It seems everyone is obsessed with high-res. Figures have to have tons of polygons, textures have to be 4000x4000 minimum, etc. Taking a look at major CG films and animation projects, some of the most realistic examples use primarily procedural shaders, which P5/P6 support very well. I've seen a few people do amazing things with the shader room, but not many people seem to appreciate it for what it is. I've seen people use mathamatical alithograms to create eye textures completely map free. I've seen decent looking skin shaders done - with practice, and a group effort, we could see amazingly realistic skin shaders without the use of a single image map. Every day I see posts complaining about "I have 4 V3's and a M3 plus a vehicle and a bunch of clothes and weapons in my scene, and poser stalls. Poser sucks, it can't handle even a few figures!" - Now granted, Poser may not handle textures and models as well as some programs, or even as well as it should. But take a look at V3 and M3 - They're extremely polygon intensive. Much more so than they need be - and their modeling, while fairly realistic in most reguards, doesn't seem to be too efficient. Throw in some morphs, some 4000x4000 textures, some weapons, some more 4000x4000 textures for them, a ton of lights, turn on ray-tracing to get those reflections going, and you're trying to render a pretty complicated scene. I'd like to see modeling trends going in the opposite direction - Lower res figures with well thought out joints and polygon distribution, procedural textures that don't take up nearly as much RAM, etc. I'm curious as to why everyone seems to be running away from these things instead of embracing them.

You can get low-res figures. I use one for animations, for example. But for character work, I want a detailed, well-defined face. But I do have one quibble when it comes to realism. Rather than just having a really large pixel-size of skin texture (which shags my computer when it comes to render) are there other ways for merchants to produce realistic skin textures?

I think so many micro-details can be done with textures and displacement, so high poly figures are really just a waste of resources. I've gotten great close-up detail with James lo-res for christ sake. ;-P That said, the problem with lo-poly figures in Poser is morphing. Because Poser is not a modeling app, and most of it's users aren't intersted in that aspect of 3D, the demand for a figure that can morph into almost any kind of character is always high. So... in order to get that morphing flexibility, the poly count on the figure needs to be adequate to assume all the shapes and expressions people want. I definitely think anything over 60k-90k polys is a total waste. If the modeling is done properly, and the mesh is detailed in the right areas, 90k polys is WAY more than you should ever need for a "naked" human model (probably too much), especially with Poser's smoothing capability.

So theres any way to reduce polys of a model?

"So theres any way to reduce polys of a model?" Sure, you can decimate the mesh in some modeling apps, but it screws up the UV's, and the results are less than perfect. You can do it by hand as well, but that would take a long time, and the mesh would need to be re-rigged afterward.

from what i have seen, the reason there has not been much in the way of procedural shaders is that P5 was largely rejected (too buggy) and p6 is still kinda new. So many people are still in the P4 material mind set and think every thing has to be done in jpg images. the procedural shaders are also a big jump from the jpg textures so the learning curve hasnt been overcome yet.

Procedural textures eat up processor time. Nothing prevents you from lowering the resolution of your texture maps in an editing program it's a pretty straightforward procedure. I for one would rather have the 4000x4000 texture to start with. so that If I want a close up render for print purposes I don't have the ugly pixellation. On the other hand bumping up the resolution of a low res texture will never give you good results. 3D rendering is a time/processor intensive process. What does eat up memory is when you try on three or four different mat poses before deciding which one you like best, because Poser keeps all the maps loaded during the session. When you do that, save the file, chut down Poser and restart it to clear those maps no longer needed. If you have multi-figure scenes render them in sections. Background characters can be rendered separately and then those renders can be mapped on to a plane in the scene to cut down on resources, there are lots of workarounds.

