Yes simulations are getting faster, but there are by no means lightening fast, and the more you ask of them, the slower they get.

Drop 2 characters in C4d, both with dynamic hair and clothing, and then tell me how fast a render you have.

Yeah, I'm doing a number of cloth simulations in Max, for a piece that will end up a poser conforming item.

There is more progress info and step by step updates and comparison to Poser dynamic cloth in this thread: http://www.contentparadise.com/forums/showthread.php?t=6151.

The reason I'm running them in Max is because it will handle folds and various more complex geometries that poser won't do at all. Each pose takes one overnight run of calcs (on my home computer, I'm not using office stuff for this.)
The main motivation for putting things into max was the ability to do more. I never compared max vs. Poser speeds on identical items.

Quote - Heh, Xeno, your post and good part of this thread reminded me of some philosophy and anthropology reading I did couple years back where it was said that men arguing about politics or business moves is a male version of girls talking about relationships or other 'more female' subjects.

I don't know all of the details -- but I"ve just partially heard a news story about the results of some study or other -- apparently directed by a woman researcher -- which has concluded that women on the average really do prefer the color pink and that men on the average really do prefer the color blue.  Then they go into all sorts of bizarre reasoning as to why this is true.......OT, but interesting.

Yes.......I understand that detailed speculation can be fun.  I do it myself.  

The ef guys probably read threads like this -- roll their eyes and elbow each other.  *"Get them.......hee!  hee!  hee!"  * :lol:

About the only way that I've ever found to speed up computations of any kind is to get a more powerful PC(s).  IMHO, hardware has far more to do with speed than software does.  At least most of the time.

I think they're all games of honing skills... for men it's usually skills that will allow them to improve their hunting and bringing home the bacon... For women it's usually the skills that will help them in a family environment...
And then there are always exceptions (like me) which don't fit either side all that well... LOL

Well, I do find little more amusement in speculating about relationships and psychology of interactions then about politics and business strategies... So, I still have a bit of a  'female slant' there ;)

Yes, dvlenk, I've been perusading myself of that. just go for speed and good general shadows with the Max scanline render engine, as fast as possible for animation. Then apply these desaturation and grain effects in AfterEffects. Voila!

Can anyone answer this: can a render engine take advantage in any way of grayscale over rgb? In otherwords, can it be triggered into dealing only with shades of black/white and thus process faster? Might seem trivial, but to an animator the saving of a few seconds on a frame can turn out to be significant. I fight for every second.

::::: Opera :::::

Quote - Yes simulations are getting faster, but there are by no means lightening fast, and the more you ask of them, the slower they get.

That's why we now have dynamics hardware:

AGEIA

(for games but they work just as well for 3D)

As well as research on doing dynamics and other calculation on display cards:

GPGPU
CUDA

We live in interesting times... :P

Quote - So, I still have a bit of a  'female slant' there ;)

We guys would be lost without that slant in our better halves.

Quote -

That's why we now have dynamics hardware:

AGEIA

(for games but they work just as well for 3D)

That is indeed interesting.  I"ve heard in the past that top-end gaming cards aren't the best for top-end 3D work.

Quote - Can anyone answer this: can a render engine take advantage in any way of grayscale over rgb? In otherwords, can it be triggered into dealing only with shades of black/white and thus process faster?

That's an interesting idea.... The JPEG file format does something similar by using the fact that humans are much less sensitive to color then luminosity to get high compression. Theoretically it should be possible to do the same for a render as well. I dont think any render engine can do this though. Better get a patent for the idea quickly before someone goes and creates one LOL.

Quote - That is indeed interesting.  I"ve heard in the past that top-end gaming cards aren't the best for top-end 3D work.

It's supported by a whole bunch of 3D companies, including Autodesk, Softimage, Cebas etc so they must be doing something right ;)

AGEIA Partners

Quote - Yes simulations are getting faster, but there are by no means lightening fast, and the more you ask of them, the slower they get.

Drop 2 characters in C4d, both with dynamic hair and clothing, and then tell me how fast a render you have.

Yes, I'll agree that those hefty polygons will drop performance in C4D (or Maya et al).  But unlike Poser, there are options to improve the calculations.  One old standby is use of proxies in place of the hefty geometry for collisions - a major part of the calculations by far.  It can be done in Poser perhaps using a low-res conformed figure or individually parented props - but as far as I know, Poser uses the figure geometry directly.  No idea how to use proxies in Poser for sims.

