front focus wrong, DOF / focus distance and zoom

[mario74]

hello,

with a nice lens, f 1.8, DOF and focus distance "ON" and zoom faktor i get wrong front focus.

the top view of cam shows me the object is still in focused area of cam, while the rendering starts to blur already - so i am out of focus.

i have to reduce the focus distance to get sharp pic.

is it cause of using zoom factor? it looks to me, like this factor stretches the exact start (and maybe end) of DOF and make it unsharp when you are very close to an object...


regards, Mario

[mario74]

maybe with picture i will get some hint...

as far as i know, DOF is the "sharp" distance. but if i use "focus distance", DOF dont come out as cam shows me.

any suggestions, please?

[mario74]

testing this sitation here, focus distance changes DOF of cam right away. (in viewport)

logically this dont make sense, cause f-stop will always be the same with lens and zoom and or focus distance.

maybe only viewport shows wrong DOF of cam, but rendering is correct?

it starts to confuse me...

[tashko.zashev]

Hi,

Would it be possible to post the scene file to see the whole scene setup?
You can replace your models with simple geometry.

[etown]

I'm experiencing odd focal distance issues with the movie camera as well. I recently updated to the latest stable build and I have to turn on and off the targeted box in the camera settings to get the depth of field to behave correctly. This seems like a bug -- anybody else have issues with this?

[mario74]

hi Tashko,

i attached a simple file with the cam and added one light, otherwise it would be dark.

as you can see, the cylinder should be in DOF, but it turns out blurry already.

[Svetlozar Draganov]

maybe with picture i will get some hint...

as far as i know, DOF is the "sharp" distance. but if i use "focus distance", DOF dont come out as cam shows me.

any suggestions, please?

[ATTACH=CONFIG]14345[/ATTACH]

Thank you very much for the scene.
Unfortunately the distance between front and back plane of the camera doesn't represent the Sharp Distance and such distance cannot be shown.
VrayPhysicalCam like the real camera could only be focused only at one distance at a time and every object away from that plane is gradually blurred more or less depending of the F-number / Focal Length and the distance to the focus plane -there is no save(sharp) volume.
The front and back camera planes are just approximated representation of the focus distance - but they couldn't guarantee that the objects inside will be always 100% sharp or the objects outside will be always out of focus.

[mario74]

Hi Svetlozar,

if i position the cam with zoom = 1, than the DOF shows right, and renders right (depending on focus distance) as far as i can see.

the usage of both, zoom & focus distance gives strange result. you may not recognize, if you use low zoom around 1,2 or 1,5 and you are not close to some object.

the factor zoom, should zoom in object and be used as factor which opens the DOF in same scale as object gets zoomed in. (i dont change focus distance here!)

i will setup a easy scene with some cams, so you can switch between them according to this calculations here:

http://www.dofmaster.com/dofjs.html

regards, Mario

[mario74]

well, lets go: focus-test for VrayPhysicalCamera!

for calculation i used a Canon 5D MarkII, crop-ones turn out almost same result, but why to use here

i renderd in 2400 pixel width. for exact measuring limits, you may render in bigger resolution but for now it should be enough.



as i supposed, everything fine with zoom = 1





top-view still perfect with zoom = 3, rendering turns out wrong





rendering is out of range with zoom = 6, front- and backfocus doesnt turn out clear anymore, but near and far limit of cam is still correct!

[mario74]

attached the high-res ones

[veselin.georgiev]

Hi,

I'm Veselin Georgiev, one of the developers from Chaos Group.

Your observations that the near and far limits of the camera do not correspond exactly to the places where visible blurriness sets in are correct. However, just deciding on what is blurry is a complex matter and cannot be decided in a discrete - "this is blurry and this isn't" - way. Perception of blurriness encompasses a whole lot of other parameters: including the geometrical complexity, the image sampler being used, the number of DOF subdivisions, DOF shape, what type of monitor are you using to view the pictures, its PPI, the conditions of your eyes, image brightness and so forth. Capturing all this in a simple pair of camera front-back focus planes is impossible!

It can also be argued that the visual determination of where the blurriness sets in is purely subjective.
One of the things is that, on the zoom=6 render, you assess the back focus to be shorter than the front-focus interval, which is geometrically impossible. The front focus interval is always shorter, and if it seems it isn't, that is just a perception thing. Or an optical illusion, if you will.
Also, given that the distances between the camera and the target are different in the three renders, and thus the perspective is different as well, the zoom > 1 renders have a longer interval where the blurriness gradually increases. Humans have the tendency to put the blurriness "threshold" earlier in that scenario.
All in all, if in the zoom > 1 scenarios the front-back planes seem to be misplaced, then they are probably just as bad in the zoom = 1 case, and by the same factor, but it's not that visible.

So anyways, as Tashko mentioned, the front-back planes exist only to give you a very approximate idea of the DOF. You have to experiment, more or less by trial and error, to find your best combination of focal length, f Number and proximity to the subject.

[mario74]

Hi Veselin,

thanks for statement!

All in all, if in the zoom > 1 scenarios the front-back planes seem to be misplaced, then they are probably just as bad in the zoom = 1 case, and by the same factor, but it's not that visible.

So anyways, as Tashko mentioned, the front-back planes exist only to give you a very approximate idea of the DOF. You have to experiment, more or less by trial and error, to find your best combination of focal length, f Number and proximity to the subject.

i just rendering the 500mm-cam with zoom = 1, DOF-subdivs 100, width 8400 pixel. it seems to look exactly the same as before with zoom = 6.

the point, that confuses me: the near and far limit seems to fit perfect at top-view and this makes it easy to adjust camera with DOF as i need. trial and error is in this case not a comfortable option, because to find the perfect DOF includes shutter-speed also and take away the feeling you already have while using a DSLR.

i could not achieve promising results in post-work, thats why i am using DOF out of Vray almost every time.

are there any plans in future, to avoid trial and error?

regards, Mario

[veselin.georgiev]

Hi Mario,

Trial and error is in this case not a comfortable option, because to find the perfect DOF includes shutter-speed also and take away the feeling you already have while using a DSLR.

Could you elaborate more on this - I don't quite grasp what you mean by including the shutter speed?


As to what can be done to avoid the lengthy trial and error process - you can setup your scene to make it quicker - just select a few very small regions of your image ("focus points", if you will), and setup V-Ray to render them only. This way you can quickly assess the blurriness levels at various points. You can do it with RT to make the process even faster.

Best regards,
Veselin