DOF in reflections

[MANUEL_MOUSIOL]

Is it possible or ever going to possible to achieve this affect: a far away object is blurry through DOF in a focused reflection?
I am currently working on a render where I actually want to focus on the sky that is reflected on a car. In real life the car would be blurry and the sky sharp - one could also change focus and make the sky blurry and the car in focus.
As I understand it, right now Vray would can only focus on the reflective object and the reflections will be in focus too...

[andybot_cg]

It depends on if you are doing DOF in the render or in post. If you're rendering DOF, this effect is absolutely possible. Just set your DOF target point distant enough past your "mirror" so that the far away object is in focus.

[Raven]

DOF is raytraced so the reflections should be out of focus according to their apparent distance from the lens.

[MANUEL_MOUSIOL]

ok cool guys. I tried it out and it actually works as you suggested.
now how far is the dome light actually?
I have a camera with a 65mm lens and when having it open at 15 aperture things are relatively easy. BUT the effect is pretty subtle. now if I wanted the car to be more blurred, I would open the aperture, as I did to 3.5
But now comes the trouble and that is focusing. Maybe having set the distance at 1000000000.0km is crazy but there is some pretty heavy guess work to do with focusing this. and then comes the trouble that I would have to set the units to km to be able to set the camera right, maybe meters would be sufficient, but how far away is the sky in 3d space?
the example here is set to maximum km so that cant be right. clouds are probably 1-4km up, and (I dont have a render for that, sorry) when setting to 2000m they seem to be more in focus, but finding the sharpest point seems to be almost impossible....
any suggestions?

[PIXELBOX_SRO]

Id try the effect in somthing more obvious than clouds. Put a checker map in the domelight and start the tests from there?

[MANUEL_MOUSIOL]

Good call! Will do so !Thanks !

[MANUEL_MOUSIOL]

Here are the results. With a perfect mirror material I get the checker more or less sharp on about 400m, but some parts are blurry, which I guess is due to the fact that the dome is obviously round. Not sure if that would be the case in real life..with 200m I can get the checker farer away a little sharper, but they dont get really sharp...couldnt find a focus point for them. at -3,5m the car is sharp...

[joconnell]

So the math behind it is the distance between your camera and the surface you're aiming at, then add on the distance between the surface and the thing it's reflecting. So say you're aiming your camera at a mirror which is 20 metres in front and you want a sharp reflection of an object that's 100 metres in front of the mirror, it's 20 + 120 metres = 120 metres for your focus distance and you'll be sharp.

As for your example, some parts of the checker are sharp, some are soft because of the above equation. All if the bits of the checker being reflected are closer or farther away from your car - the middle bit might be 400 metres away and sharp but the top parts of the checker are further - say 420 metres away and the bottom ones are closer - say 380 metres away so they'll get a bit more blurry. Focus is only razer sharp at one specific distance from the camera, softer in front and behind that so the only way to get things sharper overall is change your aperture to something higher and less soft overall!

[MANUEL_MOUSIOL]

Sure, makes sense and I'm aware, but I think I never experienced that in real life, not as strong as this at least. Usually landscape and objects which are far away are in focus if you put your lens to the furthest point (infinite) or just in front of that. The clouds aren't that far away from each compared to objects closer to the camera . Probably it has something to do with the dome being around 400m away and not like in real life 1000-4000m (at least for the clouds)...

[andybot_cg]

I am also curious about what is the focus distance that V-Ray calculates between the domelight and the camera. One thing I noticed in regards to blurriness, it seems the DOF calculations start off as blurry and then have to resolve the image to sharp - thus it's not starting out sharp like when you have DOF turned off, but instead has to calculate back to sharp. Wondering if this may be the reason the reflection isn't perfectly clear.

[vlado]

The dome light is assumed to be infinitely far away (i.e. it's equivalent to environment background).

Best regards,
Vlado

[MANUEL_MOUSIOL]

From my current testing it doesn't seem to be. At the max distance it was blurry, but sharp at around 300m...

[andybot_cg]

The dome light is assumed to be infinitely far away (i.e. it's equivalent to environment background).

Best regards,
Vlado

I figured that would the logical answer, but like Manuel, I am also finding it is blurry with a large distance used for the focal point.

[vlado]

<shrug> It as accurate though. If you put an actual sphere that is sufficiently far away, it will look the same.

Best regards,
Vlado

[andybot_cg]

Well, not really. Here's a quick test with DOF and 10km focal point distance, and DOF turned off. Doesn't seem to matter, I tried 250M, 600M, they all give me about the same blurriness.