My plugin now has proxy creation support for just this purpose.  It can use primitives (per body part) or reduced polygon meshes (figure or body part).  The one thing that I have not yet had success with is making a reduced mesh proxy that bends with the figure - reassigning weight maps is not so easy as at first thought.  But reducing a 66000 polygon V4 to 6000 polygons in proxy definitely speeds up calculations while retaining a believable silhoutte for the collisions.

I'll just say that other solutions have better interfaces and are more flexible (no pun) than Poser's while also using more up-to-date algorithms.  These guys are continually improving their own implementations.  I don't know if e-f has to wait for improved support from the originators or has direct access to make such improvement (i.e.: the actual code).  I just did a cloth sim to show gravity pulling a cloth onto and then letting it slip off an object in C4D - yeah, I did that including a 240 frame QuickTime mov file in between posts here ;P.  With the cloth plane at 200x200 polygons (40000 polys), it did take a few seconds per frame step in the calculations - but at 50x50 (250 polys), about four frames per second.  Here's the mov:

http://www.kuroyumes-developmentzone.com/private/cloth_slip.mov

Robert

Quote - It's supported by a whole bunch of 3D companies, including Autodesk, Softimage, Cebas etc so they must be doing something right ;)

AGEIA Partners

Great......that's one to note.  Thanks for the information.

These hardware physics simulators are going to be extremely useful!  My new computer is going to have a GeForce 8800 GTX which includes the Quantum Effects physics engine.  Now if only Maxon would get in gear to support it! ;)

Quote - > Quote - Yes simulations are getting faster, but there are by no means lightening fast, and the more you ask of them, the slower they get.

Drop 2 characters in C4d, both with dynamic hair and clothing, and then tell me how fast a render you have.

Yes, I'll agree that those hefty polygons will drop performance in C4D (or Maya et al).  But unlike Poser, there are options to improve the calculations.  One old standby is use of proxies in place of the hefty geometry for collisions - a major part of the calculations by far.  It can be done in Poser perhaps using a low-res conformed figure or individually parented props - but as far as I know, Poser uses the figure geometry directly.  No idea how to use proxies in Poser for sims.

My plugin now has proxy creation support for just this purpose.  It can use primitives (per body part) or reduced polygon meshes (figure or body part).  The one thing that I have not yet had success with is making a reduced mesh proxy that bends with the figure - reassigning weight maps is not so easy as at first thought.  But reducing a 66000 polygon V4 to 6000 polygons in proxy definitely speeds up calculations while retaining a believable silhoutte for the collisions.

I'll just say that other solutions have better interfaces and are more flexible (no pun) than Poser's while also using more up-to-date algorithms.  These guys are continually improving their own implementations.  I don't know if e-f has to wait for improved support from the originators or has direct access to make such improvement (i.e.: the actual code).  I just did a cloth sim to show gravity pulling a cloth onto and then letting it slip off an object in C4D - yeah, I did that including a 240 frame QuickTime mov file in between posts here ;P.  With the cloth plane at 200x200 polygons (40000 polys), it did take a few seconds per frame step in the calculations - but at 50x50 (250 polys), about four frames per second.  Here's the mov:

http://www.kuroyumes-developmentzone.com/private/cloth_slip.mov

Robert

Hmmmmm..... The proxy idea is something that could be done. Nerd did very low poly covers for the hands and feet that kept the hair and cloth from getting caught between the fine mesh of the fingers and toes. Maybe something like low poly conformers with some adjustable parameters like scaling? The more universal something like that could be, the more likely it would be to be used... Something to put on the 'Ask 'eF for' list!

Quote - Yes, dvlenk, I've been perusading myself of that. just go for speed and good general shadows with the Max scanline render engine, as fast as possible for animation. Then apply these desaturation and grain effects in AfterEffects. Voila!

I was using the scanline renderer a lot (for animated camera walk throughs) before I switched over to real-time for that. I don't use it so much anymore, except to pre-render radiosity for the real-time GI.
It makes nice shadows, and even raytraced ones are real quick. As I said, avoid the adv. lighting if you're after speed.
There is a 'film grain' render effect that it can use. It's an RGB noise a lot like the gaussian noise filter in Photoshop. Maybe that would work for you, when you desturate it will wipe out the noise color saturation too.

Quote - One old standby is use of proxies in place of the hefty geometry for collisions - a major part of the calculations by far.  It can be done in Poser perhaps using a low-res conformed figure or individually parented props - but as far as I know, Poser uses the figure geometry directly.  No idea how to use proxies in Poser for sims.

