G and B (the weaker components) are modified by multiplying them with the ratio of R'/R. In other words, after the exponent is applied to get R' from R, the other components are altered by the same ratio. This tends to preserve hue and saturation very similar to full blown GC without dealing with the incoming anti-gamma step.

It is not quite as accurate as GC. But my experiments with this indicate that it is very close. So close, that moving a light by a few degrees would produce a bigger difference than HSVETM versus GC.

Given my supposed understanding of how HSVETM works, I'd argue that the default value of 1.6 is wrong. I have been using 2.2 and getting results that look almost the same as Poser Pro with GC=2.2.

I tried it with the default exposure of 1.6 and it was ever-so-slightly different from rendering without it.  There was better handling of areas of the image that were only lit by indirect (bounced) light, but it was hard to tell.  I am re-running with exposure at 2.2.
oops, ps: mode is Exponential
edit/oops again: since Bagginsbill suggests that Exponential is pretty much useless, aborted and restarted as HSV Exponential

Smooth Polygons is often the culprit on the GI artifacts. Needs a little work there - it's the classiic REYES self-shadowing problem. There's no Min Bias parameter for GI - we need one.

One thing mentioned in another thread is to scale up by 10,000% which kind of struck a chord in what's left of my brain.

I know renderers like VRay prefer that the scene is modelled in roughly real world scale because it makes the lighting more realistic.  I'm just wondering if Posers stupidly small scale is throwing things off.
(of course - I could be talking complete cr*p here, in which case feel free to tell me to STFU noob :biggrin: )

if anybody is seeking a "GI bounces" variable (which they had in the previous version(s)), just wanted to mention that if the new renderer is like other professional renderers with GI/IDL/ISL, then there should only be ray-trace bounces IMVHO.

Why would you say that? Other renderers that can do GI treat reflection depth and GI bounces separately.

The "scale up 10,000%" method results in truly enormous render times.

"Other renderers" cost more...

no, pj, I'm just saying that, in this first iteration of their new GI GUI, they aren't gonna let users adjust some variables, perhaps because they'd go nuts trying to figure out what the best combination/permutation is.  personally I believe that in a ray-traced render, I would want to simplify it to the point where there are the same number of bounces specified by one variable for reflection, indirect diffuse, indirect specular and whatever other optical processes involve bounces.  I wouldn't wanna ask poser users to try to set 2 different bounce variables like in the earlier version(s).

Not even in "Advanced Render Settings"?

I also agree with BB. Actually I'm kinda surprised to find out Poser 8 has a GI option, yet apparently no "Advanced Render" settings.

Maybe they should have locked render dimensions too, because it may confuse some users to be forced to select a size for their output.  Probably being presented with a choice of filename is a challenge for some as well, it would have been better if that was hard-coded to "File1" "File2" File3".

PJ - Is your build number 10157?

Yes.  I think I see where this is going.

nudity

LOL!

Sorry for laughing and all, but that's just hilarious there, phionix. :blink:

I'd advise everyone to stop fucking around with whatever you downloaded today and take a break.  Start fresh tomorrow and re-download whatever the correct version is.  All this work of the past day has been a complete waste of time.

Are you saying they put the wrong version up for download?
I think I made a good decision to wait until the weekend or maybe even later.

Are you serious pj?? We have an incorrect build??

what is it that we are not knowing BB??

Ignore me, listen to Bagginsbill.  He is paid to be nice, I am not.

I dont care about nice, i just want the skinny

According to other replies made very recently, it looks like y'all are gonna need a hotfix that I think was supposed to be supplied with the download. Something like that.
I'd post a link to where i read that but I already lost it and things are moving too fast here to keep up with it all. ;-)

It was not a mistake. You got the correct release build.

There is one after the release. I thought it was public knowledge and I thought it would be included in your download list in addition to the full install, the content, and the legacy content, I thought you all were downloading it.

I can't explain why it was not published on the same day. All I can say is it has a few small changes that were corrections to the official package.

It is a HotFix - an additional download. It will be available by end of week, but probably sooner. I have it already and thought you did too.

Quote - if anybody is seeking a "GI bounces" variable (which they had in the previous version(s)), just wanted to mention that if the new renderer is like other professional renderers with GI/IDL/ISL, then there should only be ray-trace bounces IMVHO.

easy there. ok? 

you you need to have seperate control for raytraced bounces and lighting bounces. lets say i have a car in my scene. and i want 5 bounces of light. does this mean that i should render out a car with 5 raytraced bounces? are you even aware how much this is? noone uses 5 bounces for reflection.

Quote - hard to know where the tiles go when they're procedural

Does your video card properly support OpenGL? If so, you can get a very reliable preview of many procedural shaders. Under Render Settings:Preview, see if there's a bit of tiny green text that says "hardware shading supported", if so, tick the two boxes under it, hit "save settings", and then look at the viewport :)

It's pretty accurate, yeah, although depending on how fine the tiles are you may need to zoom in pretty close.

Nope, we don't have it.

:)

o.k., mike, thanks for the scrnshot. if anybody wants those items in poser, it may be possible to request those in kupa's PP2010 thread, hopefully this week, before the thread drops off the page.

I understand the part about setting bounces, but it may currently be simplified so the user only has to set ray-trace bounces, presumably because GI bounces are fixed at 3 (or some other figure designed to allow quick renders) by the software, just prior to render.  they may be able to add more  variables to the render GUI if enough people show interest and they have the budget.  but I hope most users can become accustomed to the current set-up first.  99.9% of these users have not tried to do GI renders in poser prior to the introduction of this new render GUI.

No, I actually got it direct from Stewer (Stefan Werner) that the Raytrace Bounces slider is controlling GI bounces as well as reflection depth (meaning, it goes up to 12, plus or minus one depending on what Bagginsbill was saying earlier about the exception made for GI bounces).  He said this was a design decision - hopefully they change the design and separate these functions, because it really makes no sense to keep them on the same setting.

Quote - PJ - Is your build number 10157?

10154 here, on my Mac version...