Just displacement, BB Env, and I light object emmiter (DiffuseShort Lit 5, scaled 200%).

Dis, SSS as above. I'll try adding more bounces.

I added made it two ray trace bounces, then add 2 indirect light bounces, still not seeing it.

ISTR bill mentioned SSS doesn't scatter blue well ~ 1 year ago, hence SM knows about it, and it mightn't have those blue areas if somebody tried a diffuse emitter/envsphere with no blue, which would not look good.  however, if snarly's or somebody's scene is only using one bounce, that is not enough for realism.  poser can do 16 or more bounces for extra raytrace depth in a completely enclosed building scene, but I reckon 4 bounces is more than enough for renders in empty scene. 

@ mori:  the reason I mentioned IDL IC = 100 is that it skips the precalc (as you noted) and sets GIMaxError to zero (no interpolation), but apparently the blue is a scatter error (in case any of the scatter error variables have been tried), not a GI interpolation error.  AFAIK they speeded up the no-precalc method for the next release.

Does that scatter issue happen if you use the "custom_scatter" node rather than the skin scatter node?

Quote - @ moriador

 Did I understand your screengrab at rdna correctly?

 Did you put a SSS on the light emittor? The object with ambient at 5.2?

 The SSS node should be on the recieving object, not on the IDL light emittors.......

 That would explain all your blue . . . . . . on the recieving objects

No, the SSS is on the ball prop.  The light emitters have no nodes attached at all. They are simple white.

Quote - This is similar to a bug I reported early PP2012 during beta. I found completely different colors in scatter nodes (any kind) affected by render size - no other changes required.

It was super obvious, then. Didn't need displacement to do it.

I point this out not because this is the same problem. Rather, because I learned then that scatter is affected by render size - part of it's calculation depends on that.

Just report this to SM. Quickly, before SR3 rolls out.

Thanks. Glad to know I'm not crazy. Have reported.

Quote - ISTR bill mentioned SSS doesn't scatter blue well ~ 1 year ago, hence SM knows about it, and it mightn't have those blue areas if somebody tried a diffuse emitter/envsphere with no blue, which would not look good.  however, if snarly's or somebody's scene is only using one bounce, that is not enough for realism.  poser can do 16 or more bounces for extra raytrace depth in a completely enclosed building scene, but I reckon 4 bounces is more than enough for renders in empty scene. 

@ mori:  the reason I mentioned IDL IC = 100 is that it skips the precalc (as you noted) and sets GIMaxError to zero (no interpolation), but apparently the blue is a scatter error (in case any of the scatter error variables have been tried), not a GI interpolation error.  AFAIK they speeded up the no-precalc method for the next release.

I did try a completely white emitter with a completely white sphere (terradome, actually -- no maps, just ambient value, no color). I switched to ENVSphere because most people around here have it in their runtimes.

I used 3 bounces, and I still got the issue.

When I rendered with IDL IC at 100, I got impatient and stopped after an hour and 36 minutes.  About a quarter of the ball had rendered by then, and it was apparent that the entire thing was going to render completely blue.

Quote - I added made it two ray trace bounces, then add 2 indirect light bounces, still not seeing it.

Oooh! Well, I would really like to figure out what you're doing differently. :)

I ended up trying a few higher setings but I wasn't getting it. I wonder if it's a combination your using? I was just displacement and a SSS (skin1) for testing. Off for the weekend I hope you can figure out what causes it.

mori, white light has 33.33% blue component, hence one way to eliminate the scatter failure may be to use orange light (no blue), but of course that's not realistic.  my feeling is that somebody will come up with a complex set of math nodes to compensate for it.

It's a bug and there's no way to compensate.

It's not the presence of blue - it's the absence of red.

I find it kind of strange that this is linked to render size. I don't doubt it - it's just mystifying to me why it should be so.

I wonder - is it the render size in pixels that is the determining factor? Or the calculated size in inches, based on the DPI? I know that doesn't make much sense, but no less sense than the pixel dimensions being the factor!

Just wondering if setting the DPI to a different value might make the render size smaller (in inches) while keeping the same pixel dimensions, and so fix the problem?

Weird.

Weird is the right word, yes.

Distance matters. Can be but could not duplicate.
But displacement value matters too.

My test figures render fine with "normal" displacement values, but crank displacement up, and they get "blue" ill. :-))) at the same distance....

Sierra
Tango
Romeo
Alfa
November
Golf
Echo
= Strange :-)))

OR?????

We have 2 different bugs !!!!