Procedural textures eat up processor time. My memory will reach its limit of textures long before my processor runs out of time. :) I have read a discussion on which takes more to use, textures or shaders. If i recall correclty it was determined that most shaders are not as resource intensive as a hi res texture. Some such as Reflect and Refract are very resourse intinsive. re: embracing shaders- D|S doesn't do procedural shaders. DAZ has not embraced them. From what i understand DAZ requires products to be P4 compatable. P5/6 compatability is added to many brokered products but i dont recall seeing seeing any DAZ originals that have p5/6 materials. also (to the best of my knoledge) DAZ has not even updated thier "big line" (M3 and V3) to connect their bump maps in P5/6.

I think it is easier to say "we could see amazingly realistic skin shaders without the use of a single image map" than to actually do it. There are some pretty fundamental issues with implementing such a thing - the number 1 being that uv's are not evenly spaced across the model, meaning any procedural texture gets pushed and pulled by the uv's resulting in a pretty ugly texture up close. For all packages, more maps = a higher level of realism. Hopefully hardware will develop at a faster pace than our ability to invent new maps.

"Procedural textures eat up processor time." This should only be a concern if you have an old machine with relatively small CPU. Even the most complex procedurals (excluding those with raytraced material nodes) should only take a fraction of extra time for a decent Pentium 4 or AMD to crunch out, plus they're resolution independant. I honestly don't see very much difference in rendertime, for instance, when applying Face_off's complex skin shaders to a texture, as opposed to rendering the same scene without it. At least it's not perceivable to me.

I hardly ever use high res figures. I'm always making sure there is a low res counterpart before buying a figure. That's why I haven't bought most of the new figures released. As soon as one starts to do multiple figure scenes or animations, one cannot escape from using low res figures. Unfortunally a lot of creators do not see this at all. When it comes to textures, if I can avoid them great, I'll use shaders and such instead. If I can't I'm lowering the texture size as much as acceptable. This doesn't only count for multiple figure scenes. Doing a render with rendering size 1024x768 pr even 1600x1200 and then using 4000x4000 or higher textures just doesn't make sense at all. Lowering those texture sizes would save a lot of resources and rendering time for sure. It's sad to see that most creators or poser stuff are under the impression that the whole world is using high poly and huge texture sizes.

"For all packages, more maps = a higher level of realism." Yep. So long as the basics of 3D remain constant, you'll always need maps for precise control, even regarding procedural texturing. At the very least, masks and blend maps are necessary for good results most of the time.

I have been trying to figure out how to reduce the resolution of the textures I have bought and still be able to use the MAT poses. I havn't done this yet but I am thinking to make a duplicate runtimes, one with the original textures and poses for detail work and one re-sampled down for bigger scenes. I never know what data Poser uses really, can renders be improved by hiding body parts for instance? I am all for low poly figures. Search for Nea here... biggest inovation in Poser since Eve by Travler et al. in my opinion. What I want is a low poly figure (12K polys?) but done so the edges and siluatte are smooth. Usally low res figures have no problem that I can see other than the edge of the jawline and breasts...

And don't forget that many vehicles and props are "bloated" as well. Model-makers need to realize that a 150,000 polygon tank is not "showing off your skills"- in fact, a 150,000 polygon tank only slows down rendering. Does the model have to be that detailed? Maybe. Could some of the detail be added with a texture map or displacement map instead? Probably. And what is the usage of Poser- are people using it to make mural-sized images, where an audience will see the figure's pores and skin blemishes? Or are most artists using Poser to make images to be displayed online? If so, do you really need two super-detailed 4000x4000 image maps (for the figure's head and body) when you're going to reduce your final image to 800x800 pixels? Or have we just become accustomed to "bigger is better"? Would texture makers make 6000x6000 texture maps if Poser could read them? Would model-makers make 500,000 polygon models if Poser could support them? Now, sure, the argument can be made that texture-makers want to "stay ahead of the curve" and make their textures better than everyone else's. But, how far ahead of curve is too far? At what point do the textures become too large to be used on the average user's computer? And for merchants, do they really want to cut out the potential customers who have lower-end machines? --John

Is best rule of thumb be that no texture should be larger than the final render or is there subtlty found in using large texture for small renders?

