Group f/64 • Hacker News Board

Re: Group f/64

#21

Post by m463 » Sat, May 04, 2019, 1:36 AM UTC

kazinator wrote: post #19

m463 wrote: post #18

Earlier quoted context omitted.

On the other hand, many games simulate lens flare too. I kind of wonder if there are not other "actually realistic" avenues that haven't been pursued. For example, simulating floaters. I also found it funny playing some first-person shooters in 3d and/or VR. Although it would make sense to have these effects when using a sniper rifle, you can use a sniper rifle without closing one eye! ha!

If a subject has floaters, they will be there when viewing a VR scene with goggles, and will move in the right ways when moving the head and eyes; there is no need to waste cycles on that.

Oh I totally agree, but I'm just highlighting what's actually realistic vs what's really just artistic license.

Re: Group f/64

#22

Post by trm42 » Sat, May 04, 2019, 6:01 AM UTC

w0mbat wrote: post #13

ƒ/64 may give the maximum depth of field (if that's your smallest stop), but due to diffraction you generally get higher image quality at ƒ/8 or ƒ/11.

Now you’re comparing different sensor/negative sizes against each other. Usually ”full frame” (aka 135mm) diffraction starts to affect on smaller f-values (larger aperture) but on larger negative sizes the diffraction starts to be a problem on bigger F-numbers (smaller apertures). On 4x5 format something like f/22 and on 8x10 f/45.

Re: Group f/64

#23

Post by CydeWeys » Sat, May 04, 2019, 12:32 PM UTC

w0mbat wrote: post #13

ƒ/64 may give the maximum depth of field (if that's your smallest stop), but due to diffraction you generally get higher image quality at ƒ/8 or ƒ/11.

As pointed out elsewhere in this thread, diffraction depends on the physical wavelength of visible light and the absolute, not relative, size of the aperture. f numbers are relative. A really big lens can be f/64 and still comfortably above the diffraction limit.

Re: Group f/64

#24

Post by CydeWeys » Sat, May 04, 2019, 12:34 PM UTC

m463 wrote: post #18

kazinator wrote: post #15

Ironically, an ultra high depth-of-field isn't realistic. In computer graphics, simulating low depth of field (background and foreground blurred, while subject in focus) is a considered an advancement in realism.

On the other hand, many games simulate lens flare too. I kind of wonder if there are not other "actually realistic" avenues that haven't been pursued. For example, simulating floaters. I also found it funny playing some first-person shooters in 3d and/or VR. Although it would make sense to have these effects when using a sniper rifle, you can use a sniper rifle without closing one eye! ha!

You can use gun sights in the real world while keeping both eyes open (and some schools of thought recommend it for situational awareness). It's not even that hard so long as your dominant eye is the one looking through the sights, and you don't get wink fatigue either.

I suspect the VR game isn't correctly simulating what this actually looks like, though.

Re: Group f/64

#25

Post by CydeWeys » Sat, May 04, 2019, 12:37 PM UTC

kazinator wrote: post #15

Ironically, an ultra high depth-of-field isn't realistic. In computer graphics, simulating low depth of field (background and foreground blurred, while subject in focus) is a considered an advancement in realism.

In real life, whatever I look at is always in focus. It's actually very unrealistic to simulate depth of field without good eye tracking, because the user will end up looking at things that are simulated as out of focus but should actually be in-focus because the user is looking at them.

If you don't have good eye tracking, then rendering the entire scene in focus is most realistic, as whatever I look directly at will appear in focus and everything else is more peripheral and will be blurred a bit by my optical system anyway.

Re: Group f/64

#26

Post by giobox » Sat, May 04, 2019, 3:02 PM UTC

w0mbat wrote: post #13

ƒ/64 may give the maximum depth of field (if that's your smallest stop), but due to diffraction you generally get higher image quality at ƒ/8 or ƒ/11.

Virtually no lens that hits peak sharpness at f8/f11 (usually implies a lens with an imaging circle covering the 35mm/APS-C format size) will even stop down to f64, with virtually all of them the smallest aperture size will be f22.

These chaps where dealing with much larger formats and their corresponding larger lenses typically, the usual rules of thumb you might be familiar with from the majority of digital interchangeable lens systems or 35mm film don’t always apply. f64 would be fine on 8x10, as one example, and was often used on this format.

Re: Group f/64

#27

Post by giobox » Sat, May 04, 2019, 3:21 PM UTC

kazinator wrote: post #15

Ironically, an ultra high depth-of-field isn't realistic. In computer graphics, simulating low depth of field (background and foreground blurred, while subject in focus) is a considered an advancement in realism.

It’s arguably more typically considered an advancement in the realism of simulating a real life camera with an optical lens. Not the same thing as simulating human visual perception more realistically.

Games are trying to look more like movies, hence new technologies to simulate shallow DoF, lens flare, anamorphic lens artifacts etc. None of these really help make the game world look more like how a human might perceive them, they do all help make the game look like how a typical movie camera might have perceived the scene.

For sure shallow DoF simulation _could_ be used to try and make games more human perception like, but I don’t think I’ve ever seen the effect used this way - it’s virtually always used to simulate optical bokeh/out of focus areas to look more film like.