[packetslave]

I wouldn't call it a "trick" but I suspect Apple is doing some focus stacking under the hood.

The image pipeline in the iPhone gets more and more advanced with every IOS release. The state of the art for computational photography is pretty amazing.

[dylan604]

If you're doing something under the hood without disclosing it but calling it something else, it's a trick. It doesn't have to be nefarious, but you are tricking the user into thinking they are doing something which they are not actually doing.

[packetslave]

I see where you're coming from, but here's the thing: every single photo taken on a modern phone goes through a pipeline of 10+ stages of image processing, including multiple exposure merging. The iPhone isn't even taking a single photo... it's picking the best frame(s) out of a running buffer of video.

(unless you shoot in raw, and I haven't read enough yet to know what processing, if any, Apple does to ProRes/RAW photos).

Call it "tricks" if you want. I call it using technology to give the non-professional camera user the best photo possible at the time. If that bothers you, pull out the DSLR, shoot in RAW, and spend time afterwards in Photoshop/Lightroom.

edit this is also why apps like Halide (or ProCamera or Filmic Pro) exist... if you want to control more of the options instead of letting Apple choose, the capability is out there. Most users probably don't care. They just want a good photo of their kids to post on Instagram.

[marcellus23]

How are they not disclosing it? They talk at length, every single keynote, about all the software they build to process iPhone photos.

[dylan604]

Do they really? I only ever hear about how its the best iPhone camera they have ever made. Yes, they talk about how their AI is able to adjusts exposure, color tone, etc. Do they actually talk about how a macro is taken in such detail as focus stacking etc? I tend to nod off during these videos, so I might have missed something.

[marcellus23]

> Yes, they talk about how their AI is able to adjusts exposure, color tone, etc. Do they actually talk about how a macro is taken in such detail as focus stacking etc?

Is that moving the goal posts a bit? Your point was that Apple is somehow pretending they're not using software to create this effect -- if they say they're using AI/etc. to do it, it seems to me like they're not misleading their customers. I don't think the requirement is that they explain exactly how it works in technical detail.

I don't think they did any sort of deep dive on their macro tech, but in the past yes, they have gone into some detail about how their image processing works behind the scenes.

[dylan604]

There's a difference in macro being 1:1 ratio (as understood in photography) vs having an AI do something to make something small look big in the image. So if their lenses are not delivering 1:1 but use some sort of algo, then it's misleading.

[macintux]

There are no shortage of lenses that are labeled macro but top out at 1:2, or even 1:4.

[packetslave]

They tend to have sessions at WWDC about the underlying camera pipeline, for developers who care. Product launch keynotes are the wrong forum for deep technical talks.

[dylan604]

that's not what the person I responded to implied. they specifically called out the keynote

[crancher]

The do indeed call it out in the keynote, including a slide with graphics indicating what each stage does.

[renewiltord]

DSLRs use extensive such tricks to process the image. Enthusiast groups will show the pathological cases for the image processing that happens. Ultimately, ceci n'est pas une pipe applies to the picture. It's a representation and it is meant to evoke the scene.

[dylan604]

What DSLRs use AI or other software based tricks to perform macro photography?

[renewiltord]

Name a DSLR and it uses software based tricks to perform all photography.

[dylan604]

At the risk of feeding the trolls, no DSLR software comes close to the image manipulation of a modern phone. We know this with tools like MagicLantern that would have enabled this type of stuff.

To even equate these as being the same is being disengenious at best.

[passivate]

Specifically, what is 'state of the art' in the iphones image pipeline? Many of these computational techniques (focus stacking, image averaging, subject recognition, face recognition, etc) have been in use for decades - but just not in a handheld device.

By the same token, I wouldn't call Google docs as 'state of the art' since its duplicating existing desktop software, but in a browser (not to minimize their effort, I'm sure its hard).

[packetslave]

You said it yourself: "just not in a handheld device" and not in what is effectively realtime.