A bug depending on rendersize is in itself not so strange. A bucket calculates/evaluates now a smaller area and gets more detail. Since IDL is an approximation of the light it might give a different result for a pixel. It should be the same - but that's where the bug comes in

Here's my attempt at lighting a basic scene with just IDL.  It simply uses an Envsphere and the Spotlight, Diffuse Tall and Diffuse Short photo gear props from BB and Dreamland's Car Patio... original was rendered at 4096 pixels square, at which size the little imperfections become (more?) glaringly obvious, on close up view. Took about thre hours to render at that size...

Click the image to view at 2048 pixels wide in my gallery here... that was as big as I could leave it whilst retaining the image quality and getting in under the file size limit...

Cheers

i have an image to post but first I have to get that last image posted out of my mind.

Meanwhile............................................

Point of information...........

I have PoserPro2012 sr2      9/0.2.21561

When I set Irradiance Cache and Indirect Light Quality both to 100%, I was under the impression that would tell Poser "Brute Force" and it would not process a "precalculating indirect lighting" pass. When I render with those settings, it DOES process that sequence.

Was i wrong in my information?

::::: Opera :::::

She seems quite startled-but is not looking at the black Jaguar. I think the shadow artifacts on the chair are due to shadow Minimum Bias settings-I can't remember if they should be set higher or lower than the default 0.8000 setting to correct that.

Quote - Here's my attempt at lighting a basic scene with just IDL.  It simply uses an Envsphere and the Spotlight, Diffuse Tall and Diffuse Short photo gear props from BB and Dreamland's Car Patio... original was rendered at 4096 pixels square, at which size the little imperfections become (more?) glaringly obvious, on close up view. Took about thre hours to render at that size...

Sorry.

I was in two minds about whether or not to leave the snarling big cat in there... not sure if it detracts too much or not?

Yeah... I was also confused about what was said earlier about Irradiance Cache and Indirect Light Quality.

I am simply going with Intensity at 1.4, Bounces 7, Samples 200 and Irradiance Cache 20.0.

Any suggestions what I should adjust those to to improve the results I'm getting? Is it intensity and irradiance cache I should amp up to the max?

I'm using the Dimension 3d script to configure render options by the way.

Anyway... one of the other issues niggling me with that render was getting the eyes to not glaze over in such a bright scene.

The girl's eyes, not the viewers

So I'm going to bite the bullet and try playing around with the BBEye I reckon...

Cheers

Quote - She seems quite startled-but is not looking at the black Jaguar. I think the shadow artifacts on the chair are due to shadow Minimum Bias settings-I can't remember if they should be set higher or lower than the default 0.8000 setting to correct that.

 

Quote - Here's my attempt at lighting a basic scene with just IDL.  It simply uses an Envsphere and the Spotlight, Diffuse Tall and Diffuse Short photo gear props from BB and Dreamland's Car Patio... original was rendered at 4096 pixels square, at which size the little imperfections become (more?) glaringly obvious, on close up view. Took about thre hours to render at that size...

Ah, but what is the cat snarling at?

... I think I need to show a close up of her eye reflections perhaps?

Yes the chair artifacts are also bugging me. I have no light so where is shadow min bias set if not in the light? Is that "shading rate"?

Many thanks

o.g., if these are your settings, they set IDL IC to 95, causing precalc.

use d3d's renderFF script to set IDL IC to 100.

monkey, one sets the accuracy of occlusion (IDL) shadows using IDL samples (script), IDL IC (script), shading rate, pixel samples et al.

Thanks - in that case, yes... the "Min Shading Rate" in the standard Firefly options dialog is the same thing as the "shading rate" value shown in the D3d script.

Set off rendering now with min shading rate of 0.2 and 100 for IR cache and Intensity... will see what the difference is.

Cheers

Thanks Miss Nancy, that worked. You don't save 100% of the time, right, because the IDL calculation is performed during render. But is there a net resultant saving?

:: og ::

Okay... so upping the "Intensity" setting in the Dimension 3d Render FF dialog to 100 was just resulting in a massively over-exposed looking image. I have just set that back to 1.4... so either I was mixing up this setting with something else or it just doesn't need to be so high for my scene.

Irradiance cache at 100 was just taking silly amounts of time to render. I've gone back to 20 there. From looking at the render (as far as it got before I lost patience, about 5 hours in, and aborted it) I just couldn't see any clear benefit.

Obviously my scene isn't particularly enclosed other than by the envsphere, which has an exterior skyscape panorama on it.... I guess I should try making up some sort of room prop with just a window or two and / or skylight andsee how that differs?

I realise I need to try a lighting set up that creates more "chiaroscuro", in order to really test the use of IDL only lighting... after all, it's primarily this feature that I have still been using Poser lights for, on top of using IDL for ambient light.

What else am I missing on both these points regarding forcing the ommission of the IDL precalc??

