Help with spheres.

[Charolotte Caxton]

Hello. I am an amateur photographer in second life and I am having an issue with my spheres. I dont know if I am just now noticing this again or if something else is going on. By again I mean I had previously noticed this issue in my shoulders and had that resolved, but now I am noticing it everywhere.

In the photo I have included is a sphere. Notice it is not round but rather made up of many angles, polygonal I believe is the term. I understand about the mesh, insofar as when I view the world in mesh (ctrl alt t) that is exactly what that pseudosphere actually looks like.

What I would like to know is if there is a way to deceive myself into believing I am actually seeing a sphere, ie antialiasing type stuff that I am just barely understanding of.

If anyone could post examples of spheres that look like spheres that would really help out a lot. I am including my system info just so you know what kind of graphics processor I am working with. My graphics settings under preferences are Ultra. Hardware settings are Anisotropic Filtering enabled, Antialiasing 16x, VBO enabled.

Thank you for your time.

[Dora Gustafson]

---

jjccc Coronet wrote:

there is a way of getting smooth spheres in sl i think as a particle bubble ive seen them and they are smooth all the way round

---

All particles are flat as freckles, even bubbles:smileyvery-happy:

But the viewer always shows you the full picture no matter the viewpoint.

Just like floating/hovering text

[koto Nizna]

I think there's no perfect curve in secondlife. If you take a look at your own avi it also has sharp corners. My suggestion is.... Fix it in photoshop (liquify) .

[Knowl Paine]

When I zoom in it is round, when I zoom out it reduces quality. There is a way to stop it but idk. I do believe that all objects will retain their qualities which could cause some serious lag if too many objects are in view.

The squared effect can be helpful when aligning cylinders, in the past I would reduce graphics settings to highlight the build but this current lack of roundness seems to be more noticeable then it was in the past.

The feature is like Render avatar imposters effect. If I can find it I will update. 

[Charolotte Caxton]

@ Knowl, thanks. I needed someone to let me know that they noticed the lack of roundness had become more noticeable, thank you.

@ Koto, yes I had indeed first noticed this lack of curves in my own self, but had resolved myself to it, but I used to be able to see spheres as spheres and now lately I can't help but notice the angles. For photography your photoshop suggestion is spot on. A lot of my work is untouched so I was looking to avoid post production, not that I think there is anything wrong with that, quite the contrary, I just dont have the experience.

@ Coronet, lol. Thanks for showing me that its not just me that sees the spheres that way, whew. I suppose I'll have to start importing my spheres from Google

@ Dora, good point.

[Josh Susanto]

I would go with the particle idea.

OTOH, you can always sculpt and duplicate a spherical section of greater resolution and join them all together :-/

[Josh Susanto]

OR something like this solution... 

The sculpt map above produces the sphere on the right, which looks slightly rounder from almost any angle.

With a little more fuss, you can probably do even better.

[Charolotte Caxton]

Thanks Josh. 

[Luc Starsider]

You may be thinking of LOD (Level of detail). Which works so that when you move farther away from an object, another instance of the mesh with fewer polygons are used to save how much resources are needed to draw the object. This is to make more horsepower available to render the nearby objects, and make sure everything render as quickly as possible. Perhaps not a good, or 100% accurate, explanation, but hopefully it is understandable.

In preferences, the Graphics tab, the slider for 'objects' under 'Mesh detail' affect how quickly LOD kicks in. The higher the value in the slider, the farther away you can be before the next LOD happens. A Debug setting is also available which make it possible to set an even higher value than the slider in prefs; RenderVolumeLODFactor. A value of 4.000 is regarded as giving the best results, although I have seen it said that 3.000 is just as good.

Sculpties often have a higher number of polygons than regular prims. I don't know if this is the case for spheres, but that would explain why they sometimes look better than regular prims.

- Luc -

[Josh Susanto]

That's not the difference in this case.

I just bent the default sculpt sphere around 3 axes to produce pervasive vertex disjunction without changing distance of the sculpt surface from the gray point.

What you're actually seeing at the perimeter of the sphere is more than one set of limits. That probably sounds intuitively like a bad idea. But it should work with practically any surface texture because the layering occurs at a point where you're looking at the texture from an essentially sideways perspective, and the layers will tend to look much like a single layer for that reason.

The downside is that your surface texture will map non-orthogonally. OTOH, the two puckers will be easier to hide because they are closer together.

If you need a sphere with no seam and only one pucker, let me know.

[Luc Starsider]

That flew right past me in a hurry. I'm going to need that in non-expert language. If you are referring to what I said about sculpties sometimes having more polygons than regular prims; that's something I read and don't know anything more about. I thought maybe if that is the case for one type, perhaps it also is for spheres.

- Luc -

[Knowl Paine]

I think LOD might be it, I will look at my settings. ty 

I'm not following Josh either, I did read it twice. Sounds like he measured something.

Did I mention I like to build? I know about that. Round prims are less round than they were a few weeks ago. As to why, idk.

 

[Josh Susanto]

In this case it is not essentially  a question of the number of verts. 

It's more a matter of filling in some of the visual gap between a straight line at the edge and the curve which it is intended to approximate.

While a normal sphere, even a normal sculpt sphere, basically presents one visual perimeter, the sculpt I created has all the available polygonal perimeters twisted around so that they somewhat slop into the negative space tangent to the predominant perimeter as seen from practically any angle.

Try to imagine the perimiter of a normal sphere as being one of a set of polygons that can intersect a wall perpendicular to camera view. With a normal sphere, the other polygons closest to the widest one are of similar geometry, and so leave basically the same gaps as that polygon. On the sphere I provided, the smaller polygons immediately in front and in back of the largest one are nonetheless large enough that, rotated as they are, their corners push subtly into the gaps formed by the other two. The end result is that they form as a total perimeter an irregular aggregate polygon with many more sides than the main one alone.