It's not a common practice in Poser as far as I know, but in principle it's pretty easy if you wanted to take the time to set something like that up - you could just parent a primitive (e.g. cylinder) to a given body part that you want to be a cloth collider, x/y/z scale the primitive so it roughly fits that body part, and let the cloth collide against the primitives instead of the body.  Set the primitives to have a transparent material or some such to keep them from being visible in the render.  As the figure animates, the primitives will follow their parent body part.  It never occurred to me to try this approach when I used Poser heavily, and now that I no longer do, I don't have a lot of interest in trying it.

I have to agree with Kuroyume though that dynamic cloth and hair simulate FAR faster in C4D than they ever did in Poser, even when using the same polygon counts.  Render speed is a bit hard to judge, but I'll say for sure you can get much, much higher quality renders out of C4D's Advanced Render than you can get out of Poser - there are things Poser just does not do at the moment, e.g. area lights, area shadows, GI.  I consider the implementation of GI that's built into Poser 7 to be very much beta, and it's already noted to be extremely slow (it's not even officially part of the product, so that's not really a slam against Poser).  When it comes to GI, we're talking minutes versus hours.

I've about gotten to where I don't use Poser for rendering -- other than for testing / trial purposes.

Here's hoping that Poser Pro has a good renderer.

Most of you are probably aware of this, but in the fall of 2004 an engineer/artist, Kirwyn, worked up a proxy system for Poser dynamic hair. He has posted a version for various subsequent versions of popular models every since.

His V4 version is with the others...in freebies....
http://www.renderosity.com/mod/freestuff/index.php?user_id=19480

I have not tried this concept, but I have heard it works well. 

:::::: Opera :::::

Quote - I've about gotten to where I don't use Poser for rendering .

...............   Whaaa!!

When a 3D mesh becomes a rightously defended personification of your personal desires
It has become an IDOL and YOU have become its worshipper,slave and servant
-Tain-

Hell yeah! ;D

Collision detection is based on the object surface(s) - which are usually polygonal.  Put simply, the more polygons involved, the longer the algorithm takes to determine where the collisions occur.  If you can provide less polygons to a collision detection algorithm, all the better.  Some solutions allow for 'primitives' (mainly spheres) because they can compute the collision based on simpler math - a procedural sphere is far faster than any number of polygons in a closed object.  This is exactly what C4D allows for if you can use a sphere (proper: ellipsoid) as a collider proxy or direct collider - and is good for a simplified head representation for instance.

This is exactly why I went for the reduced polygon mesh option in my plugin - you get a good approximation of the high-polygon mesh with much fewer polygons (user set reduction percentage) to increase dynamic calculation speeds.  Procedural proxies are great but in close up situations requiring more accuracy you can see the procedural shape (as the hair bounces before hitting the actual mesh, for example).  Along with proxies is limited regions.  If you are doing somewhat short head hair, you don't need to include the legs, right?  If you can only include the polygon regions that will be in the range of the hairs' contact, you can increase simulation speeds.  In C4D, one can use polygon selection tags for this purpose.  In Poser, you may be able to specify groups which would be very helpful - need more info on that possibility.

The basic premise is this: how small a region of an object is affected by the dynamic object calculations and how simple can you make that region's representation to get the desired results asap.  I'll give that Poser's dynamic cloth and hair may be just as fast as other applications - if one could apply this premise to the dynamic calculations.  I'm sort of surprised that the developers hadn't considered an auto-proxy solution - considering that they succeeded with dynamic SPD for displacement and such.  Real-time LOD is a quite well established form of mesh reduction (hint). ;)

Quote - While you all have good points, I am in the Dale camp....I think it might NOT be FireFly. 

Who owns/develops FF? Making it 64-bit, render in background and netrender savvy...these are substantial upscale shifts. Is E-F responsible and capable of a substantial revamp of the external and/or owned-by-someone-else FireFly? Dale, how do you know these features have already existed and are just now being 'turned on?'

And they may be saving the 'brand' of the render engine for a PR bomb, indeed, like Dale said. In other words pjz, yes they are going to make significant hay out of it, but the mower has not entered the field yet. Further evidence: if it is FF, why are they saying "a" 64-bit renderer, not touting "a completely transformed FireFly making it a world-class engine" or something like that. I don't think the true puffery has begun, yet.

On the other had, the bit about switching from REYES (render everything you see, right?) to raytrace...that sounds ominous. Yet...what is the explanation of the current ability of FF to render raytrace and the existence of a nascent GI stub? 

