Poser 8 Impressions so far

PROS

• I couldn't believe it, I was able to render four fully clothed characters in a complex scene in under 7 minutes.  The memory didn't blow up and I have a quad core + 4gb ram.  These guys just took a step in a professional direction.
• Getting poser 8 for the improved rendering speed is the ONLY plus so far that I've found.

CONS

• They need to fix small picture  icons in the library.
• The search feature doesn't work at all (maybe cuz I add runtimes from previous poser versions).
• I personally think the new IDL sucks.  Is there something I'm missing here?  I turn on Indirect lighting and leave the samples at 6. Is there another option somewhere to turn on Global Illumination because nothing in an IDL render resembles anything close to real deal GI ... and yes I know GI means indirect lighting i'm just trying to figure out if i'm doing something wrong, because IDL SUCKS!! lol
• Having the library & properties window undocked causes crashes when going to material room.

 I don't yet own a copy of P8, but from my understanding the included IDL is far from the sort of global illumination you would find in a higher end rendering package...

I would highly recommend you back read the other threads concerning IDL setup and evaluation.  Bagginsbill has been working closely with SM and users to resolve app issues and lighting.

Quote - I would highly recommend you back read the other threads concerning IDL setup and evaluation.  Bagginsbill has been working closely with SM and users to resolve app issues and lighting.

Yea BB rocks and the guys are really making a good effort.

But let's face it.  This is not real GI.  That's fine though. I'm happy with the fast rendering times.  That is a HUGE deal for me :)

I've seen written in other threads that the render setting screen doesn't include all the editable settings for IDL, there is an included script (scriptspartnersDimension3dRender FireFly) that has more options for your tweaking pleasure, I haven't touched it yet as I have no idea what I'm doing with the standard settings as it is.

Quote - I turn on Indirect lighting and leave the samples at 6.

This is close to minimum quality, of course it's going to look terrible.  Maybe the default should have been more towards the middle of the slider, but I'm just pointing out that the current default of 6 is basically "draft mode" minimum quality in favor of speed.

Quote - > Quote - I turn on Indirect lighting and leave the samples at 6.

This is close to minimum quality, of course it's going to look terrible.  Maybe the default should have been more towards the middle of the slider, but I'm just pointing out that the current default of 6 is basically "draft mode" minimum quality in favor of speed.

hmmm

I'll try kicking that up.

Raytrace Bounces also controls the number of bounces of indirect light (GI) that are calculated (currently, anyhow).  If you have it set to 1, you will get low quality results in favor of high speed.  Generally, a larger number of bounces will give you smoother GI lighting, true in any GI renderer.  Just be aware this also includes reflection depth; if you want these counted separately, so you can have small # of reflection bounces but a large # of GI bounces, look at that render interface script by Dimension3D.

A good point Bagginsbill made early on that may have fallen in the cracks is that, all materials throughout your scene really ought to have their Diffuse_Value setting reduced from 1.0 to something like 0.8.  Practically all Poser content has its materials set with Diffuse_Value at 1.0.  The reason is that, in the real world, as light bounces off of some material, some of the illumination is absorbed by the material.  With Diffuse_Value left at 1.0, the amount of light bounced is too strong and it doesn't fall off with successive bounces as it ought to.  With 0.8 or so, it gradually falls off in a more realistic way.

It's tedious to change all your materials settings on everything but it really should be done.  This isn't specific to Poser either.

Nevertheless, the current iteration is certainly useable to get to know it. Better to have something to practice and learn on than nothing.

Here is a test at Indirect Light Quality = 6, the default. This is clearly draft quality.

Note added: These images use only an IBL for lighting. The IBL is flat, with no image attached, so any small variations are magnified. This is a worst case scenario.

Normal lighting regimes with a directional light involved will provide superior quality at much lower quality settings.

IMO, this is a huge improvement in lighting. However the shader is blown out, and needs adjustment. It is a GC shader, and shader GC with IDL is not such a good idea.

Looking over my scene, I found that the ground was white. Now without GI, that really doesn't matter, but with GI on, this was producing an excessive amount of ambient light. I changed the ground to a dark gray.

Further, my scene is a studio with walls. IBL was there to simulate the reflection of light off the floor and walls. But with GI, I have the real thing, and the IBL is just pilling on, making altogether too much light. So I simply turned off the IBL, and let the GI take care of the ambient light bouncing off the walls and floor. So now I just have a single spotlight in the scene.

Finally, the GC was producing sRGB values for all the surfaces, which compensate for how luminance works on a screen. We want the output of each shader to be a linear luminance, not an sRGB luminance. So I turned off the output GC stage of my shaders. This was just a few clicks using VSS. I left the input stage anti-GC on the shaders. This lets the shaders anti-correct the incoming textures so that they are linear color space values.