"At what point do the textures become too large to be used on the average user's computer? And for merchants, do they really want to cut out the potential customers who have lower-end machines?" That's not true. You can reduce the texture, but you can't increase it very well. I would rather have the big map and use as I need it. If you start making low rez textures, merchants would start losing the majority of buyers. P4 users do need higher rez figures. It's a fact, P4 dosen't have smoothing like P5/P6, Shade and many other apps do. I don't know if DS has it. I agree that many of the props have gone overboard a bit in the wrong places. To little and P5/P6 smoothing can do some strange things and look bad in P4. I think many modelers don't understand the apps that the model is going to be used in. Question would be, how do we get a standard set of does and don'ts everyone can follow.

I have taken to resizing textures manually. As far as Poser has been concerned, it is documented by EF that your texture does not need to be larger than the rendered image resolution. You can also reduce how much of the resolution you use in P6 in the manual settings for the FireFly engine (Max Texture Size.) I set this number to the same as the largest part of the render res. Ex. for an 800x600 render, set the Max Text Size to 800. (I don't recall whether P5 had the same ability.) I do agree more with beryld. I would rather have the high-end resolution and just scale it down for my own needs. You never know when you are going to want that huge texture for a render, and it's better to have it, than not!

So its the P4 anchor that is holding us back.

The texture sizes now are ridiculous, a texture sized 2000x2000 is probably bigger than is needed. Huge textures look sharper because they are being made smaller to fit the figure, it is not better quality.

"P4 users do need higher rez figures. It's a fact, P4 dosen't have smoothing like P5/P6, Shade and many other apps do. I don't know if DS has it."

That's true, but we're well into two official upgrades beyond that application now (three, if you count ProPack as a "version"). How long will content development need to worry about being 100% compatible with P4 standard? Will it last into Poser 7,8,9, and 10? The last unofficial survey I saw mentioned that there were more existing P5 users than P4 at this time last year, and with the way they're hyping P6 advertisement, the number of Poser 4 users will likely be dwarfed even further. What I'm trying to say is, there are some obvious disadvantages to the majority of P5/6 (and eventually P7) users by continuing to create "bloated" content that was designed for an application that's at least 4 years outdated now. Message edited on: 09/08/2005 12:32

It's a fact, P4 dosen't have smoothing like P5/P6 Last I heard, the smoothing angle was over 179 degrees for P4! Does that mean that P5/P6 is over 180?

Poser Artist is still being sold... Stay P4 compatible and get the DAZ Studio market, too... Btw, a good displacement maps will use the same amount of memory as a good texture map. Mesh detail will show always, displacement only when rendered... Displacement uses render time, too. Look at the Terai Yuki release, they added a higher res version afterwards... The low res really was too low res...

"How long will content development need to worry about being 100% compatible with P4 standard?" I think only CL/eF can answer that one, how long will they be selling Poser Artist. Don't foreget Daz has not embraced P5 well, not to mention P6. DS is more of a P4 type app, untill they start selling the add on packages, I'm sure will happen eventually. "What I'm trying to say is, there are some obvious disadvantages to the majority of P5/6 (and eventually P7) users by continuing to create "bloated" content that was designed for an application that's at least 4 years outdated now." I can't add anything to that. I completely agree. Modeling so much of my own stuff now that I have learned where to cut corners and still have a very well made model that works well in P5/6.

Slow dialup, cross posted with Dizzi.

"Stay P4 compatible and get the DAZ Studio market, too..."

Hmmm. Stay compatible to capture the market of a FREE application? That's not going to be very easy. It's like Lightwave trying to capture the Wings3D market. Free app users are usually extremely loyal, otherwise they'd have the more expensive and well-known counterpart by now. There's some reason dedicated users of D|S went with that app (be it financial or otherwise), and creating MORE content that they can just use in D|S isn't going to bring them over to your side. You need to show something different... something 'better'.

Mesh detail will show always, displacement only when rendered... Displacement uses render time, too.

