[blahblah4332]

As someone with a professional photography background, this article was actually super upsetting.

The author is like "look, this gas station one is the only one that worked".. well yeah, the camera's white balance algorithm saw that the scene was about 80% daylight, so it set the white point at around 3500k, the rest of the scene is recorded accurately as a result.

The thesis of the article "I don't understand white balance, so smartphone cameras are bad"

[perardi]

I know I should defend the common user…but c’mon, if you’re going to write an article like this, maybe spend like 5 minutes figuring out why the camera is “struggling”.

Broadly, the camera is doing the right thing here. Nobody developed these auto white balance algorithms to factor in the sky turning orange. The camera/imaging pipeline is seeing this crazy orange cast, assuming it’s deep incandescent light, and then trying to compensate, which is generally what you want it to do.

[blahblah4332]

Exactly! You could measure the collective eye-rolling of all the electronics engineers who work in the photo industry on seismographs around the world. (hopefully overstating the reach of this garbage article)

Edit: Not to mention that the auto WB algorithm out dates smartphones by more than a decade.

[Someone]

Also, before smartphones or even digital cameras, people sent camera rolls to laboratories, and got prints back.

If, in winter sport season, you sent your photos from the Sahara to be developed, chances were you got what looked like snow landscapes back. Conversely, sending winter sport photos in july/august might give you ‘beach with skiers’ photos back (I think that was because of operator error, not automation)

[perardi]

Most of my experience with film was long ago, and mainly through those execrable disposable cameras my family used when I was a kid. (https://en.wikipedia.org/wiki/Disposable_camera) I doubt there is a single indoor photo of me as a child with the proper white balance.

(Which is for the best. I was an extremely homely child. It's for the good of all we shot with slow, blurry film.)

[blahblah4332]

The fact that you had to add a wikipedia link for disposable cameras makes me feel old

[perardi]

A bit of disambiguation to make clear I wasn’t talking about Polaroids.

[SpicyLemonZest]

I think you're multiple steps ahead of the curve here. Most people think that an ideal camera captures things the same way you see them with your eyes, and any differences reflect a flaw in the camera. (I personally thought that was true until yesterday!)

[blahblah4332]

Yeah, I can agree with that. Before I got into photography I believed the same thing. In fact I was driven to it by the disparities in my own images when compared to professional examples.

The reason this article is so frustrating to me is that instead of the author highlighting what has been true since the dawn of photography - that cameras never truly capture the scene as we view it - they chose to baselessly attack a specific set of devices which happen to have cameras attached to them.

If this were to happen 10 years ago to someone using a Nikon Coolpix 1234Whatever, they would be equally disappointed, and someone using an Olympus Stylus somethingorotherD may have been very pleased. The color representation in images has always been the property of the medium's manufacturer, from Kodak to Polaroid and so on.

It would have been nice to see the author share information instead of just ignorant disappointment

[Leherenn]

Could you please explain for those not in the know?

I realise that why I see is just one "interpretation" of the reality, but it seems sensible that one possible ideal for cameras would be to perfectly match my perception of the world. Is it just impossible because the difference in perception is just so big between individuals that what would be perfect for me would be way off for someone else?

You can also create (at least theoretically, no idea how they compare currently) cameras that are better than human eyes at things like resolution, contrasts and so on, and that seems like another possible ideal; but which one is more relevant seems very context dependant.

[SpicyLemonZest]

The challenge here is that your perception of the world itself performs contextual color correction. If you've ever seen those optical illusions where you see two different colors but they're actually the same (https://www.syfy.com/syfywire/another-brain-frying-optical-i... is my favorite recent example), that's a demonstration of the effect. So a camera that's trying to match your perception of the world can't just identify the "true" color; it has to identify what color it should print on your screen to generate the same perception as if the contents of the picture were outside your window.

I wouldn't necessarily say it's a bad ideal, though. Smartphone cameras do a very good job at automatically performing all the right corrections in most circumstances. It's just nontrivial, and thus unsurprising that the normal algorithms fail in weird edge cases like "the entire sky is deep red".

[svachalek]

Smartphone cameras are not for people who know what white balance is, so that's a very solid thesis.

[blahblah4332]

Unfortunately this reality applies to _all_ digital cameras, not just the oversimplified versions found on many smart phones.

Also, with smartphone sensor/lens quality improvements as well as computational photography enhancements, smartphones are for _all_ picture takers. Point and shoot cameras are no longer relevant.

A better version of your reply is: Expecting accurate results in photographs is not for people who are ignorant of general photographic principles and technology.