Anyway, I brought the "shading rate" down to 0.2 and upped the "Samples" in the IDL section of the D3d dialog to 500 and that looks to be helping with those artifacts, far as I can see from the IDL pre-pass. I'll play around with that some more and post the results later today some point

Does mesh topology perhaps contribute to these shading artifacts, such as I got under the cushion of that chair prop?

Also doing a birds-eye view render of the light set up I've used to make this clearer....

Cheers

Cheers

This one is not really an advancement of ideas in this thread...just a render in appreciation of the Dreamland Models line of furniture, very cool. Shaders by Baginsbill. This is their newest addition, a lingerie chest. Magenta Starman Globe sold seperately.

::::: Opera :::::

monkeycloud, from my experience I definitely wouldn't go high with the Intensity. That's a way to make your light settings be completely unpredictable. I think I've rendered as high as 1.2, but normally I think you want 1. If that doesn't let the IDL spread enough light around, you can up the IDL bounces (although I would think with 7 you're going to be reasonable in that department).

Mesh topology does matter. If you have concave vertices somewhere, Poser definitely doesn't like that.

Attached Link: http://www.renderosity.com/mod/forumpro/showthread.php?message_id=3817130&ebot_calc_page#message_3817130

when IDL was introduced in poser 7, stefan said to use IDL intensity = 0.8, as this is a good fall-off value per bounce, giving a curve that is approx. inverse linear for large surfaces in an enclosed room.  however, if all diffuse values on all posersurfaces are <=0.8, then IDL intensity can be left at 1.0.

when IDL intensity is >1 and some posersurfaces have diffuse value 1, this causes IDL to multiply (rather than falling off) such that some areas that should be occluded become intensely bright.

there was a bug in earlier versions where intersecting orthogonal surfaces and narrow triangles converging on a polar vertex would show artifacts at IDL IC<100, but this may have been fixed in one of the last or next releases.

 Hi Monkeycloud

It had not occured to me that the Jaguar was a pet-I thought it was the Jag that had her startled!I guess Naked Punker Girls usually have Black Jaguars as pets.:-) Maybe there is a Giant Cockroach on the floor?

I thoought the minimum shading rate just affected the fine details- with a lower setting-more fine details can be seen. I think increasing samples-as you have done-should reduce the shadow artifacts at the line on the chair.

Overall I think the render looks very good indeed!

Quote - Yeah, for some clarification, here's a screen grab of my render settings, per the Dimension 3d Render Firefly script UI, such as I'm now going with...

 

Cheers

OperaGuy

That is some nice furniture and BB's shaders add quite the realistic appearance! I like the dynamic cloth clothing hanging out of the drawers![

quote]

This one is not really an advancement of ideas in this thread...just a render in appreciation of the Dreamland Models line of furniture, very cool. Shaders by Baginsbill. This is their newest addition, a lingerie chest. Magenta Starman Globe sold seperately.

::::: Opera :::::

Here's a bird eye view of my scene... this should help clarify a few things?

the nightgown is one of the marvelous freebies from carib98

http://www.renderosity.com/mod/freestuff/index.php?user_id=92438

they are great for animation too.

anotber of my renders with these nightgowns....

Are all of you that are able to skip the precalc at IC 100 using a beta of SR3 or something?  Because I just tried a IC 100 render overnight and the next morning it had stalled about 75% of the way through the Indirect Lighting Precalc...  (I'm still on SR2 btw)

Many thanks for the info guys.

I'm waiting for a render to complete, with intensity set to 1.0, at present... further to Believable's post.

All the shaders are BB's, effectively. So I should be safe with the diffuse I think... however I'll perhaps go with 0.8 here on in, based on Miss Nancy's excellent further info there...

Miss Nancy helped me with that....

use d3d's renderFF script to set IDL IC to 100.

see her post above.

Quote -   the nightgown is one of the marvelous freebies from carib98

http://www.renderosity.com/mod/freestuff/index.php?user_id=92438

they are great for animation too.

 

Yes, got that too... it's a great freebie

Quote - Miss Nancy helped me with that....

use d3d's renderFF script to set IDL IC to 100.

 

see her post above.

Yeah, you need to use the script.

But skipping the calc will not speed things up for your.  My primitive ball got about one quarter rendered in 96 minutes.  It's excruciatingly slow.

Hi All

Just to confirm - it is the LOWER Irradiance Cache slider on D3D's render settings dialog that you set to 100 to put Fireflyin in the brute force mode.  Then tweak the Samples as required.

In the brute-force case, there is no initial indirect light calc - and I'm not seeing the SSS calc output (it's black).  There is a substantial rendered difference (SSS wise) between IC 99 and IC 100 (100 has much less red).  I wonder if the SSS pass needs the Irradiance Cache to work properly?

