Quote - All I was looking for when I started this thread was how to deal with 90 degree corners and all I heard was I am smart but I am not going to explain how to deal with 90 degree corners.

Several people gave you advice on how to deal with this actually.

edit: it actually kinda boggles me that you come here, ask a question in kind of a confrontational tone ("is it okay if I ask?"), are given a lot of helpful information, and then say you didn't get any answers.  wtf dude, read the thread, you got handed 4-5 ways to deal with this.

Yes and I was the one that first told you about smoothing groups.

I'm really amazed at what's going on here. The confrontation over n-gons had nothing to do with you, DM, but certainly made the conversation unpleasant, I'll grant you that. I'm pretty upset that two people who know me really well refused to believe the object was made of quads. I'm pretty upset that people who know me insisted that I don't know what an n-gon is and told me to go read the definition of the word, instead of counting up the vertices in my prop. And I'm pretty upset about being called a smug asshole.

I don't argue techincal points like what the angles are or how many sides are in my polygons because I like to be superior. I argue because people are trying to understand things, are saying incorrect things, and are departing from the truth instead of arriving at it. This isn't about ego. It's about subjective reality, instead of fantasy. As long as people keep insisting on falsehoods, I keep talking.

Now you say "All I was looking for when I started this thread was how to deal with 90 degree corners and all I heard was I am smart but I am not going to explain how to deal with 90 degree corners."

I never gave any opinion of your intelligence or any other aspect of you as a person, and frankly you seem pretty whiny when you say shit like that. I said that you (and others) have wrongly assumed the corners are 90 degrees when they are not. I said that you guys would not believe me without some illustration of the situation. I spent an hour producing illustrations to show that they are not 90 degrees. You still think they are 90 degrees despite the evidence before your eyes.

We have people who still think my prop has n-gons, and you still think the corners are 90 degrees, and you still think there is a problem with 90 degree corners. None of those things are truth. They are an invitation to further discussion, as far as I'm concerned. If my repeated explanations seem "long winded", it's only because you're not getting it.

I'm not making any judgement about why you still believe the mitered corners are 90 degrees. I'm stating a simple fact, with evidence. Despite whatever you insist on believing, they aren't, just like my quads are not n-gons. I make predictions about what works, I try those things, and then I demonstrate those things. Whatever prediliction you may have for avoiding confrontation, that's fine, but if you're going to ask for information and reject it when truth is presented to you, then you're going to remain confused. Not stupid, just confused and unable to produce what you want when you want.

Your attitude is preventing you from seeing the truth. Regardless of how you think the information is delivered, it is the truth. As long as you ignore it, you'll be unable to understand what is happening.

Quote - When all things come out in the wash you have to be humble enough to admit that you don't know it all. I have been involved in graphic work in one way or another for more than forty years and I am still learning.

Isn't that the pot calling the kettle black?

Your'e the one that refuses to learn the angles at the corners. It doesn't matter how long you've modeled. Those are not 90 degrees, and you could say otherwise after 4000 years of modeling and all that shows is that you're not humble enough to admit that you don't know it all.

What an impossibly obnoxious thing to say to me. I have never claimed to know it all. I claimed to know the angle of that corner. You claim it is 90. It isn't.

Quote - And with the center polygon un-welded from the others.

I see what you're showing me, and acknowledge that the steps you're taking to alter the mesh result in it being ... altered. But that's not the essence of my question.

What I was asking boils down to this: When you render a rectangle in whatever program it is you're using, does it look like a rectangle, or does it get rounded like you're showing here?

When you render four rectangles arranged in the three ways I uploaded above, do they look like my render or do they not look like rectangles anymore?

I am suspicious that my understand of the term Catmull Clark is incorrect. But if it is incorrect, and a CC-enabled renderer will round the corners off any un-welded edge, including the outer edges of the simple 3x3 group you showed, then I don't understand how people get rectangles to show up in such renderers. To get a simple rectangle, do you have to use edge loops?

OMG...it's not like Tom comes in here and posts all the time asking questions. And yet ya'll are still acting like dicks. It really reminds me how much I hate this place anymore. Too many egos and it's so miserable I can't even read a fucking thread anymore without getting a headache. We all just want information when it comes right down to it. Why can't we get just that anymore? Nope, has to be served up with a side of self righteous, sanctimonious fucking "Look at ME, look at ME!!!"