Then I enabled Exponential Tone Mapping. Not HSV Exponential, but Exponential. As far as I can tell, the math here is identical to the final GC stage of Poser Pro. So my linear values coming from the shaders are finally converted to sRGB color space by Poser 8's tone mapping feature.

The resulting render is more realistic than what I started with. The work required was modest, but necessary. It was wrong to combine IBL (fake ambient light) with GI (real ambient light).

 "With new tools, we must use new techniques.
It isn't fair to just flip on GI, and ignore how that changes things."*

That is the MOST Reasonable Statement I have read yet today :biggrin:

bagginsbill   after i rendered anything in poser with IBL and AO i always had to use curves in photoshop to make a bigger contrast.

i see that GI in poser 8 does this for us. in your new render it loosk more alive and not so flat.

but you said Exponential. didnt you say some days ago that HSV is like GC? 

Quote -

for a render without SSS this looks pretty good.

There are two versions of Tone Mapping in Poser 8. Very kind of them to give us both.

HSV Exponential is not the same as Exponential.

We should use Exponential if we want to do full GC + IDL, combining anti-GC in the shader with GC in the renderer.

We should use HSV Exponential if we do not want to bother with anti-GC at all.

With shader-anti-GC + Exponential Tone Mapping + IDL we get exactly what Poser Pro 2010 will do.

My god, I have a headache already, and I don't even have the program yet.

Laurie

Quote - My god, I have a headache already, and I don't even have the program yet.

Laurie

You have a headache because my P8 render is so beautiful and you cry cry for not having good tools? teehee Just teasing.

You so Crayzzziiee

Tone mapping and gamma correction are two separate things.

Gamma Correction will leave white as white and black as black, leaving the overall dynamic range of an image as is and only affects the mid levels. Gamma correction is to compensate for the different color spaces in which the textures, the renderer and the output image live.

Tone mapping on the other hand works more like film - with brighter tones, it becomes less sensitive and thus compresses the dynamic range. Exponential tone mapping will, just like film, fade overbright colors to white. If you look at photos taken with a strong flash, bright spots get washed out to white. Now if you try the same with a render engine without tone mapping (just dump RGB clamping), you'll get some really ugly color fringes around a white spot. Turn on tone mapping, and it will nicely fade to white without those strange discoloration, behaving much more like film material.

HSV tone mapping is, quite frankly, a completely non-physical unrealistic hack that just happens to look really, really good in many cases. It's applying the same tone curve as the exponential mapping, but instead of applying to the RGB components of a color, it does it only to the brightness of the color and leaves saturation untouched, therefore preventing all washed out colors.

That in cases, HSV tone mapping  can look similar to gamma correction is nice, but mainly we wanted to have something to give users the ability to get great looking renders without having to worry about which textures are in what color space. The math behind the two is actually quite different.

BB, your middle pic shows exactly why everybody is crying.

They see the button click it, and ignore the rest.
They suppose ‘demand” that Poser should do the rest.

Your statement is so true
"With new tools, we must use new techniques.*
It isn't fair to just flip on GI, and ignore how that changes things."*

But, three basic settings should be included.
Draft, medium, and high quality, like the slider control on Auto Settings.
Introduce a slider control for Indirect light.
So a new user has at least an idea of what is happening.

Now, without help, and a new user that does NOT have the time to browse all the forums, (to find the briljant help here) there are so many options, possibilities that new users do not see the tree in the woods any more.
They are overwhelmed.

Click, click, try, shit, again, and again, and then they blame the program.
just an idea, and my 2 cents

Stewer, GC final output step is just a Pow function. So is exponential tone mapping, no? They both tone down the hyper colors. I looked up HSV Exponential tone mapping and built my own several months ago. It too is a Pow function, but it doesn't do Pow to each channel, only to the brightest, and then scales the others to match.

You've not published the precise formulas so I'm guessing, but the web published formulas I found for exponential tone mapping is identical to GC, but people write it differently. No?

Tone curve is a fancy term for function applied to luminance data. Exponential is a specific funciton, the Pow function, no?

Stewer, never mind. I just graphed the two functions - they aren't identical. However, they have the same overall shape.

Hmmmm. I'll have to noodle on this. Not sure it really matters though, because the shapes are so similar.

In GC, the heart of it is x^p, and in ETM the heart of it is 1-e^(-qx), where p and q are user-definable parameters. Theres more to the ETM equation, but that's the jist of it. The clear difference is whether x is in the base or the exponent.

Blue = ETM
Red = GC

Dang I should have graphed them a long time ago. Again, I saw similar output, and made an assumption.

