> There were jokes aplenty about the ultra-wide distortion on the camera, but this is all a fairly cheap jab at what is really a very impressive bit of software.
No, you know what, it isn't. I'll just say it, it's not impressive software at all, and the fact that they can't apply simple lens correction is a little embarrassing. In high school, I worked in a FIRST robotics club where we had to design, program and test our own computer vision models on robots, which would need to complete various tasks to varying degrees of precision. To that point, I have seen high-school Juniors write a program that can track your face in Python, it's really not some sort of mind-blowing, next gen thing.
Overall, I'm starting to really hate these Halide reviews. Their lifeline is based on the success of Apple hardware, why should they consider themselves a credible and unbiased source in these reviews? It's glaringly unprofessional at best, and downright cringe-inducing at worst.
You're confusing ultra-wide rectilinear lens distortion with barrel or fisheye distortion due to lens construction limitations. The latter is correctable in software, whereas the former is what they're joking about.
> it's really not some sort of mind-blowing, next gen thing
Sure. But why has no-one else implemented it in a selfie camera yet? I'd put $ on this camera being a major marketing point for either this or next year's iPhones.
If that's the case, what can be done about reducing that distortion? My layman's understanding is that it's like mercator projection - you simply cannot make a two-dimensional photo of a wide field of view without distorting.
And that’s exactly what the face tracking feature does. That shot is a still from the wide angle cam, and not in a framing you’d normally use for video calls.
These are indeed impressive but one thing I know for sure is that from the iPhone 7 onwards I never gotten a single picture that looks nearly as good as anything that these professional review sites get.
Phones are now as tricky as DSLRs to fully exploit.
Of course, if you're not a professional you'll never get as good photos as a professional. A lot of what makes a picture good doesn't have anything to do with the camera itself to do. Composition has a major affect on how you perceive a photo.
Also having the right subject matter at the right time. Wander around Dallas at noon and you probably won't get many great shots. Denali National Park at daybreak? Might be hard to not get a good shot.
Yes, it can definitely uplift an image. But if you change too much, then you have very little of the original image - and at that point - is it a photograph or is it just digital art?
But to be sure, post-processing is an optional step, not a requirement - and in some cases - its banned (e.g. photo journalism).
When it comes to programming impressive software, choice of language is a greater factor than the computing platform which is a greater factor than the IDE.
Language > Platform > IDE
Using a good programming language is the primary differentiator.
(Alright gatekeeping elitist programmers, pick a different battle, pointing out the value of skill, experience and technical knowledge is not a productive addition to this reality, thanks.)
What I offered was an (admittedly strained and tedious) analogy to demonstrate how utterly strange your post was. Your prior parenthetical was nothing more than poisoning the well against responses you presumably expected were coming.
It’s a bad habit fairly unique to enthusiasts photographers
I think it’s fine to nip it in the bud, just lets them think twice about responding as they’re not usually disagreeing about the academics or state of technology, just getting into some vendor war
The fact that you think photographers might have responded with “some vendor war” strongly suggests that your assumption about what photographers would take issue with in your post was seriously misguided.
28 comments
[ 2.9 ms ] story [ 73.9 ms ] threadNo, you know what, it isn't. I'll just say it, it's not impressive software at all, and the fact that they can't apply simple lens correction is a little embarrassing. In high school, I worked in a FIRST robotics club where we had to design, program and test our own computer vision models on robots, which would need to complete various tasks to varying degrees of precision. To that point, I have seen high-school Juniors write a program that can track your face in Python, it's really not some sort of mind-blowing, next gen thing.
Overall, I'm starting to really hate these Halide reviews. Their lifeline is based on the success of Apple hardware, why should they consider themselves a credible and unbiased source in these reviews? It's glaringly unprofessional at best, and downright cringe-inducing at worst.
> they can't apply simple lens correction
You're confusing ultra-wide rectilinear lens distortion with barrel or fisheye distortion due to lens construction limitations. The latter is correctable in software, whereas the former is what they're joking about.
> it's really not some sort of mind-blowing, next gen thing
Sure. But why has no-one else implemented it in a selfie camera yet? I'd put $ on this camera being a major marketing point for either this or next year's iPhones.
Do you know if my assumption is incorrect?
If that's the case, what can be done about reducing that distortion? My layman's understanding is that it's like mercator projection - you simply cannot make a two-dimensional photo of a wide field of view without distorting.
Phones are now as tricky as DSLRs to fully exploit.
But to be sure, post-processing is an optional step, not a requirement - and in some cases - its banned (e.g. photo journalism).
Take a little wet'n'dry screen cleaning wipes to the lens cover and you'll suddenly get way way better photos.
Until you put the phone in your pocket just once, and suddenly you'll be back to the not so great photos.
Lighting > Lens > Sensor
Recognizing good lighting conditions is the primary differentiator.
(Alright gatekeeping elitist photographers, pick a different battle, your edge case is not a productive addition to this reality, thanks.)
Language > Platform > IDE
Using a good programming language is the primary differentiator.
(Alright gatekeeping elitist programmers, pick a different battle, pointing out the value of skill, experience and technical knowledge is not a productive addition to this reality, thanks.)
Not a valid path to constructive discourse.
I think it’s fine to nip it in the bud, just lets them think twice about responding as they’re not usually disagreeing about the academics or state of technology, just getting into some vendor war
It’s not complicated.
Although, sure, the lid is thinner too.