I don't -know-, as in have facts. Those were based on the behavior that has been observed since eF took over Poser development (and just before, actually). There were shader nodes in P5 that didn't get enabled properly until P6. P6 apparently had the basics of the animation layer, which wasn't switched on until P7. P7 is the first version to have a decoupled, stand alone render option; and here comes P7Pro, which has a standalone renderer and network rendering....which would require a decoupled renderer if you intended to do anything else while rendering on the network. This seems to be eF's modus apperandi; place new tech in a release in an inactive or disabled state, see if the users's systems accept the stubs, take any wierd bug reports and check to see if the new tech is responsible, then switch the 'new' feature on in the next release. Using that as my basis for conjecture, then the questions they ask about features could very easily be indicators of what is to come. I'm also keeping in mind that eF has a good working relationship with eon.....who has apparently been getting some tech infusion from unnamed sources. And the FireFly renderer is already a bastard, being a REYES engine with raytracing functions. So if the basic raytracing architecture is there already, it should be expandable, true?

Quote -
Can anyone answer this: can a render engine take advantage in any way of grayscale over rgb? In otherwords, can it be triggered into dealing only with shades of black/white and thus process faster? Might seem trivial, but to an animator the saving of a few seconds on a frame can turn out to be significant. I fight for every second.

::::: Opera :::::

If you're using bitmap textures that have RGB color information in them, then forcing the renderer to turn it to greyscale is going to require more processing time, not less.  You can turn a color map to greyscale using shader tricks, but again, it's more process intensive than just rendering out color information.  It may not even be noticable in time per frame however.  If you start with greyscale bitmaps, or shaders without RGB color info, then you're probably saving on RAM because of the smaller memory demand.  But I would think any minimal gain in render time would be lost in preparing a scene to do this anyway.  You are much better off to just render normally, then convert the scene to greyscale in post.  You will also have more control over the end result by doing it in post.  3dsmax has a maxscript that can desaturate your frames right in the renderer, and there's free scripts and plugins that can desaturate based on z-depth, object ID, material ID, whatever you want.  Using them actually adds to render time, but gives you unlimited control, even deeper than processing the frames later on in some post software.

My suggestion?  Render everything in passes.  Max g-buffer allows for some very deep g-buffer pass control for compositing together later.  You can render entirely seperate passes for lighting, shadows, z-depth, atmosphere, diffuse surfaces, blend surfaces, specular, reflections, self-illumination, and even velocity (for adding post effect motion blur).  Check out the render elements in the render dialogue.

You guys hoping for a brand spankin new render engine, you ought to hope for a rebate for $50,000 to be included in every purchase - either is as likely.  Nothing wrong with hoping, more power to you, but there's simply no evidence to lead any reasonable person to suspect that.  Look at how much bragging was done about "universal" poses and multiple undo in P7's pre-release marketing.  You don't think something as huge as "we're adding GI" or "we're switcing to VRay" would merit a little heads up?  People would crap in their pants over news like that.

Quote - In C4D, one can use polygon selection tags for this purpose.  In Poser, you may be able to specify groups which would be very helpful - need more info on that possibility.

You can use polygon selection tags in C4D for defining what hair collides with?  I need to investigate that a little closer obviously, I didn't realize that but now that you mention it I think I remember where it would go (a field inside the Collider tag properties).  In Poser one defines collision for cloth by picking from the scene hierarchy, which boils down to basically the same thing (collide against only head/neck, basically polygon selections).  I believe Poser's dynamic hair works the same way, although I'm not 100% certain.

Just as a 3rd party spectator, I highly doubt Chaosgroup would bother creating a plugin version of Vray for Poser, at least at this time.  3 reasons come to mind immediately.  1) Vlado is already busy getting the renderer to work like it does in 3dsMax with other apps like Maya, Rhino, and soon Cinema4D.  They also have a Truespace plugin, and a standalone version is said to be in the works too.  2)  Chaosgroup is known for their extremely helpful community and forums.  I think the last thing they would want or need right now, considering all the new formats they're supporting, is to be flooded with bewildered Poser users who have absolutely no experience working with a high-end raytrace renderer.  3) Poser's shader structure is based on the renderman-compliant shader system, if what I'm hearing is true.  A new shader system would have to be integrated into Poser or some kind of conversion system would need to be implimented, and I simply don't see this laborious task happening any time in the near future.

Far more likely, if there's any change at all to the renderer, it would be that they are adding some more advanced controls to the Pro version, and exposing more features for advanced users to tweak and fiddle about with speed and. quality parameters.  If GI is exposed in the render features, this alone will require a new set of advanced parameters in the render dialogue.

