Bug Report - DOF

Fri Feb 01, 2013 10:28 pm

I was playing around just now and ran across a major bug with the DOF. Take a look at these two shots. The shot called "Blurry Ledge" has the focus in the background yet the reflections in the window are wrong. Those windows should be equally blurry. The inverse (In Focus Ledge) is wrong too. The reflections in that one are blurry yet the glass is in focus. So the reflections should be sharp too.

This really needs to get looked into.

Any thoughts?

Thanks! Still loving the Octane.

Alec

t_3 · Fri Feb 01, 2013 11:13 pm

hm, but that's like it actually works? the light rays travel through the mirror into the camera, so the object in focus will stay focused, even if the mirror is closer to the camera, at least if the distance object -> mirror -> camera is not very different from object -> camera...

t_3 · Fri Feb 01, 2013 11:17 pm

note: this might only look more extreme, thus a little distracting in a rendered image, using flat mirrors of course, since they are _perfectly_ flat, what you will never find in real world...

Sat Feb 02, 2013 2:40 am

t_3 wrote:hm, but that's like it actually works? the light rays travel through the mirror into the camera, so the object in focus will stay focused, even if the mirror is closer to the camera, at least if the distance object -> mirror -> camera is not very different from object -> camera...

Ah, I'm not buying it. In the in focus example the camera has the ledge and nearby elements in focus. Including the glass. The Octane render is blurring the reflection. Why? The blur in the background is happening in the camera. Not the environment. So the camera should have crisp reflections. Right?

It's as though the info getting to Octane's camera is post blur. Can someone explain why I'm wrong here? I would love it if I were dumb and Octane is right.

Thanks!
Alec

Sat Feb 02, 2013 2:58 am

ASyme1 wrote:I was playing around just now and ran across a major bug with the DOF. Take a look at these two shots. The shot called "Blurry Ledge" has the focus in the background yet the reflections in the window are wrong. Those windows should be equally blurry. The inverse (In Focus Ledge) is wrong too. The reflections in that one are blurry yet the glass is in focus. So the reflections should be sharp too.

This really needs to get looked into.

Any thoughts?

Thanks! Still loving the Octane.

Alec

Nope, you got it wrong and t_3 got it right. Try it out with a camera

For example: Take a photo of yourself in a mirror. If you are in focus, the mirror is not and if the mirror is in focus, you won't be.

If Octane would calculate DOF in post it would be as you said, i.e. reflections on nearby objects would have the same "blurriness" as the reflecting object. Since we don't do it in post but via ray-tracing it's behaving correctly.

Cheers,
Marcus

Sat Feb 02, 2013 3:49 am

The correct way to think of a mirror is like it is a window into a new world that is equal to the "real world", just flipped. So the rays will travel through the "window" and find focus spots inside the new world. The focus spots inside the new world will be at the exact same travel distance as in the real world. Please look the following examples i just too with my camera:

Scene:

Camera in front of mirror:

Mirros is blurred, reflection is not.

t_3 · Sat Feb 02, 2013 4:44 am

abstrax wrote:If Octane would calculate DOF in post it would be as you said, i.e. reflections on nearby objects would have the same "blurriness" as the reflecting object. Since we don't do it in post but via ray-tracing it's behaving correctly.

Cheers,
Marcus

and it should be noted, that this is one (of many

) points what's so cool about octane: real world dof with a negligible performance cut, with a glimpse into those areas which are obscured by foreground objects (=like with a real world camera)...

now we'd only need some bokeh - if this is possible at all

@nagboy: great example!

Sat Feb 02, 2013 7:42 pm

abstrax wrote:
ASyme1 wrote:I was playing around just now and ran across a major bug with the DOF. Take a look at these two shots. The shot called "Blurry Ledge" has the focus in the background yet the reflections in the window are wrong. Those windows should be equally blurry. The inverse (In Focus Ledge) is wrong too. The reflections in that one are blurry yet the glass is in focus. So the reflections should be sharp too.

This really needs to get looked into.

Any thoughts?

Thanks! Still loving the Octane.

Alec
Nope, you got it wrong and t_3 got it right. Try it out with a camera For example: Take a photo of yourself in a mirror. If you are in focus, the mirror is not and if the mirror is in focus, you won't be.

If Octane would calculate DOF in post it would be as you said, i.e. reflections on nearby objects would have the same "blurriness" as the reflecting object. Since we don't do it in post but via ray-tracing it's behaving correctly.

Cheers,
Marcus

Well, I'm very happy to be wrong on this one. It's way easier for me to be wrong than fix some major code. Thanks for all the explanations on this. I've got to do some reading on this. Still isn't making sense to me.

Thanks all!

Alec