Yep, I'm pissed. Can you tell? I'm sick of it.

Laurie

Ok...now that I've vented...

What does it matter who told him what as long as he got the info he was looking for? I mean, really? Isn't that why this forum exists? So that ppl can ask questions and get help?

And for those who are keeping track: BB mentioned smoothing groups - that's true - but did not explain how they are accomplished. raven was the one who pointed Tom to a thread where Stonemason apparently DID explain them. Just in case you were keeping track.

If I may make a suggestion...

Let's all park our massive egos at the door when someone genuinely needs and asks for help and not belittle them or make them feel stupid while you're doing it. This would be a much happier place.

Laurie

"We're all ignorant, except on different subjects"

• Will Rogers

Because Max can export smooth-groups to an obj-file and Poser is able to understand this, I've tried to find something about smoothing groups in this document: http://local.wasp.uwa.edu.au/~pbourke/dataformats/obj/

Seems that having a group tagged with "s 0" stops smothing this group (e.g. anything following until "s 1" occurs).  I don't have Max, so I'm not able to explore an obj-file written by Max.

But if this is correct, a python-script can alter an obj-file while identifying named groups ("g noSmoth") while adding "s 0" right under this "g"-entry. All other group definitions needs a following "s 1" if I understand the document correctly.

(untested hack to get the point):

fh=open("original.obj")
fhout=open("copy.obj","w")<br></br>for line in fh:
  print >> fhout, line<br></br>  l=line.strip().split(None,1)
  if l[0]=="g" :
    if l[1] == "noSmooth":
      print >> fhout, "s 1"
    else:
      print >> fhout, "s 0"    <br></br>close(fhout)
close(fh)<br></br>

Attached Link: http://www.renderosity.com/mod/forumpro/showthread.php?thread_id=2800349&page=2

@raven: Seems to be exactly as I guessed :)

<pre style="white-space:pre-wrap;">
g Sphere01
usemtl wire_227153153
s 1
f 1/1 2/2 3/3 
f 1/4 3/3 4/5 
...

<span style="font-family:Arial;">There is a "usemtl" between "g" and "s", but the order after "g" shouldn't matter.</span><br></br>

 Here is the important part from the OBJ-Definition:

s group_number

    Polygonal and free-form geometry statement.

    Sets the smoothing group for the elements that follow it. If you do
    not want to use a smoothing group, specify off or a value of 0.

    To display with smooth shading in Model and PreView, you must
    create vertex normals after you have assigned the smoothing groups.
    You can create vertex normals with the vn statement or with the
    Model program.

    To smooth polygonal geometry for rendering with Image, it is
    sufficient to put elements in some smoothing group. However, vertex
    normals override smoothing information for Image.

    group_number is the smoothing group number. To turn off smoothing
    groups, use a value of 0 or off. Polygonal elements use group
    numbers to put elements in different smoothing groups. For
    free-form surfaces, smoothing groups are either turned on or off;
    there is no difference between values greater than 0.

Sorry, I have no Poser here yet. Can't explore how Poser deals with the defined groups at the moment.

Problem solved, at least for Poser 7/8.

I called a friend of mine and he did some tests.

The "Smoothing Group ID" is just a number to identyfy a seperate group. You can not define a certain "crease angle" for one of this group.

What you can: Define a crease angle to a prop/actor "exept for one or more groups". The exepted groups are set to 0. Sharp edges (based on normals) are defined for each defined group.

Load a sphere as a prop.
Open the group-editor.
The whole sphere is defined as one group.
Now create an empty new group and select it.
Mark some facettes to become part of the new group.
Give this group a "smothing group ID".
Repeat to get another group with another smoth-group-ID.
Close the group-editor.

The preview will show you the groups without the common ugly black-shading.

That means:

It is possible to create "sharp edges" directly from within Poser (tested with 7/8). This is good for simple props.

Or we can create obj-files with each modeler able to define standard-groups. A script like the one I posted already can change defined groups to entries Poser is able to use for (not) smoothing.