The ETM curve produces more contrast than the GC curve. No wonder ice-boy liked it better.

Stewer, why did you only give us one of the two control parameters in ETM? We have Exposure, but not Exposure_Gain. Why?

Could you add the output GC to curve to Poser 8 as a 3rd option? It really sux to not have the final correction to sRGB or close to it.

Quote - > Quote - My god, I have a headache already, and I don't even have the program yet.

Laurie

You have a headache because my P8 render is so beautiful and you cry cry for not having good tools? teehee Just teasing.

That and I just learned how to use POSER 7's materials and lighting!!! lol ;o)

Somehow, I'm going to have to go through the forums BB and bookmark EVERY post you've ever made on materials and lighting. You've helped me so much already....thank you :o).

Laurie

 Bagginsbill, if you are using normal non-IBL lights with IDL, do you use HSV or Exponential exposure?

Quote -  Bagginsbill, if you are using normal non-IBL lights with IDL, do you use HSV or Exponential exposure?

I had an answer to that, but it was based on incorrect assumptions. I've been posting along about what I've learned and what I think, and sometimes I've made mistakes.

There are some subtle issues here I don't fully grasp yet, but at the moment I'd say HSV Exponential. The only way I see to use Exponential at the moment involves extra work in the shaders. HSV ETM is the way to go if you're not going to get into shader hackery.

That was what I have been thinking too... However, I've noticed that with non-IBL lights, if you use HSV exposure then you need to use a limited number of lights otherwise your characters will appear washed out and a lot of subtle details (reflection, ambient occlusion, and shadows) will start to drop out.

Quote - Blue = ETM
Red = GC

Dang I should have graphed them a long time ago. Again, I saw similar output, and made an assumption.

Are you sure that's the y=1-e^(-qx) graph? That function never exceeds y=1.0, your graph however does.
  > Quote - Stewer, why did you only give us one of the two control parameters in ETM? We have Exposure, but not Exposure_Gain. Why?

The gain parameter is neither practical nor useful. Since the exponential function never exceeds 1.0, there's no use for gain < 1. Gain > 2 on the other hand blows out really, really quickly, if you then lower the exposure to compensate, you end up with a linear function = no tone mapping. Play with if you like, it's exposed through Python and pz3.

I'm lost...what is the difference between exponential function and exponential exposure?

Quote - Are you sure that's the y=1-e^(-qx) graph? That function never exceeds y=1.0, your graph however does.

No, I'm not sure. I'll do it again. What you just said is correct - that function never exceeds 1. I'm starting to look stupid.

I'm going too fast today, aggressively multi-tasking, and not being methodical. I'm conversing in at least a dozen active threads and staying up to date on another 10 or so. I'm also tracking threads at dpreview about Nikon cameras and lenses. I get one or two new ebots every minute. I'm also reading and thinking about how to port a complicated web app from SQL server to DB2. I'm also writing the GUI for a digital signage configuration tool. And, I'm constantly running rendering test cases, to understand GI better. And, as can be seen above, I'm occasionally making and posting demo renders about Poser 8. In between all that, I'm having email conversations with two clients, and I had several in-person discussions in my office about a product for another client. Very shortly I'm going to start howling like my dog.

Stewer, can you write for me the exact formula, so I don't do any more stupid guessing and assumptions?

I've found that it's best to keep Indirect Light Quality set to at least 25, unless you're doing very fast test renders.

@ bagginsbill

The 3 renders of the female figure show the classic shredding of the edges of the render that I mentioned in a couple of other threads.

I may be wrong.... but I don't think anybody confirmed that SM were aware of the problem.

Do you know if this will be fixed in SR1..?

I'm pretty sure it will be fixed.

I'll be getting a pre-release of SR1 soon, and will be able to tell you what I see. That doesn't mean SR1 is ready. Just that a huge number of fixes are already in the queue for testing.

" Exponential,  HSV ETM, shader hackery, ETM versus GC,  x^p, 1-e^(-qx), where p and q are user-definable parameters"...Bagginsbill you're making me dizzy! You obviously know your stuff, but I agree with vilters..Poser definitely needs a few more buttons to press. :-)

Quote - With new tools, we must use new techniques. It isn't fair to just flip on GI, and ignore how that changes things.

I find that in any 3D application:  The more real a scene is, the more sense using Global Illumination is.

Well, Poser 6 and 7 forced me to learn new ways of doing things. I don't see why Poser 8 should be any different ;o).

Good thing we actually have people that know what they're doing so that they can guide the rest of us! lol.

Laurie

*"Good thing we actually have people that know what they're doing so that they can guide the rest of us! lol."

I eagerly await the detail rich tutorials. :)

LMK