Ok, you are all probably right. I just saw trueSpace get VRay for a list price to licensee of $300 extra, and in the ChaosGroup forums they are saying Caligari did most of the work. I also saw the language 'a' 64-bit render engine instead of "a far more powerful FireFly". I am letting my five bucks ride and you all have to give me 1000-1 odds.

Meanwhile....who owns FireFly? Is it used in any other applications? Would e-f be doing the upgrades, or the owner/developer of FireFly?

::::: Opera :::::

**devilsreject,
**thanks for the info. If I thought it would really advance render time I would indeed desaturate all texturemaps and remove shader color info, thus going all greyscale.

However, if there is an option for more control, precision and effects by rendering normally, I'd have to measure the trade-offs, speed vs quality.

I will be looking into the multipass scheme you outlined. Thanks.

::::: Opera :::::

Isn't Firefly a scaled-down version of Pixels3D's Tempest renderer?

Well, we know that the Material Room is Pixels3D all the way! ;)

Quote - Well, we know that the Material Room is Pixels3D all the way! ;)

A-ha, then I imagine they based the renderer on Tempest as well, being that seems the most efficient thing to do.

Yes, a quick google search brought me to this page:
http://www.keindesign.de/stefan/poser/firefaq.html#10

The renderer is a "sibling" to the Tempest renderer.  The shading system is akin to ShaderMaker Pro.

Edit:  apparently, this information is no secret.  It's stated clearly on the box and documentation in Poser that they licensed the Tempest engine.

Quote - Whaaa!!

Didn't mean to make you cry...........

Nope: these days it's usually Vue 6I that I use for rendering out Poser scenes, other than for checks & testing.

IMO, Poser's two main weaknesses are its render engine and the fact that Poser doesn't play well with certain other apps.  But I'm hoping that Poser Pro will go a long way towards addressing both of those issues.  We'll see.

Just out of curiosity, what do people find wrong with Poser's renderer?  I just found that it's based on Tempest, and further reading into Tempest seems to indicate that it's a pretty solid reyes renderer.  Aside from having more control over the render process, I don't see why so many people would be unhappy with this.  This kind of renderer, at least the technology it's based on, is production-proven.

The main problem is speed.  C4D's renderer is considered very fast.  Maya, XSI, 3DSMax are at about the same speed.  External renderers are superb.  Vue's renderer is molasses - but it does support such cool features as Ecosystems (very nice).  Poser's implementation is aggregiously slow compared to most other 3D apps - sorry.  Firefly might be a great render engine, but Poser's implementation calls for coffee, a second hobby, some free reading time, gardening.

The problem is that for the results that you get in other apps in minutes or hours, Poser consistently takes hours or days or weeks or months.  Note that there are no 30 minute high-quality animations from Poser.  Oh, you can get basic crappy animation renders in that range (~54000 frames - 30 minutes * 60 seconds * 30 fps) in a week or so.  For a quality animation render like that in Poser, expect to be rendering for some months!  Poser Pro (v7) does seem to
offer a means to do this in days - which is good!  Can't complain about that.

Again, I believe, you get what you pay for.  You spend $5K on a top-notch app, you get high quality rendering engines.  You spend $300, you get a good render engine that takes a long time.  You can't expect Maserati performance from a Toyota Celica!

IMO it's not the speed, it's the various huge render errors and unexpected behavior with lighting and shading, some of which I detailed here:
http://www.renderosity.com/mod/forumpro/showthread.php?message_id=2976801&ebot_calc_page 

Poser's implimentation of the Tempest renderer, AKA Firefly, may very well just be simplified to death, because they know the majority of their user base doesn't like to be burdened down with complicated parameters they can tweak.  Obviously this doesn't please advanced users.  I expect a Pro version to be much more like Tempest, which I never really used myself, but have seen screenshots and read discussions about it.  Pixel3D users seem happy.

If the vast majority of users who export their scenes to other highend apps do it mostly for the faster rendering, then that explains why Poser is adding the export options to output scenes into highend apps, and serves to futher convince me that they aren't switching to some other kind of render engine natively.  They'd have to ditch the majority of the shading and rendering technology they already have and re-code everything new, or invent some kind of conversion script, which still wouldn't be perfect.  Seeing how quickly they announced this Pro release after the version 7 release, I can't even imagine that's the case.  If it is, expect huge problems in the way of bugs, and some people's heads to explode here in the forums.

