Basic smooth shading not working

Hi,
New scene, add sphere, set smooth shading, render in cycles (2.72) - gives incorrect shading. I’ve seen this before I think - so not new to 2.72.

Any ideas?

Thanks
NQL


Something to do with lighting falloff in regard to normals when using smooth shading. It’s doing ok on per-face basis, and somewhat on edges, but pretty obvious there’s not much weighting or averaging done with it at the vertices. So those corners really stick out in the transition to shadows.

I think there’s a similar problem with this in regard to reflective shaders, however with reflection it’s more obvious in the corner cases where theres a transition to sharp edges from rounded surfaces.

As far as I know, it’s been in Cycles since who knows when. I’m sure the developers are also well aware of it. As an end-user artist, just gotta hope they’ll get around to it sooner rather than later.

Now if somebody knows a work-around, then I’ll be watching for it too.

i tried it, and this flaw only occurs when lightning with emission planes or lamps. when i use just the world settings, for example with an hdri, i did get a correct render.

Thanks. It is strange to see such a fundamental thing like this last so long as a bug. It must show up in all renders, in some form.

It isn’t a bug, it is the well known termination issue. So common to pathtracers like octane, cycles etc etc
Try to subdivide more, or/and increase the size of lamp.
A nasty issue, try some other render engines, and, good luck.
welcome to the world of the so called “physically correct” render engines.
Please, try not to post conclusions like “this is a bug” or so. Not that easy.

yeah, it a patchtracers thingy.

you could use the compositor to separate all the render passes and add them all together and then blur/multiply the shadow pass into the image (some bleeding issues here), have fun! :slight_smile:

It is a bug and other render engines have fixed it. Come on we have to be able to call bugs bugs. It is the shadow terminator bug (I’ll be back!) or more appropriately (You’ll never fix me!)

One solution is to use Blender Internal and not use Area lights.

1 Like

@Atom
it isn’t a bug, rather a feature, we don’t have it. Cycles is not able for such tasks, like toon shading under certain lighting conditions, by example. Not so far. Pity. A must-to-have, not a bug. Some dirt is what we need, some hacking, considering the so called “physically correct” concept.

Of course it’s a bug. A “bug” isn’t necessarily “incorrect code” and I’ve no wish to insult the developers! However, if I set “smooth shading” on an object and I get the result above then that’s NOT the result I require and is, by definition, a bug in the process, if not the code.

However, if I set “smooth shading” on an object and I get the result above then that’s NOT the result I require and is, by definition, a bug in the process

I don’t like it, it is an ugly artifact. In many cases makes cycles useless. (see Toon shading issues)
However, by “definition”, it is not a bug.
A useless cycles, yes it is.

I had this problem for over 10 years and still have not found a good solution, if anyone got anything id love to hear it.

It’s not a bug, it’s a part of path tracing. Any corrections to it (as with Thea) are non physically accurate and therefore add bias and possible artifacting to renders. You can’t trace rays around a curved surface that isn’t actually curved. These shadows are absolutely what’s intended from raytracing shadows against low poly objects.

Any corrections to it (as with Thea) are non physically accurate and therefore add bias and possible artifacting to renders

Possibly, however this nasty thing acts like an artifact under certain lighting conditions.
See toon shading. Cycles may become completely useless under such conditions.
A rasterizer like BI is a better choice.

It’s remarkable to see the focus on semantics. I, like you, refuse to let go of the issue :slight_smile:
It is a bug - because a “bug” is, exactly, a general term to describe something that does not work as it should because the process and/or code is incorrect. You can call it something else if you like, but that’s just your term, not THE term. The point is that the problem is difficult to solve because (after reading about it) it seems the calculation required to ascertain the correct point on the virtual surface with which to calculate the shading, is non-trivial and is therefore approximated by the software, as “path tracing” dictates. But just because the entire process has a name, “path tracing”, does not immunise it from criticism as a system that contains bugs.

BI is pretty much on the way out. It has its uses, just like the horse since the automobile, right?

“Smooth Shading” is the major problem with this “shadow terminator bug cases”, I guess, and indeed “Smooth Shading” is one big old fake still available in cycles, for good reasons though.
Possible “Smooth Shading” need some adaption to physically correct pathtracer ?

https://www.youtube.com/watch?v=L0MK7qz13bU

@iceking - I see what you’re saying - she has incorrect shading on her chin at 44 seconds. Good catch.

i don’t see the incorrect shading.
But i have also just noticed this “not-feature” in my last render. never before.

horses are useless xD

To be blunt, horses need be killed if injured.