How many polygons do I need?
Experimenter
Posts: 162
I have a question about creating furniture and Iray rendering:
With the new Iray render I am not sure how the number of polygones influence the look in the final render. My situation; I creat a table very simple. Every part of the table has the minimum of polygons (legs and plate are made of 6 polygones). Than I made the table with more polygons as necessary (not know the exact number). I rendered both single and together. I did not see any difference. But I know there is a difference.
I want a realistic furniture which looks real in iray renders. How many polygons do I need when I creat a table?
I know this was asked before but I did not find a good answer for my question. Thank you for your help
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
As few as you can get away with and get the shape right. The more you have, the longer it takes to render. It's not the number that is important, but how and where you place them. If the geometry is clean and the textures fit correctly, you'll get a good result.
I know this is not the answer you are looking for, but, like in all things CG, the answer is usually...it depends:)
Does that mean, there is no difference between one and one hundret polygones on an flat plane?
Typically, no, there is not.
That will depend, however, on certain other conditions. For instance, certain rendering options that depend on the use of vertices for calculations, like "Smooth Shading", will only have four sets of vertice normals, for instance, with which to run such a calculation. If more vertices are needed to get the desired effect in Smooth Shading, then... more are needed. :)
Largely, what determines how many polygons you use will be left up to the intended use of the object. In a still rendering application, where having a lot of processing overhead dedicated towards animation and effects isn't going to happen, a model that uses more polygons to reflect object detail is usually not an issue at all. In a game environment, however, memory is king and many details are not actually physically present in the model, but are done with normal maps and other "procedural" effects.
So, for example, with this in mind, let's say you're going to use this table in a scene with a couple of chairs and a couple of figures sitting in them, having dinner. The total polycount for the scene isn't going to be that heavy, so you can afford to add some nice geometry details to the table. But, let's say you're making a table that is a student's desk in a classroom and you want to fill the classroom with student desks. Then, your considerations would be a bit more stringent and you may choose to limit the polygons so you can put in lots of tables without taking up too much memory space. (Note: "Instanced Objects", if you can use those in DS or whatever you're using, are handled differently. But, it takes a specific sort of operation/program to use Instancing effects.)
Also - One face for the top of a table might be just fine. However, what's usually most important in modeling for Poser/DS are the "Edges" of surfaces. In order to present a good edge for rendering, you may need to do some Chamfering, adding some extra polygons to really define the edge of the table and to prevent certain rendering operations from getting confused about where the edge of the table ends and the rest of the object begins. (The middle object in this pic shows chamfered edges - http://3d-kstudio.com/wp-content/uploads/2014/02/edgechex_04.jpg )
Lastly - It is possible to have procedural shader effects on an object that rely on polygons to determine the direction of certain effects or how patterns will be displayed. If you have planned for some materials like that, you need to be sure they'll have the faces, and thus the UVs, necessary for their own operations.
It's not really all that complicated. Just plan your object according to your intended use. If it is for other people's use, then try to create an object that is equally suitable in whatever the most likely of useage conditions would exist.
LIke I said...the answer is always..."it depends"....:)
When I first started out modelling, my mentor advised me that any edge that does not define a change in surface level has no right in the mesh... Of course, that advice was meant for low-poly models, on old, slower machines, but it does hold true in general.
I am a Hex/DS 4.8 beginner. Would it be fair to say that a number of DAZ models have kind of skipped on the chamfering, with the result that the wood on the railing of an old sailing ship say, or weathered boards on a beach or boardwalk come across as having razor-sharp edges more like new, freshly milled lumber. Old stone planes on the parapet of a castle look like newly-done plasteel (that's what they called it in "Dune"), that sort of thing.
I've been wondering if it might be worthwhile to go back to some of these sharp edges and dull them up a bit with a few more polygons? Or should the end-user just stick to the AI routines found in the latest generation of postwork software. If you already have a model and you buy "version 4.0" it would be specially marked in the catalog and there would be a special "upgrade" price and so on.
Of course if you're saying that the sharp edges are a physical requirement for some shaders say, then postwork is definitely the way to go.
Another thing you have to take into account (Daz Studio) is the user can add various types of smoothing to an object, which will ruin the shape if not built properly.