The fact that it will show "always" is part of the problem. Super-high res meshes bog down the viewport, you can't add a ton of things along with them, and they render very slow. Animating a fully loaded V3 in a viewport is kind of a drag when you have a complete environment for her to interact with. If there was a way to decimate the figure temporarily in the viewport, that would be good (and no, working in bounding box mode isn't the answer). ;-P Firefly is a REYES renderer with non-view-dependant smoothing. These kinds of renderers are known for their displacement capability, because it's extremely fast compared to sub-divided displacement that other renderers use. It may take a little extra time, but if you're not raytracing anything, it's not extremely noticable.

Message edited on: 09/08/2005 13:37

I also like to have that 4,000x 4,000 tex avail, in case I want to render half a face at monitor-filling size. The textures that I use most often, I've reduced myself to more reasonable sizes. However, 4K is not always 4K. I have a number of textures in my Runtime that pretty obviously were blown up by 4 times or more...some of them have a bit of high-detail work added on top of the blur, but others, it's just a big texture -- NOT a detailed one. Proceedurals are time-consuming to write. They also need a lot of tweaking and layers...and very possibly a few lower-resolution image-maps as part of the layers and controls...to get truly realistic results. I have a peculiar attitude about textures, myself. I find the majority of work out there, proceedural or not, has that "flood fill" look. The real world doesn't do surfaces like that. And real light in a real space remains more complex than our available render engines. I don't find a texture realistic unless it has some "dirt" on it, and unless it makes sense within the surface; a brick wall is never flood-filled; it has corners, it gets stained under fittings and spattered near street level -- and all of these are dependent on some method to fit the texture (proceedural or not) specifically to the surface. I also have a peculiar attitude about Poser4. It may be correct to say the user base has mostly migrated to 5. It is probably correct that the vocal users are now on 5 or 6. But thinking of this as "lazy people who haven't upgraded and should" is an apples and oranges situation. Think of Poser4 as a different program; as different from Poser5 as DAZ Studio is. You don't need PhotoShop to take the red-eye out of a vacation photograph. You don't need Max to make a simple model of a sword. For many users, Poser4 is the method and the technology that works just fine for their needs. And even if they are technically using Poser5, I think you will find a great many are basically using Poser4 methods and materials and so forth -- with only occaisional forays into the new functionality. We simply haven't migrated in useage yet. Dynamic clothing is still a minority, dynamic hair definately so -- and proceedural textures are an almost untapped resource. The average user as of this moment, regardless of the program version they are using, is still looking for JPEG texture maps. Until the entire Poser community has advanced to where proceedurals and dynamics are the norm, it is silly and self-defeating to look at texture maps as "legacy support" and "the great iron chains of the past."

Attached Link: Nice Mesh by obm890 in MP

maxxmodels you cant blame poser 4 for big files, some of the content paradise stuff is big. The idea of making models that are low resolution to be used in poser 5 and 6 with smoothing is wrong and probably wouldnt work. It also would make poser less popular, files that only work with poser 5 and 6 would be no use for vue 5 infinite and the other programs that can import poser stuff.

Poser4 is the method and the technology that works just fine for their needs. Just so long as new content is catering to their needs.

It also would make poser less popular, files that only work with poser 5 and 6 would be no use for vue 5 infinite and the other programs that can import poser stuff. But if Vue5 and other programs cannot import the poser 5 and 6 stuff the they are not "programs that can import poser stuff"

At the moment vue5 can import poser stuff, changing that seems foolish and short sighted.

At the moment vue5 can import poser stuff, changing that seems foolish and short sighted. so vue5 can import poser 4 stuff and its inability to import poser 5/6 stuff is foolish and short sighted.

Well it is in a way, the less people who can use poser 5 and 6 means obviously the less it will be used. Poser 6 is the worst version for exporting because of the pmd files and other newer stuff.

