Hey everyone,

I've asked something like this before, but just want to make sure it's not my machine that's being fussy.

I am in the process of a final render for a scene that I'm making in Poser, I really like the lighting effects I finally accomplished.  The run down on the scene is this:

About 30 different props all together, and about 6 lights.  I have turned up the settings on light mapping to 1024, and I upped all my render settings to just about the fullest possible 4 for Raytraces, almost full on IDL calculating, etc.  Can that take over 12 hours for a render? 

I no longer get the dreaded non responsive Poser error after upgrading, thanks to Bagginsbill, and John Csaky (I hope that's his last name) in the support center, Bill here in the forums.

I really like my lighting effects I finally figured out -- I now have individual lights lighting up the headlights of the car and parking lights using "Point" lighting, have it adjusted so it doesn't bleed on other aspects of the environment.  I even have the lighting for the environment to look "night" colored,  but the render times are driving me nuts!

I guess everything worth waiting for is worth the trade off. 

My question is this -- does anyone else experience the long render times for the best results on their images? 

Just curious to see what everyone else does. 

My system specs if helpful:  Dell Inspiron 531 2.88ghz Dual Core AMD processor 3 gigs of RAM, and 1gb GTS 250 by MSI Factory Over Clocked.  Windows Vista Home Premium 32 bit.  I plan on upgrading to 7 64 Bit after my business gets going. 

Thanks in advance everyone, you've all been very helpful to me here in the forums :-)

In addendum to above:

Now I realize I have some "post" edit work to do on the cop car, I have to add a badge on the door, which I'm sure is easily done in Photoshop, but hope the props in front don't get in the way. 

Anyone have any good post editing tips in that case? :-/

Don't use depth mapped shadows.  Ray tracing takes longer but if you're looking for the best possible render, it's the only way to go.

 IIRC, BB stated you shouldn't need more than 2 Raytrace bounces. 

If you are using IDL, make sure you set irradience cache down, not up.  It will make a large difference in the IDL pre render step.

Bill

consider adding the door badge as a tatoo. use a blender node and mask. Postwork is fine if nothing is in the way.

That's true on the Ray Trace bounces --- I'm not going to touch it now but forget if I did actually change it or not. 

Bill, with the IDL up though, doesn't that have better calculations for more accuracy?  That's what I am looking for, and hoping that's the case. :-).

Thanks Mark for the photoshop Tip -- I think the spot on the door is free where the badge goes, so I should be all set. :-)

 Sam, any reason why not to use Depth mapped shadows?  Just curious on that.  I thought Depth mapped would add more realism to the scene......

Well, it is not supposed to make a difference, but it does.  if you use the 'reneder firefly' script, you can get better results, because it has allows you to set both regulare cache and idl cache seperatly

scripts>partners>dimention3d>render firefly

Bill

Quote - About 30 different props all together, and about 6 lights. 

Prop count is not important - it's polygon count. I can render 1000 Poser boxes, but not 1000 Minicoopers. (I have 25 pieces of paper in my pocket - who has more money, me or you?)

Quote - I have turned up the settings on light mapping to 1024, and I upped all my render settings to just about the fullest possible 4 for Raytraces, almost full on IDL calculating, etc.  Can that take over 12 hours for a render? 

If you ran out of physical memory and are now swapping to disk (thrashing), it could take 12 weeks.

The phrase fullest possible is not consistent with 4 raytrace bounces. The fullest would be 20. I consider 4 to be the minimum for IDL, although for a draft and for specific situations, 2 can be enough. I usually want 6 for a quality render. Be aware that there are actually two different "bounces" settings in Poser. One for reflect+refract, and one for IDL. We often only need 1 or 2 for reflect+refract, but need more for IDL. Setting up more than you need for reflect+refract does not cost anything, unless you happen to have two bi-reflective surfaces interacting with each other. Then you'd want to deal with the two settings separately. They can be found in the D3D render firefly settings dialog found in the scripts menu.

You mention "light mapping to 1024" - I assume you mean shadow mapping, therefore you're using depth-mapped shadows. These are faster than raytraced and 1024 is also rather small. A "fullest" setting would be 2048 or perhaps even 4096, but can't remember if Poser will even let you do that. To get over 1024 you have to modify that parameter's maximum allowed value, and then there is an internal maximum you can't go past.

Quote - I really like my lighting effects I finally figured out -- I now have individual lights lighting up the headlights of the car and parking lights using "Point" lighting, have it adjusted so it doesn't bleed on other aspects of the environment.  I even have the lighting for the environment to look "night" colored,  but the render times are driving me nuts!

Sounds like you enabled inverse square falloff, which is a good thing for small light sources. But note that it doesn't add to render time at all. It just changes the equation used for lighting.

Quote - I guess everything worth waiting for is worth the trade off. 