Extend the script with 

...
  if l[0]=="g" :
    if l[1].startswith("groupID_"):
      groupid=l[1].split("_")[1]
      print >> fhout, "s %d" % int(groupid)
...

With your modeller create some groups with name "groupID_1", "groupID_2", "groupID_nnn".  
With a modified script it's also possible to use material-zones insteed of groups.

Sorry, can't explain it better because i'm not good with the english language.

Another "trick" from within Poser:

If you need "sharp edges" for an area with a seperate material, you can use the group editor.
Define a new group and click the button "add Material". Select the material you want and give the new group a "Smoothing ID".

This thread has finally become informative and worth reading. Thank you :o). I'm actually learning something ;o).

Thank you adp...raven :o).

Laurie

Hey Laurie,

All good things come to those that ..... keep reading ...... :biggrin:

It has been said ...

When you get to the end of your rope ...
... tie a knot and hang on.

Quote - Hey Laurie,

All good things come to those that ..... keep reading ...... :biggrin:

It has been said ...

When you get to the end of your rope ...
... tie a knot and hang on.

Well, sorry for popping off like that. Was not very classy. I apologize to the forum ;o). I didn't lie when I said I was sick of it tho...lol. I just could have worded it better.

Laurie

Laurie,

No prob ..... I don't blame you.  I think you said a few things that I was thinking, also.
I was getting a little miffed at the way this thread was headed (without regard to anyone in particular) but I am glad that things got toned down.  Less than pleasant words never really accomplish anything other than (usually) escalate tempers.

Ok, off my soapbox now.

Thanks Laurie, you're beautiful.

The obj on the right will render fine in Poser (with smoothing turned on)
The obj on the left will not.

exactly (Laurie showed that earlier too).  Poser and other Reyes renderers really require different modeling techniques if you want to make use of polygon smoothing with a minimum number of polygons.

i didnt read teh whole thread. so i dont know if i missed something.

why would turn on smoothing for this object?

i modeled this and crease angle is 80.

Awesome :)

Something I've been getting headaches over for years and it's that simple? Unbelievable...lol.

Laurie

Quote -
 Something I've been getting headaches over for years and it's that simple? Unbelievable...lol.

There is a good chance this method can't be used with P5/6. Anybody able to test this? And can DAZ Studio handle "s"-tags in an obj-file?

I know for a fact that you can assign smoothing IDs in Poser 6. Not sure about anything earlier than that.

Laurie

This smoothing business doesn't arise in DAZ|Studio because it doesn't do Reyes polygon smoothing, it does Catmull-Clark subdivision.  I don't know whether it reads smoothing groups in or whether D|S subdivision cares about them (I bet Stonemason does though).

Quote - All vertices are welded.

The obj on the right will render fine in Poser (with smoothing turned on)
The obj on the left will not.

Question - isn't the artifact here related to the ambiguity due to poles? On the left there are poles at the corners with 6 faces sharing a vertex. I've read that 3-poles and 5-poles cause trouble in Catmull Clark. Perhaps 6-poles cause trouble in REYES.

I will grant that all the apparent unwelded joint angles are 90 here, but joint crease angle is not the only thing that causes unwanted smoothing artifacts in a REYES renderer.

The n-pole problem is extremely obvious with the newest Poser Primitive cone. It isn't because they are long thin triangles on the cone. It's because there is a huge n-pole at the top.

Quote - I know for a fact that you can assign smoothing IDs in Poser 6. Not sure about anything earlier than that.

Thanks, Laury.

It's a shame - I've used P6 for years and don't remember.
The truth is, I never have had a problem with this (same workflow since P4), because I didn't count facettes ;)

Quote - I've read that 3-poles and 5-poles cause trouble in Catmull Clark.  

Not at all.  In fact they're straight up unavoidable for all but the simplest models - this topic doesn't really have anything to do with poles.  The smoothing bulging thing isn't an artifact either, it's just how Reyes renderers handle polygon smoothing.

Quote - I know for a fact that you can assign smoothing IDs in Poser 6. Not sure about anything earlier than that.

Laurie

You can't assign smoothing ID in Poser 5 (and earlier).
BTW Blender does export smoothing groups.