As of this moment in time, .obj is a standard able to be read by most applications (with some differences in detail). Picture textures can be used by almost all applications. Proceedural textures do not yet have a standardized form and language; with some exceptions, proceedurals must be treated as specialized to the application that wrote them. It is not surprising that heavy use of P5 proceedurals and methods, and P5/P6 file compressions, is not readily importable into other applications.

IM GLAD YOU MENTION THIS. many of the native features of a program are proprietary and NOT tranferrable to other programs even on the high end. Take FBX for example, it is universal file scene exchange format between high end programs. but even FBX wont send a liquid simulation from MAYA unlimited over to MAX or Lightwave and Cinema4D cant bring in lightwave's hypervoxel volumetric shaders either through FBX. so it not fair to just say Company X is "refusing" to support poser5/6 shaders it is not that simple.

More polys is better, without a doubt. New video games have 40k+ poly models. For CARS. A human figure is significantly more complicated than a car. What Poser needs is hardware acceleration. An x86 CPU is not exactly well suited to 3D work. That's why there are graphics cards. And it's getting cheaper every day to get a hell of a lot of rendering horsepower. nVidia now makes some of the most powerful cards for CG work, and you know what? They're just modified versions of their standard consumer cards. If only Poser, Vue, etc, supported hardware acceleration, you could have a hell of a lot more polys and much higher resolution textures in each scene. And we're talking single frame renders. Current cards can nearly do Poser level work in real time. These software companies need to get on the hardware acceleration bandwagon. A few Linux OS's already have hardware acceleration for the DESKTOP. Windows Vista will make it a commonplace thing when it debuts next year. I'm willing to bet that come next year, our 3D modelling applications still won't take advantage of our 3D accleration capable video cards.

Message edited on: 09/08/2005 20:14

Vue CAN import Poser 5/6 stuff. Just not all of it. Vue 4.5 Pro without the Mover plugin can import static scenes that are P4 compatible, whether they're made in P4, P5 or P6, doesn't matter. Vue 4.5 Pro with the Mover plugin can import Poser animations, including dynamic cloth and hair. Vue 5 d'Esprit: same story Vue 5 Infinite has the Mover plugin built in, it can import P5 and P6 scenes. P6 external morph targets are not supported, but it's no problem to save a scene without external morph targets, and it will not diminish the quality of the scene. Dynamic cloth has been changed in Poser 6 SR1 and can't be imported right now in Vue. It seems e-on is working on an update (though they're taking their sweet time). Vue does not import the Poser 5/6 procedural materials. And Vue doesn't smooth the meshes as well as P5/6 do. So for closeups rendered in Vue hi-poly figures are still needed. By the way, a portrait rendered at high resolution benefits from hi-poly meshes too - P5/P6 polygon smoothing cannot completely eliminate the typical polygon edges in a lo-res closeup. Still, V2 in P5/P6 with polygon smoothing renders smoother than V3 in P4. I'd love to see more products that take advantage of the P5/P6 capabilities. While I think that it's mathematically quite tough to reproduce the irregularities of human skin, a combination of procedural shaders with lower resolution texture/displacement maps should be able to produce very realistic skin - more realistic than a 6000x6000px texture map alone. Same goes for dynamic cloth and hair. Well made dynamic cloth beats conforming clothing any time of the day when it comes to loose flowing cloth. But there's so few merchants who dare to take the plunge, and so we still see skirts and dresses appear on the MP that need dozens or even hundreds of morphs to make them drape more or less realistic. Twiddling the dials to get them right usually takes much more time than a straightforward calculation in the cloth room.

There is nothing wrong with the newer features of poser 5 and 6 being used, but the low resolution figure idea is a bad one. Vue doesnt have mesh smoothing but most other programs dont either unless you physically change the mesh. It would be better, if poser 7 could be written to use all the memory that is available to use.