Almost. And you must mean everything worth having - it's a source of endless argument in this forum as to what is worth waiting for, or what is worth putting effort into. Gamma correction is free. Inverse square falloff is free. These are things worth having, and you don't have to wait when you use them. As you talk to various people, some will be insulted, injured, or threatened by views that they fail to use features they should be using. As you read stuff regarding what's worth doing, keep in mind that realism is hard, and if you want it, great. If you don't, great. Some people don't feel that way, and they are utterly unable to let a discussion of difficult realism techniques go unanswered. When they try to tell you it's not worth it, tell them to shove it. I hate ugly fake looking pictures.

You, of couse, don't have GC with Poser 8, so you'd have to put more effort into using it. Sounds like you already put considerable effort into lighting and ran into some difficulty getting realism at night. That comes for free with GC.

Quote - My question is this -- does anyone else experience the long render times for the best results on their images? 

I often render over night. I never find it unfinished in the morning. In the past, I did quite often, but I have learned what render settings and other things will cause that. For example, if you use a very low min shading rate and lots of small feature displacement on a large polygon viewed from very close, you force the number of micropolygons into the millions. That sounds unusual but it isn't. A single large ground square with a pebbly texture using displacement and the camera close to the ground will use all your memory.

Once that happens, 12 hours isn't even a start. Could be weeks.

Quote - My system specs if helpful:  Dell Inspiron 531 2.88ghz Dual Core AMD processor 3 gigs of RAM, and 1gb GTS 250 by MSI Factory Over Clocked.  Windows Vista Home Premium 32 bit.  I plan on upgrading to 7 64 Bit after my business gets going. 

3 GB RAM is weak. You need that much for Poser alone, and with the OS using probably 1 GB, you don't have enough.

Your processor is unspecified, but I'm guessing it's 20 times less powerful than my I7 860, so your 12 hours is my 36 minutes, even if you're not thrashing to disk.

Quote - Don't use depth mapped shadows.  Ray tracing takes longer but if you're looking for the best possible render, it's the only way to go.

Yes.

Quote -  IIRC, BB stated you shouldn't need more than 2 Raytrace bounces. 

For reflections. That's not the same as for IDL. Sometimes IDL really should have 6 or 8 or 10. (If you want light to go around multiple corners in a corridor, for example.)

Quote - If you are using IDL, make sure you set irradience cache down, not up.  It will make a large difference in the IDL pre render step.

 

Bill

This is true - the dial is actually an accuracy dial. Setting it to 100 disables caching altogether and is most accurate and also most slow.

For some scenes, I can get away with 7, which is very fast.

No lights - zero - nada.

Environment sphere provides all the lighting.

Raytrace bounces = 2

Irradiance Cache = 50

IDL Quality = 7

Renders in a couple minutes on my I7.

Can you spot the differences? They are small in size, but enormous in terms of the amount of luminance being altogether wrong.

Bill,

Thanks again for that indepth detailed explanation.  I have one of the lower end processors, am aware of that, so I think (actually know) I'll have to upgrade. 

I have put TONS of effort into the night lighting effect.  Just out of curiosity, what is GC? If you'll pardon my asking, still relatively new at Poser and all these terms. 

I'm hoping it won't take weeks to finish this -- Im trying to get it done, it's going on 16 hours now of Render time :-O). 

When I do the quick and dirty render, works great for the testing of it.  It's now officially 3/4s done in the IDL set up, so I'll let this finish.  If it's not done by tonight, I hope it will be, I'm hoping it's worth it.  Then I'll full around with adjusting more of those settings you mentioned.

Glad I'm doing it right and rendering over night.  Thought that would be the best way to do it.

Wish there was a batch way to render, would be easier and be able to get multiple edits done. 

Would love to know what GC is -- not sure what you mean by that, and am curious on anything to help with night lighting -- planning on alot of night scenes in the future, now that I have the lighting where I like it in Poser, but if it makes it easier will be worth it :-)

GC = Gamma Correction.  Tons of posts, pros & cons, about it in the forum.

BB, when you talk about IC you're talking about the one in the general render properties or the dimension3d render settings ?

GC is Gamma Correction.  I don't know enough about it to tell you anything useful, though.

Ray Traced shadows are better than Depth Mapped simply because Depth Mapped shadows aren't accurate.  Even with a large shadow map they aren't anywhere near as accurate as Ray Traced and eventually you'll run into the physical limit set by Poser on the size of the shadow map.  You'll also start to lose out on any time gains you may have had because large maps take time to calculate.  All that for something that's no good anyhow is a waste of time. 

GC is gamma correction. In a nutshell:

Digital images express brightness of each color component using a non-linear curve. This is done so that shadow details can be represented better, and because historically it's how phosphors worked in television. All cameras, printers, TVs, and computer screens follow this rule. The characteristic parameter that describes the non-linearity is called gamma. Everything that follows the sRGB color standard assumes roughly a gamma of 2.2. Older Macs had a gamma of 1.8. (This is why Mac images from the pre-internet area, when viewed on a PC, looked like crap. Gamma 2.2 versus gamma 1.8.)