Yeah, I tried it last night on a 300 X 400 test image. It skipped the IDL precalc alright but the subsurface took forever. Something more than an hour and a half later it was almost finished rendering. It was just insane.

Quoting myself from other threads due to lack of time.

IC is the source of artifacts in the white chair.

Quotes follow:

IDL with Irradiance Caching (IC) is a stochastic approximation. (Look up stochastic if it's not familiar. Without that word I'd have to say a hundred things.)

The more you rely on IC (for speed) the more error you should expect in the lighting. Since the error is stochastic, it's not the same error from one render to another.

Therefore, even if you're doing a still image, two renders in a row will not be the same.

IDL + IC contains artifacts, by definition. These artifacts will move in random ways, even if you change nothing.

It is possible to implement deterministic pseudo-randomness (repeatable, but still unpredictable), which would still have errors but they would not move around. I don't know why Poser is not implemented this way.

If you don't want stochastic approximation artifacts, set IC = 100. Prepare for very long renders.

IBL+AO is also stochastic.

The IC value represents the desired accuracy, which is inverse to how much the cache is used. When IC=100, you're saying you want 100% accuracy, which means 0% use of cached values. Lower values use the cache more, accepting less accuracy in exchange for greater speed.

When you lower the IC value, you're accepting huge inaccuracies, which predominantly express themselves as missing shadows in a lot of places. Which is another way to avoid shadows, right? So in a sense a very low IC does improve consistency, because it consistently fails to react to occlusion. This does not contradict anything I said, but does provide an alternative explanation to why you thought your results contradict what I said.

Basically you turn turn off small-scale occlusion when IC is set very low. If you elide all but the biggest and most obvious shadows, then yes it is very consistent. However, highly inaccurate.

Quote - Hi All

Just to confirm - it is the LOWER Irradiance Cache slider on D3D's render settings dialog that you set to 100 to put Fireflyin in the brute force mode.  Then tweak the Samples as required.

In the brute-force case, there is no initial indirect light calc - and I'm not seeing the SSS calc output (it's black).  There is a substantial rendered difference (SSS wise) between IC 99 and IC 100 (100 has much less red).  I wonder if the SSS pass needs the Irradiance Cache to work properly?

Strange things were also happening for me in the SSS pass, when I set IC to 100.

Well, I say that... all I mean is that the SSS pass output looked very different. So perhaps when IDL is enabled, SSS does rely on it??

Clearly SSS works without IDL, so any condition must only be active if IDL is on? (and no other lights are present perhaps).

Well, I say that... all I mean is that the SSS pass output looked very different. So perhaps when IDL is enabled, SSS does rely on it??

I have logged it as an issue with SmithMicro.

Based on cspear's graph, as posted above by Miss Nancy, an IC value above about 60 or 70 is not likely to render any faster than IC=100... am I interpreting that correctly?

I seem to be getting more black artifacts with higher Pixel sample / shading rate / IDL sample values, if I leave my IC at 20... and whilst there was a little improvement with the original shading distortions under the cushion of the white chair in my render, this approach hasn't resolved the issue altogether... and more black hole type artifacts are now appearing, scattered more randomly around the image.

So I may have to just bite the bullet and try again with IC at 100 I guess. Or perhaps look at the geometry of the chair mesh, to try adjusting what I think are the problematic polygons. I might try just subdividing it and see if that reduces or impacts the original distorted shadows at all...

Can adding AO nodes to your shaders, with "Evaluate in IDL" ticked, help to compensate at all for the loss of small scale occlusion when IC is set low??

Many thanks

Quote - Can adding AO nodes to your shaders, with "Evaluate in IDL" ticked, help to compensate at all for the loss of small scale occlusion when IC is set low??

Yes.

Many will tell you never to use AO with IDL, and indeed if used indescriminately it can ruin your renders, but used prudently it can work well.

See this post about the subject by Blackhearted who, in my opinion, tends to produce very nice renders :-)

Hmmm, I wonder then how Blackhearted has worked material AO into his SSS setups? I am very curious to see a screen grab of that. Snarly, I'm also awaiting your new EZSkin with baited breath!

Cheers Snarlygribbly

I'll experiment a bit more with some carefully placed AO nodes set to Evaluate in IDL then I think...

Due to the complexity of most of my scenes, exponentially adding to the render time, by setting IC to 100 and losing the IDL prepass, just isn't really feasible, for the most part.... aside from smaller experiments like my girl in chair with cat render.

So I think using more "artistic workarounds", such as careful placement of AO nodes, or indeed a Poser light or two, in addition to IDL, is probably where I have to continue to aim, for the time being...