Quote - What I was asking boils down to this: When you render a rectangle in whatever program it is you're using, does it look like a rectangle, or does it get rounded like you're showing here?

When you render four rectangles arranged in the three ways I uploaded above, do they look like my render or do they not look like rectangles anymore?

I am suspicious that my understand of the term Catmull Clark is incorrect. But if it is incorrect, and a CC-enabled renderer will round the corners off any un-welded edge, including the outer edges of the simple 3x3 group you showed, then I don't understand how people get rectangles to show up in such renderers. To get a simple rectangle, do you have to use edge loops?

If you render a rectangle with Catmull Clarc, you get a rounded shape. In some modelers, you have the choice to make an edge hard, which means CC subdivision will not smooth it. But I don't know if every renderer honors this. There should not be a difference between modeler and renderer. 

Catmull Clark subdivision is a whole different beast then Reyes subdivision. It's made for modeling and morphing organic surfaces with only a few controll points. Reyes subdivision is just helping to smooth small areas.

At least that's what I know about this. I'm sure that your unwelded wall would not work with CC subdivision - you would get holes in the model. 

Quote - > Quote - I've read that 3-poles and 5-poles cause trouble in Catmull Clark.  

Not at all.  In fact they're straight up unavoidable for all but the simplest models - this topic doesn't really have anything to do with poles.  The smoothing bulging thing isn't an artifact either, it's just how Reyes renderers handle polygon smoothing.

OK - I'll skip any more discussion of CC, since 1) I don't know it and 2) this thread is about REYES smoothing anyway.

So if we assume the pole has nothing to do with it, then why does the mitered corner (trapezoids) matter? After all, in that particular mesh, every shared edge angle really is 90 degrees. Why does the renderer think some of those edges are less than the default crease angle of 80?

Note: I tested that particular topology and found that it renders sharp and no bows with a crease angle of 30. I didn't have time to try other values and narrow it down. Why is the renderer measuring the crease angle at between 30 and 80 when the trapezoids are present, bot not with box corners, where there is a 3-pole there instead?

which OBJ was this again?

Nobody posted the obj, but bevans84 showed this very interesting comparison.

I had to make my own. The one on the left shows bowing and shading artifacts unless the crease angle is set to much lower than 80. I did not build and test the one on the right - I'll take bevans' word that it renders correctly.

In both, all vertices are welded, there are no holes like in my partially welded props, and all the faces meet at 90 degrees, but the one on the left (with the 6-pole corners) produces a different outcome.

I don't model either of these ways - in all the models I've made, my corners are 3-poles, or they are not welded. When I make a door, the four side pieces making the edge of the door face are all rectangles, partially welded or not welded at all.

When I make a picture frame, the corners are mitered (as on the left) but they are not welded in the corners - just across each piece.

I do the various shapes according to the shapes of the real pieces of wood used to make such things. I UV map them so that wood grain looks like the real objects. Doors are not mitered. Picture frames are.

And don't forget the question about the cone - the n-pole at the apex is a problem. Why? Crease angle should take care of it.

I'm gonna hazard a guess that old versions of Poser judged whether to smooth or not based on edge angles, not face angles.  The left object has 45 degree edge intersections vs. all 90 degree edge intersections on the right.  I think somebody we all know has a working relationship with Stefan and can simply ask ;)

Here's the prop I built. See what you get.

Before the top was bowed. Now it is the left, back, and right sides.

This is not logical. The topology is identical.

Hmmm, is there ANYthing logical about Poser?

How many peeps?   ... How many hours? ... I rest my case. :lol:
(it wuz gittin' purdy tired anywho)

I've been noticing something disturbing lately...

Putting UVMPro into my workflow sometimes creates the bowing problem.

For example, I make a door. I set up the hard edges in Wings. If I bring the door straight from Wings to Poser, Poser respects the hard edges and renders well. If I run it thru UVMPro and do any mapping, material zones - anything where I have to resave the model in UVM, then when I bring that into Poser, there's problems. If I do all the uvs and material zones in Wings to Poser, no problems. I suppose it's wiping out the hard edges I created in Wings.

Laurie

Were you doing that in wings by splitting the edges, or what method?  If you're splitting edges maybe UVmapper (which I don't use) is re-welding them on import or export.