Any clearer?

 

[Josh Susanto]

>Sounds like he measured something.

Not this time.

It was pure eyeballing and I stopped tweaking as soon as I was confident it would snapshot as any more round than the template, which practically right away.

Just try it yourself. The conceptual part is the only hard part, and I've already done that for you.

[Charolotte Caxton]

---

Josh Susanto wrote:

In this case it is not essentially  a question of the number of verts. 

It's more a matter of filling in some of the visual gap between a straight line at the edge and the curve which it is intended to approximate.

While a normal sphere, even a normal sculpt sphere, basically presents one visual perimeter, the sculpt I created has all the available polygonal perimeters twisted around so that they somewhat slop into the negative space tangent to the predominant perimeter as seen from practically any angle.

Try to imagine the perimiter of a normal sphere as being one of a set of polygons that can intersect a wall perpendicular to camera view. With a normal sphere, the other polygons closest to the widest one are of similar geometry, and so leave basically the same gaps as that polygon. On the sphere I provided, the smaller polygons immediately in front and in back of the largest one are nonetheless large enough that, rotated as they are, their corners push subtly into the gaps formed by the other two. The end result is that they form as a total perimeter an irregular aggregate polygon with many more sides than the main one alone.

Any clearer?

 

---

Hi Josh, thanks for the explanation. While I was wondering if the spheres I see are now not so spherical, your answer is very intriguing.

For your sphere, with the twisted polygons, are you saying you twisted each peak? or do you have spheres inside spheres? or did you create a sphere made of many pieces? if all those questions are way above my level, please excuse. 

[Luc Starsider]

---

Knowl Paine wrote:

Did I mention I like to build? I know about that. Round prims are less round than they were a few weeks ago. As to why, idk.

 

---

 

Yes, you did. What I said above, apart from that about the LOD setting, was ment for anybody in general, not you specifically. I apologise if it came across differently.

I haven't noticed what you say about spheres being less round these days, but it has been a while since I did any building. I'll have to pay attention the next time. Like you, I can not imagine why that would be?

- Luc -

[Luc Starsider]

Yes, thank you. Very much clearer. (But don't ask me to explain it back to you; Limbs will break if I try... )

I will have to get my hads dirty and try this to fully comprehend the concept, but I think I got it.

- Luc -

[Charolotte Caxton]

---

Luc Starsider wrote:

---

Knowl Paine wrote:

Did I mention I like to build? I know about that. Round prims are less round than they were a few weeks ago. As to why, idk.

 

---

 

Yes, you did. What I said above, apart from that about the LOD setting, was ment for anybody in general, not you specifically. I apologise if it came across differently.

I haven't noticed what you say about spheres being less round these days, but it has been a while since I did any building. I'll have to pay attention the next time. Like you, I can not imagine why that would be?

- Luc -

---

Hi Luc! I noticed it not in building, but in general photo taking, curved objects weren't as curved anymore, so I rezzed a sphere to check it out. Globes dont look like globes, curved architecture looks jagged, I wondered if it was just me or what. That's why I posted in the photography section, cause that's how I noticed it and wondered if any other photographers did as well. I'm not a builder, yet.

[Luc Starsider]

In many ways, I guess things like this could be more apparent, or visible, when you do photography than when building. At least the way I build (which is a messy affair ) When taking pictures, you focus more on the objects in the shot, I suppose, and notice things that seems 'off' easier. Other photographers would have to say if this is in any way true.

I know when I build something, it is using the standard prims with the boring plywood-texture on them, which, in my mind, hides some of these 'imperfections'. By the time I get to do any textureing, I'm tired of the thing and just want to finish it, so I don't really think about it's roundness or anything.

(I'm rambling. sorry )

- Luc -

[Charolotte Caxton]

lol, its ok sleepy Luc.

I notice this all the time now when I am wandering through sims. Where there used to be round gears and spinning globes, now there are multiangled gears and spinning multipolygonals.

Again, I wasn't sure if I was just starting to notice this more because I was paying closer attention to detail, or if something had indeed changed.

Take a picture of a plywood sphere if you wish and lets see if we notice the same.

[Luc Starsider]

---

Charolotte Caxton wrote:

Take a picture of a plywood sphere if you wish and lets see if we notice the same.

---

But now we are aware of it, so I think we'll notice anyway, plywood or no plywood. .

-  Luc -

[Josh Susanto]

If you load the sculpt I provided, you can see the distribution of verts.

If necessary I'll load it as non-temporary and send it to you. Just ask.

What I did, really, was very simple.

I opened Sculptypaint (092 in my case), and went straight to the arch tool. The program opens with a sphere as a default shape, and I just pressed the arch tool rotate buttons for red, green, and blue a few times each.

The coordinates for the surface stayed at the same distance from the center of the sculpt, so a sphere shape was maintained, but the actual RGB coordinates got pushed around a little bit. Some verts got larger and some got smaller, so you may be able to find some kind of angle where one side of the sphere perimeter is actually less round, but these will be the exception. You'll have to actually look for them.

Honestly, I think this has to be the simplest original sculpt I've ever exported from Sculptypaint.

[Charolotte Caxton]

Hi Josh. May I have a sculpt as well?

[Charolotte Caxton]

---

Luc Starsider wrote:

---

Charolotte Caxton wrote:

Take a picture of a plywood sphere if you wish and lets see if we notice the same.

---

But now we are aware of it, so I think we'll notice anyway, plywood or no plywood.

.

-  Luc -

---

Now I am curious as to whether your viewer sees the spheres like mine.

[Josh Susanto]

I just sent it to Charlotte.

Please understand that it's only an example.