Oh-oh... that's something else that I haven't been up on. Let's see... I've noticed the sharp edges happen a lot on lumber, so how about some barn boards... ok, looking at the DAZ barn model, if I go Edit > Object > Geometry in DAZ Studio and I select "Apply Smoothing modifier", a lot of stuff happens. Is this what you mean? Wow, news to me thanks.
No...not the Smoothing modifier, but Subdividision (Convert to SubD). The smoothing modifier is going to affect how something will deform when in contact with something else. It's a collision system, just not a very advanced one. It doesn't matter if the object is constructed with one or many polygons or 'properly' constructed...if it is supposed to be a solid object, like a wooden plank, smoothing will mess it up, no matter what you do to it.
SubD on the other hand, will round the corners/edges or even break the model into pieces if it isn't constructed properly (bevels/edge loops to preserve hard edges, welded parts/vertices, etc).
Sorry for the delay in replying.
The fact is that most of those "inorganic" models are not modeled specifically for Poser and, I suspect, neither for DS as well. Basically, they just didn't bother with paying attention to the requirements of the render engines. Though, I have almost no experience at all with DS, so take anything I have to say about it and its rendering with that in mind. I will try to explain from the viewpoint of Poser, organic renderers and smooth shading concepts: (Pack a light lunch... ;) )
Poser and its Firefly rendering engine is primarily an "organic" renderer/posing/modeling sort of application. It, and other renderers like it, specialize in rendering organic shapes. These are nice flowing shapes with gentle edges, smooth lines, and delicate twists and the like. You know, flowers, a nice face, beach balls, boobs, legs.... whatever. These are "organic" geometries. "Inorganic" geometries are things like most tables, square railings, calculators, houses, machinery, giant-communist-robots... that sort of thing.
Many 3D modeling apps owe their roots to CAD architectural and mechanical apps from years ago. And, most of the models you'll find on the web are for those sorts of modeling/rendering apps. (Architects love having lots of free architectural furniture and greebles to put in their building renders) And, aside from furniture cushions, almost all of them will be "inorganic" objects. (Several reasons for that, the primary being that inorganic modeling is more difficult to do in detail, in general, for things that people would actually want to have as a free 3D model.)
OK, Poser specializes in inorganic renders/models, right? Well, one "trick" that is used to render smooth surfaces is "Smooth Shading." After all, inorganic objects have smooth surfaces, generally, and if we modeled every single face of a smooth surface, the models would be... ginormous in size! Smooth Shading is accomplished by a complex algorithm that takes a measurement of all adjacent vertices in an object and "shades" them based on their relative angles to each other. This is why, for instance, in Poser, a ".20" shading rate is default, since a 20 degree variation in angle is, here, what determines the "Smooth Shading." And, if you increase that "angle" of "Smooth Shading" what do you get? You get smoother surfaces in a render for vertices with relative angles determined by the smooth shading setting!
BUT, without enough vertices, what happens? In order for anything at all, ever, to be "smooth" it either has to have lots of small changes in relative angles of faces/verts or... none at all, right? That is how the algorithm is structured - It is based on having enough things to measure in order to render a "Smooth Shading" effect between relative vertices with different relative angles. (Non-planar)
When it does not have enough vertices to measure (The vertice normals, actually, but that's not important atm..) then what happens? Well, it still tries to render a smoothing effect, since that is what it was told to do, but without enough geometry to actually do that... so it borks the heck up and renders "bloated" edges and tortured geometry that's all twisty-looking and borked up...
To keep that from happening with Smooth Shading effects in Poser or anything that relies on similar Smooth Shading algorithms to produce a smooth surface, giving the impression of much more geometry creating a nice, smooth, transition, one has to have "guide edges" to guide the effect. (ie: One has to give the algorithm more things to measure) Typically, this is done with "Chamfering" since that is an industry standard modeling procedure and can be easily done with automated processes. (But, a simple "chamfer" tool will not always produce desired results, especially on sharp corners, where newly created edges will intersect and produce "puckering" when Smooth Shading is in effect unless they are corrected, typically by manually cleaning them up.)
