Some good points and excellent examples. My contribution - V4, Poser render and finished image (cropped from original). Hopefully she's got a bit of character ...

Would have been good if the OP had come back to clarify whether they meant realistic faces or just taking V4 away from that default look. Quite different things - Jessica Rabbit is a character!

wimvdb - love that render - is that V4 in Octane?

 

oops - xpost! I see that it is ....

Attached Link: Groboto

My understanding is that Poser doesn't like long thin polygons of the kind shown in your 1st screenshot - a model that contains these will have shading issues. So boolean operations that result in this kind of topology will be problematic but booleans that have a different topology will work fine.

I found a useful and cheap modeling app a while back called Groboto (see attached link) which is based on using primitives plus boolean operations. I've found the interface takes a bit of getting used to, but it's quite addictive, can make very complex shapes and can auto-generate quad meshes with pretty good (though not perfect) topology at varying resolutions. The helmet in the attached pic was 90% modeled in Groboto - some spheres and other default primitives plus several boolean operations. It also auto-generates UV maps based on the 'patches' formed by the primitives. Might be worth a look?

The zdepth mask has very jagged edges in my experience. I use semidieu's ARS2 for masks which I've found gives a smoother result, but I still find issues. Sometimes making an alpha for the affected area can help with cleanup in photoshop.

DoF at render is superior for this reason IMHO, but I've found that 16-20 samples is best for smooth results and even with the speed boost in PP2012 it can be very slow.

Attached Link: BB's Light Meter

The best way to judge an image is from it's histogram, something that the vast majority of image editors will have (Poser doesn't, unfortunately, but then it's not an image editor). It will display the RGB values for each pixel in the image graphically, and is completely independant of hardware or subjective opinion. Each pixel has a value between 0,0,0 = black to 255,255,255 = white. The histogram for the image you've posted shows that all the data is way down compressed into the far left so the image is too dark and without contrast. Ideally you should have data spread across the range without slamming into the far right or left (called clipping and results in detail being lost) - exactly how it's distributed depends on the image though.

The best tool I know of for judging lighting intensity in Poser is Bagginsbill's free Light Meter (see link). This will tell you if your lights are too weak or too strong very easily. Depending on which version of Poser you're using there are also useful tools like gamma correction and HSV tone mapping which can help.

Images displaying consistently across different setups is an issue. The other valuable thing that can be done by a lot of image editors is to save an image with a colour profile, the most useful one for web posting being an sRGB profile as that is common to just about all display devices. In effect it tells the device how the values in the image should display, so if you have a good range of values without clipping on the histogram, plus an sRGB profile, you've done as much as possible (without buying a hardware calibration device anyway) to ensure that what you see on your screen is going to be consistent on other screens too.

So yes, check your calibration, but I'd suggest trying that Light Meter too 

Sorry - what I was trying to say was that the histogram measures data irrespective of how an image appears on any monitor. I have a calibrated monitor (i1 device) and to me the render appeared underexposed. The histogram (attached) confirms it - sure it's a black horse, but it has a white blaze and socks, with the strongest light source coming from slightly behind so I would expect the edge highlights there to be approaching white.

Don't mean to come across as pedantic, but I know a little about correct exposure, and a low-key image should still have a full tonal range or it'll look technically wrong ...

Worth having a look at the histogram for the render in an image editor - the dark areas aren't clipping, but there is a lack of contrast. The highlights are barely above mid-gray. Such a compressed tonal range makes the image appear much darker and less detailed than it actually is.

This is going to enhance skin so much! It adds a subtle but powerful level of realism, I'm looking forward to seeing more results - thanks for these renders zev0!

That does make sense, and I can see the benefits to editing from your workflow (and it would be simpler if Poser supported transparencies!).

Really looking forward to this, the potential is amazing - would love to see more renders in the near future if possible ;)

Zev0, this is awesome work - I'm sold regardless of price! This will add another level of realism to figures, and I'd love to see a merchant resource kit too. There is nothing like this out there right now, and good reference is very hard to find, I've looked a few times, so even just for ref this will be sooooo useful.

Interested in your workflow too - have you tried exporting a 16-bit displacement map from Zbrush and using this both for displacement/bump and as a mask to add colour to an existing texture at all?

From the PP2012 manual - "SSS only works on objects that are in direct view of the camera. Objects in reflected surfaces will not be calculated with proper SSS". So I did a quick test with blurred reflection on a sphere prop and a reflective lens prop, and when rendered the sphere reflection is perfectly sharp. So I would assume that what is true of SSS is also true of blurred reflections and 3D motion blur.

Given that Poser is a biased renderer and doesn't do physically accurate calculations in the way that unbiased renderers like Lux or Octane do, I would call this a limitation, not a bug.

 

Great thread. Haven't played with light emitters before so think it's time I started ....

My bad, could have put that a little better :)

Can't even blame alcohol .... just over-excited. Shae has been made by an artist as a morph of a figure that seems to be generally regarded as not the best that could be achieved, so the prospect of a figure designed by an artist (who from the wips posted looks like having the same level of talent) from scratch .... 

Enough to make a non-artist like me totally incoherent.

 

Looks like this is going to be an exceptional figure and thanks for sharing your wips. I'm enjoying using Anastasia and especially Shae (as above - with Irina's shorts and boots), so to have a figure designed by an artist of your calibre from the ground up promises to be rather special.

Had a thought - how about using an ultra low-res version as a proxy for dynamic collisions? So a user could pose Lucas, load and conform the low-res figure then use that in the cloth room as the collision object for the sim, then parent the resulting cloth to Lucas and delete the proxy figure. The morph tool could be used to clean up any poke-thru, and given that the proxy wouldn't need details like individual fingers/toes, teeth, eyes etc it could be very light in terms of poly count which should result in faster calculations.

Another thought - would it be possible when rigging the figure to use bones in the face to drive basic expressions? I'm thinking of things like mouth open, brow raise/lower/squeeze etc, and then use morphs to add the detail? Have no idea how much work that would involve though or if it's practical - I remember diogenes talking about those possibilities for rigging somewhere though.

Having high and low-res versions sounds great - I use PML to go from Poser to Zbrush and back, so from my point of view 130k of well-laid out polys sounds perfect ....

Jumpsuit-type clothing? Have a look at Tool Boi for M4 at RDNA by Badkittehco. There was a V4 set along these lines at Daz by the same vendor called Welder Girl which was also great, but it doesn't seem to be there anymore.