In the past, whenever I've had issues with a given Poser scene -- I'd estimate that approx. 90% of the time the problem cropped up at the end of the scene-creation process: during the final rendering stage.  This can be quite frustrating.  You've spent a considerable amount of time in tweaking a scene until it's just so.......and then the render dies about halfway through.  Or else the render won't even begin.  Or else -- as I've had happen also -- the scene will render out to 95% complete, and then stop cold & refuse to go any further.

None of these things make for a pleasant experience.

I need to add this caveat: it's likely that 90% of Poser users will never see any problems of the kind that I've listed above.  I tend to do scenes with multiple Mil figures and lotsa background props.  A lone V4 with hair & great shaders typically shouldn't take very long to render.  But throw in a house, cars with drivers visible on the street outside of the window, plants, other houses, several other Mil figures in the room, furniture, etc., etc. -- and then Poser's renderer starts to yell "Uncle!".  Even on a PC with plenty of system resources.

I support Poser and ef -- and I think Poser is great.  But I will readily acknowledge legitimate limitations.  And Poser's rendering capabilities have limitations.  All softwares do, of course -- but Poser starts to hit its limits at a lower level than some others.

Once again: Poser Pro might serve to fix that.

I have to hit the hay, folks.  Y'all have a good one!

And thanks again for all of the great information that many of you have shared here.

I think those who constantly thrash FireFly must be challenged for unfair comparison. I think they do not fairly compare apples to apples.

Many FireFly deprecators believe that good rendering is heavy raytrace and GI of architecture, cars, planets and environments, creatures, machines and landscapes. They are solid in their conviction that if a render engine does not perform these well, it sucks ipso facto. They are blinded by photo realism as if it is the be all and end all of render style.

What if these critics were to yeild to what Poser does well, the simple portrait or body study of a human model with a non-complex background? I think you begin to level the playing field.

I went to the Maxon gallery. I found very few such intimate renders. Here are the only ones, after viewing 160 or so total images in 23 gallery pages
http://www.maxon.net/pages/gallery/pix/waiting_in_the_fog.jpg
http://www.maxon.net/pages/gallery/pix/gallery9/blasi_puyi.jpg

plus a few of the 'gentler' room illuminations such as
http://www.maxon.net/pages/gallery/pix/gallery13/mueri_zivilisation.jpg
http://www.maxon.net/pages/gallery/pix/19/serteco_country.jpg

I did not travel over to CGSociety to look for Cinema4D renders there.

Now I have no idea how much postwork went into those four renders, nor what the render time was. I submit to you that Poser7, with FireFly, can go head to head in that kind of shot with no raytrace or GI engaged. I'd call it 'evocative realism' as opposed to photorealism. 

Anyone care to engage in a challenge in work such as my image in this post?

::::: Opera :::::

Quote - In the past, whenever I've had issues with a given Poser scene -- I'd estimate that approx. 90% of the time the problem cropped up at the end of the scene-creation process: during the final rendering stage.

That's usually the case, even in highend apps.  Even in some studios I've worked at, sending frames off to render was always the part of the process where we held our breath, and always the part that presented the most opportunity for something to go wrong.  Technical Directors were always pacing the floor, ready to come up with a solution to a render issue.  That's why rendering in passes and having more tweaking options is something all professionals and serious hobbyists should want and need.  We always did a lot of the major color adjustments, render effects,  and lots of compositing in post, and rendered according to that theory to get the most consistant results.

Quote - Once again: Poser Pro might serve to fix that.

Yes, it could.

I have NEVER had a Poser7 render stop once it accomplished the first bucket. I've had it refuse to start, yes, because I exceeded the poly burden or the texture cache or whatever. This includes a 23 hour render I just did for fun on an unsuported Poser feature, GI. No glitches.

Reflection/refraction raytracing, and especially true GI were never something used a lot in major productions until fairly recently.  Even now, they're still used sparingly.  They were available and being explored for a long time, but often dismissed as too slow for any serious production work.  Advancements in render technology, as well as hardware improvements made some of these things possible for a big production budget, but still many pro houses choose to fake what they can, or do it in post.

There's an old saying in CG that goes something like: "Never model what the camera will not see, and never raytrace what can be faked".  The first part, at least, still hold true.

On the contrary, I pick on Firefly because I do not like great thick black lines fucking up a simple flat plane of polygons in a scene with one light.  I do not like sharp polygon lines appearing in dynamic cloth simulations because phong/smooth shading inexplicably fails to work.  Maybe this kind of thing will get fixed "eventually", but it's enough to make me switch to different software.  Photo-realism is not a big priority for me, but work that doesn't look like shit is.