The rendering equations are all linear - i.e. if light reaching some point is half that of another point, the numbers representing that will be exactly half. They assume gamma is 1. In the world of gamma=2.2, 1/2 is approximately .73, not .5. This means that when comparing the RGB value 73 versus 100, the 73 looks half as bright. If you used 50, it would look much less than half as bright.

The fact that the linear rendering equations are being given non-linear image data for lighting and texures, and storing the results in non-linear image data for output, creates disparities. These disparities are most noticed in things that are dark - they appear too dark.

There are quite a few effects that create "dark". The most important:

• Inverse square distance falloff from point and spot light sources. Unless focused, light energy spreads out as it proceeds away from its source. Things farther away receive less light per unit area.

• Shadow and partial shadow obstruction of light.

• Turning away from a light source (curved surfaces) creating changes in energy-per-unit-area.

All of these effects are directionally correct either way whether you pay attention to gamma or not. Gamma does not make up seem down. If you increase a light source intensity, everything it reaches gets brighter. If you block the light source with another object, things get darker.

But overall, everything is wrong, and if you're paying attention to details, you notice. Failure to address gamma means things are out of balance, always. If you arrange your lighting carefully, the imbalance is small. But, if you're doing something really demanding in terms of dynamic range, such as rendering a night scene where some things are dark and others are really bright, then it becomes extremely obvious that something is wrong.

The solution is you either use the right equations, or you compensate somehow.

Compensations (also known as hacks) include adding more lights, boosting shader diffuse values, etc. Hacking involves guessing and testing and refining for hours. It means things don't work as you expect. The more demanding the scene (dynamic range, lots of curves, variations in light in different areas and directions) the more difficult the hacking becomes.

The sound and easy way to compensate is gamma correction. All incoming material (colors chips in the material room, and color maps from images) should be converted to linear intensities. This is called anti-gamma correction, since these things already included the gamma factor. Then all the lighting effects are applied, and the final pixel color is gamma corrected before being stored in the render.

Poser Pro does this automatically. Poser (regular) does not. However, the functionality can be largely duplicated in regular Poser using gamma-correcting shaders. I am a heavy publisher of such shaders. The most famous among these are my skin shaders, which most people recognize as producing the best results with the least amount of work on your part, and do not require that you buy Poser Pro.

Quote - BB, when you talk about IC you're talking about the one in the general render properties or the dimension3d render settings ?

For the numbers I quoted in this thread, I was using the general built-in render dialog.

Quote - > Quote - If you are using IDL, make sure you set irradience cache down, not up.  It will make a large difference in the IDL pre render step.

 

Bill

This is true - the dial is actually an accuracy dial. Setting it to 100 disables caching altogether and is most accurate and also most slow.

For some scenes, I can get away with 7, which is very fast.

I just want to pick up on this, please ...

If rendering with IDL enabled ( + using D3D's Render Firefly Script), lowering the Raytracing IC to a small(er) amount, but therefore incrasing the IDL IC, and perhaps the IDL Samples, is what would be suggested?

There are two irradiance cache parameters - one for AO (used with IBL), one for IDL - normally only one applies as using AO with IDL creates redundant shadows, but artists like to do artistic things like exaggerate the shadows, so I suppose it is possible. The two are combined in the standard dialog, but completely change meaning when you enable/disable IDL.

The D3D dialog separates them. Assuming you never use AO and IDL together, only one matters.

If you use the built-in dialog to change IC, you can then open the D3D dialog to see what it did.

The main dialog seems to directly set the ray-tracing (AO) IC value, and then a translated version goes into the IDL IC value. The translation seems to be 25 + (.7 * V) where "V" is the value you used in the main dialog. So IDL IC goes from 25 to 95 - you cannot set other values from the main dialog.

Again - to be clear - when rendering with IDL, and not AO, the first IC does nothing at all. When rendering with IBL+AO, and not IDL, the second IC does nothing at all.

Quote - Can you spot the differences? They are small in size, but enormous in terms of the amount of luminance being altogether wrong.

Side by side compairson, yes.  The dolphin comes out a bit better but that's the only real change I noticed.  If I didn't put them side by side, I probably wouldn't have paid enough attention to notice anything.

Thanks BB, for this detailed explanation.

Bill,

Just got home a while ago and had to thank you for your very detailed explanation of CG.  Much appreciated on that, and I am going to do your new settings now.  Was still pre-calculating indirect light for when I got home, that's rediculous. 

Saved your settings as a text file on my other computer to use as reference.  Woul you mind telling me what EVERYTHING is on your settings?  They look quite good with the different pictures, the first is really excellent.

Thanks again for everyone's assistance here in these forums :-)