Attached Link: http://forums.cgsociety.org/showthread.php?t=187263

Reducing an object vertex count does not mean you are going to have to re-rig it in poser. take for example, my last go at this: Ultimate Wings are EXTREMELY over detailed, they are a figure with bones and groups etc etc. I loaded the obj file into PolyTrans, used the option to split object into groups (this splits the mesh into the groups that are setup for the different bones) then used the Polygon Reduction option to reduce the models polys to 30% of its original mesh count. after doing this with each of the groups, I then re-exported the mesh replacing the original .OBJ file that came with the wings. (backing up the original of course) jumped back into poser, double-clicked on the icon for Ultimate wings. BOOM in came my newly mesh optimised wings. applied textures (UV's stayed just the way they were) jump into my gallery if you want to see these reduced mesh wings in a render its not hard, and there was no loss of quality in the poser render, even after removing 70%!!!!! of the original model.

"its not hard, and there was no loss of quality in the poser render, even after removing 70%!!!!! of the original model." That's great. I was thinking you may at least have to adjust JP's, because of possible problems with bending after you reduce the polycount, but apparently not. Have you tried this on a human figure yet? 70% reduction is awesome.

Just got an idea, don't know if it's feasible or not. One of the major reasons to have a higher polygon density at the joints is those bending problems. That's where JPs and JCMs come in. Since materials in P5/P6 are animatable, and an animated material produces a value parameter dial on the figure, would it be possible to do Joint Controlled Materials? In that case, Joint Controlled Morphs could be replaced/augmented by Joint Controlled displacement - and some other things too, since skin stretched taut over bones (bent elbows, knuckles) looks different from looser skin due to the displacement of blood.

"Since materials in P5/P6 are animatable, and an animated material produces a value parameter dial on the figure, would it be possible to do Joint Controlled Materials? In that case, Joint Controlled Morphs could be replaced/augmented by Joint Controlled displacement - and some other things too, since skin stretched taut over bones (bent elbows, knuckles) looks different from looser skin due to the displacement of blood." Interesting idea, definitely worth deeper exploration. I don't know if Poser's 8-bit displacement will work well for this, but I know there's a workaround involving the use of floating point 16 bit displacement maps, but could they be animated in this way? I don't know if I'm explaining it right, but hopefully you know what I mean. Messiah:Studio has some kind of clever implimentation of animated displacement that can do things like this. I'm sure it's quite complex, but I don't have that app.

I'm inclined to the view that people do some slightly silly things with textures and UV mapping. It's worth looking at the changes there have been in the UV map layout since P3 with its one-map-per-object limit. The modern standard should have body, head, and eyes as seperate texture maps. On some older figures the eyes use a corner of the head-map, and a choice of eye textures can easily waste disk-space or render-time memory. Lips have the same problem, but it wouldn't be impossible to re-arrange the UV mapping yourself.[1] You'd need a huge close-up to justify a 500-pixel eyeball. I think you could do good things with a seperate UV map for the hands. What I've found annoying is those clothing sets which have multiple components, so you could use the skirt from one set and the top from another, but use a single UV map for the whole set. You only use 500 pixels for each component, and you;re tying up render-time RAM for maybe 4 times that.[2] [1] Open the figure in UVmapper and use select by material on the lips. Drag the handles to maximise, and save the .obj under a new name. Save a .uvs as well. And don't forget a texture template. [2] UV mapper again, and you get much better tshirt slogans afterwards.

"Since materials in P5/P6 are animatable, and an animated material produces a value parameter dial on the figure, would it be possible to do Joint Controlled Materials? In that case, Joint Controlled Morphs could be replaced/augmented by Joint Controlled displacement - and some other things too, since skin stretched taut over bones (bent elbows, knuckles) looks different from looser skin due to the displacement of blood." this is entirely possible within Poser,I've slowly been playing with the idea after hearing Stewer mention it a while ago & after seeing the animated displacement in messiah. I'm using easy pose underground to set up the dials..but it's just a case of making the material animatable in poser then saving the figure..each animatable material now has its own group that you can slave to whatever dial you want though like Max says there is the 8bit issue :

With polytrans you might be able to convert the morphs the same way to apply them to the lower resolution figure.