As for whether C4D can render human figure studies well - I don't pretend this example is all that incredibly praise-worthy, but make your own judgment.

http://www.renderosity.com/mod/gallery/index.php?image_id=1503999
*nudity

Not saying that example is especially superior to Poser's output (although, imho, the shadowing is much better than anything I ever got out of Poser) but - woo, guess what, I didn't have to postwork evil shadow map incorrect offset or the like.

PJZ,

That's a large render, 1200x1344. What was the render time?

Certainly you have a scene with skin and shadows on skin. Very well; I hope you influence more Cinema artists to attempt this type of work. In my opinion it is good, but not better than what Poser7 can do with FF.

I have not drilled down into your lighting/render troubles link, so I can't comment on it tonight.

::: og :::

Actually it's a crop of a 6000x5000 render (nearly life size on a 20" monitor).  Render time was about 1.75 hours as I recall, but I don't know if that's really a measure of anything - just two 60k or so figures, subdivided once at render time for smoothing (maybe 240k polys in the render).  The main hit on render time there was area shadows - rendering the same scene with depthmapped shadows takes maybe 5 minutes.  The GI phase of the render took maybe 3 minutes.  There is a floor in the scene, even though it is cropped out, which would speed things up if it wasn't there (not necessary for that framing).

*What if these critics were to yeild to what Poser does well, the simple portrait or body study of a human model with a non-complex background? I think you begin to level the playing field.

Sorry to say so, but Poser is capable of much more then that. We're not using Poser 1 anymore..... we're on Poser 7 these days and Poser 7 should be able to handle complex scenes, not a one figure render with a simple background only and that is not what EF has been improving Poser for all this time. It's obviously they want to break away from the nviatwas images mentality and improve Poser in such a way that it will be capable of much much more. You may use Poser for nothing more then that and that's fine, but don't day that Poser these days is intended for only that. You don't need Poser 7 for that, Poser 3 will suit that need just fine.

Granted, EF hasn't arrived yet, but they're heading in the right direction. Poser is growing and improving to handle larger scenes and if you go from P4 to P6 you'll be able to handle a lot more. The only real problem is the way the firefly engine has been implemented seems to be holding Poser back from what it really can do. I personally believe that the firefly engine could be optimized much more.

It may sound odd, but whenever I'm rendering larger scenes, I revert back to the P4 render engine. I know the quality is less, but it allows me to render much larger scenes then Firefly. I need to lower the settings of firefly so much that the quality becomes unaccepetable, so instead I use the good old P4 rendering engine. Capable of handling large scenes, quality ok and a lot faster.

But.... I rather use Firefly, but to do so it needs to become a lot faster and able to handle larger scenes without compromising quality and loosing lot's of rendering speed. It's kind of odd that to use all of Poser's new power, I still need to revert back to the old rendering engine, because the new one can't handle it. Shouldn't be like this.

The problem is that the performance gap is pretty extreme even with basic features.

Take this simple render for example. 2 lights, 25 apollo max meshes (total 2 million polys) and soft shadows. In XSI this renders in about 80 seconds on a single core system. RAM stays at about 600 MB. Now try this in FF. If it doesn't outright crash i doubt it would render in under an hour. And i didnt even tune it that much in XSI, i could probably get it down to under a minute if i really tried.

**aeilkema,
**sounds like you want a free lunch. You want better quality, but are not willing to pay for it. How do you know, precisely and objectively, that FireFly is underperforming? I am not talking about how it fails to meet your dreams or wishes, I am talking about how it fails to actually perform, based on the demands you feed it. Are you measuring poly throughput and total radiosity bounces and then comparing apples to apples numerically?

Your statement "The only real problem is the way the firefly engine has been implemented seems to be holding Poser back from what it really can do. I personally believe that the firefly engine could be optimized much more" is unfounded in this thread. You have not laid the factual basis for it, therefore it is only your perception until you substantiate. You are, of course, free to state your opinions without providing the foundation, if there is one.