With Smooth Shading, you can not easily "Postwork" the resulting mess into something that is suitable. How, for instance, can a Smooth Shading effect be done on a surface that has only two sets of coordinates for the algorithm to process that rest on a 90 degree angle? ..... It ain't gonna happen. :) But, you've got Smooth Shading you would LIKE to have on that model, right? (Some round pegs or a bowl shape or something?) And, therein lies the problem - Most typical 3D inorganic models are not made by people who are modeling the model for Poser or any other program that uses similar smooth-shading effects. (Or, they don't know how to.)
For instance, take a table, virtually any 3D table model you can find. Let's say it has round legs, OK? Well, they're "supposed" to "look" round, but they're actually just low-poly, 8 face, tubes intersecting a 6 face square. Someone slapped some textures on it and it looks cool, right? Sure! Until, that is, you render it with smooth shading so the legs actually "look" round, instead of looking like 8 faced cylinders slapped into a box that is now mangled...
See? They modeled a table, created some cylinders as legs, thinking they'd "look rounded" in a render, then slapped a six-faced box on the top for the "tabletop", which has no hope in heck of ever, ever, rendering with a nice, clean, edge with smooth-shading effects turned on, since that portion of the geometry would not have enough data for the smooth-shading algorithm to function well with. The cylinders might render.. fairly roundish, but the tabletop would look like something Bigfoot just got through humping...
What to do with existing content being sold by various sites? KNOW THE ARTIST! Do not buy dooky unless you can see a wire-render, first. (Or, PM/Email the vendor/artist and ask about the object construction.) And, if you buy an inorganic model that is obviously modeled by someone who doesn't know how to model for Poser or anything else that uses smooth shading in the rendering process like this, return it as a dissatisfied customer or, alternatively, fix it yourself. :)
Yes, it would be nice if such things were fixed. But, it ain't gonna happen, since it'd be too much work and most of those sorts of models are not actually constructed for the purposes they are sold for... There are, of course, a few exceptions of notable artists who do a darn fine job - Stonemason, of course, is one. RPublishing, IIRC, is another. DM would probably be one, too, though I have none of their stuffs. (I hear they're nice, though.)
Years ago, I bought what appeared to be a nicely detailed interior living space set. It looked "well done" and I was intrigued, so I bought it. I thought, "Surely, this artist did a very good job with attention to detail, scale, functionality and the like. They must have prepared the models to be rendered with Smooth Shading, so I didn't have to go through and correct every single one..." FAT CHANCE! It was, in regards to Poser inorganic model products, a nicely painted piece of crap that was obviously first modeled for some other application and then "ported" to be Poser compatible. I never, ever, bought anything from that particular vendor again. And, that taught me to be very, very, discriminating whenever buying any inorganic models for Poser, which I don't usually do, since I make my own. My advice is - Do the same and make your own unless you know the quality of the vendor you're buying from.
On "Smooth Shading" vs "Subdivision" - Smooth Shading only effects the rendering process and how light interacts with the surface in order to yield the impression of a smooth transition between surface angles, based upon the availability of relative geometry (vertice normals) to help determine the result. Subdivision is a mathematical modeling process that creates new geometry (Generally using user-selected algorithms) based upon a base geometry. Because of the way most of the algorithmic functions are calculated, this usually has the result of adding more geometry, which is averaged, resulting in smoother topological transitions and, consequently, usually reducing the volume of the surface, if applicable. One does not create actual geometry, but gives the appearance of it existing, and the other actually actually creates entirely new geometry, based upon a pre-existing set of reference geometry. BOTH processes, however, can be somewhat controled by the user and the use of "guiding" geometry constructed for that purpose.
Not sure if this will help, I did a tutorial a while back (with help from Patience) on making a sword - http://www.daz3d.com/forums/discussion/44296/hexagon-sword-tutorial/p1
Also check out Garry Miller's tutorials (sets 1-4) over at GeekAtPlay.com - http://www.geekatplay.com/hexagon-tutorials.php
Garry Miller's stuff is great. Though, I admit, he does (did) some things sort of differently than I'm used to in regards to UV'ing, but.. hey, if it works, it ain't broke, right? :)