Additionally, you miscast my statement. First, there is an implication I am speaking about "nviatwas images mentality," which is summarily rejected; I am talking about art nudes or clothed character studies in fairly simple settings, as opposed to grand architecture with GI and a crowd and detailed machines etc. Second, it is not 'merely' a single figure, etc., it is a single figure, or small group, with supurb lighting, skin realism, highly detailed hair that casts shadows and interesting shadows/lighting. Third, I am not saying "Poser these days is intended for only that"; I'm saying if you want it to do more than that be ready to pay for it in time. And ask yourself if Poser is REALLY an app intended to render TLSGWGRHASIACAGCWPRSAA? (two lovers saying goodbye with great realistic hair and skin in a crowd at Grand Central with perfect radiant sunbeams all around)

In this paragraph...
"It may sound odd, but whenever I'm rendering larger scenes, I revert back to the P4 render engine. I know the quality is less, but it allows me to render much larger scenes then Firefly. I need to lower the settings of firefly so much that the quality becomes unaccepetable, so instead I use the good old P4 rendering engine. Capable of handling large scenes, quality ok and a lot faster." 
...you do not make it completely clear if Poser actually refuses to render on your system with your 'larger scenes' at great quality or if it just takes a long time to render and you are hurt that it does. In either case, does it occur to you that you need to get a better computer or else have the patience to let it take as long as it is going to take? To pay for your lunch, in other words?

::::: Opera :::::

Quote - I am talking about art nudes or clothed character studies in fairly simple settings, as opposed to grand architecture with GI and a crowd and detailed machines etc. Second, it is not 'merely' a single figure, etc., it is a single figure, or small group, with supurb lighting, skin realism, highly detailed hair that casts shadows and interesting

Ah i see what you mean now. But still i dont think poser guys should be limited to doing single character studies either. Not when decent render engines are dirt cheap these days.

Sure it's intimidating to move out of poser, but the rewards are worth it IMO.

Quote - Your statement "The only real problem is the way the firefly engine has been implemented seems to be holding Poser back from what it really can do. I personally believe that the firefly engine could be optimized much more" is unfounded in this thread. You have not laid the factual basis for it, therefore it is only your perception until you substantiate.

Please, please don't be so silly.  At the very least, the multithreading implementation is dumb as hell and has enormous room to be optimized.  I don't think anyone here is putting Firefly forward as any kind of leader of the industry for final quality or render speed - and frankly one shouldn't expect too terribly much in comparison, at the price anyhow.

operaguy you're still limiting poser a lot, way too much. Also the price isn't the issue either. Carrara 5 standard has a pretty good rendering engine and yet is in the same price range as Poser. C6 standard will soon be able to anything that Poser 7 can do and add tons of stuff to it and an excellent rendering engine, yet for the same price (or even less) then Poser costs. My C5 can handle TLSGWGRHASIACAGCWPRSAA pretty well and I've paid less for it then I've paid for Poser. My Vue 5 can handle that and I did pay about the same for it as for Poser. So, price isn't the issue at all. Poser can have a good rendering engine and we wouldn't have to pay much more for it at all, it's just that EF's choice for a new rendering engine wasn't the best one.

*...you do not make it completely clear if Poser actually refuses to render on your system with your 'larger scenes' at great quality or if it just takes a long time to render and you are hurt that it does. In either case, does it occur to you that you need to get a better computer or else have the patience to let it take as long as it is going to take? To pay for your lunch, in other words?

*That's really besides the point. Everyone (except for you perhaps) knows that the firefly engine in Poser is way to slow and no matter what system you use, it will always be too slow.

Depends on what your intended goals ultimately are

ONE Nekkid Poser figure on  simple back drop?? easy if thats enough to please you
its abundantly clear by now that poser is capable of that  much at least.

but the reason one might find a dearth of such***"Intimate"***
renders posted in the Cinema/MAX/MAYA gallery is
Due to the fact that those programs are capable  of and intended for rendering
much more.

Example 1): I rendered  this "simple" scene "The Gardener"
Last year. SIX  DAZ M2 Figures at around 25,000 polys each
wearing full body suits for  total of 12 figures in scene.

RDL7 Set pieces from Sanctumart
very high quality but INFAMOUS for there ultra poly count and utra hires texture sets. add the piece way in the back from stonemason.
Cinema reported this scene at well over one **MILLION ** polygons.
the online gallery version of this scene took a little over and hour to render

the 10 FOOT by 5 FOOT wall mural version that was printed on one piece of roll canvas on a Gandi "Jeti 3000 grand format printer , took over 8 hours to render at actual size.
RENDERING HARDWARE at the time: first generation Apple E-mac 700mhz
with 256 MEGS OF RAM!!

When a 3D mesh becomes a righteously defended personification of your personal desires It has become an IDOL and YOU have become its worshipper,slave and servant
-Tain-