WOW...ISO 4,000,000 and 250 MEGAPIXELS ? !
CanonUSA claims they're working on see-in-the-dark super-high resolution sensors with effective high ISO up to 4,000,000! In their video below CananUSA mentions their sensor work in the realm of:
While we don't normally get into the DSLR weeds here, we are, though, knee deep into the two most limiting aspects of mobile photography...namely:
Here's their short video. Check out their dark(!) night scenes. We hope market forces (i.e., $) will bring these advancements to smartphone mobile cameras soon.
The article "ISO 4,000,000 and 250 MEGAPIXELS" first appeared at ProMobile.Photo.
EVERY THING YOU NEED TO KNOW ABOUT ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING
Here is all you need to know about three simple questions:
These days, nearly all advertisements for new smartphone cameras include claims of innovation in artificial intelligence, machine learning, convolutional neural networks, tensor flow analysis, etc.
Adding to our joyful bewilderment, some of these ads are simply misleading. Some of these ad's offer exaggerations of the extent to which your hard-earned dollar is buying substantial new capabilities for you to improve the quality or convenience of your photography. In other words you should be able to answer this simple question: "If I had bought this camera-'X' one year ago, how would it (could it) have improved _my_ photos, in the way _I_ capture them?"
Here are some examples of the jargon terms from recent advertisements (and supposedly independent) product reviews:
Here's two major things you should consider in sorting all this out. First, while the mathematics, physics and computational statistics underlying these AI concepts and implementations present a bewildering display of Greek letters and special symbols, the simple question is, "How, specifically, does artificial intelligence or machine learning actually improve my smartphone camera?"
We simply want our smartphone camera's artificial intelligence algorithms to perform a few straightforward end-goal tasks that many infants, toddlers and adolescents normally learn by an early age. And if we separate child-learning skills into the objective level (facial recognition, object-differentiation, food identification,...) most of us have known at least one infant who has mastered these mechanisms for triggering "adult action". The more qualitative camera/software artificial intellilgence skills (judging image quality, facial beauty, animal threat, human emotional state,...) parallel the development of skills in humans that can span several (or many) years.
As for the first question above, namely: "How do artificial intelligence, machine learning, neural networks, etc. actually help my smartphone camera perform better?"... Here are some ways that artificial intelligence and convolutional neural network algorithms are now being used in our mobile smartphone cameras:
Still, the second major question about artificial intelligence algorithms in smartphone cameras remains. That is: "Even if artificial intelligence helps my smartphone camera handle images quickly and accurately, how much will it improve the quality and appeal of my final photographs?"
Enormous advances have been made in artificial intelligence theory and groundbreaking applications since I took my first graduate course in artificial intelligence as part of a master's degree program in physics and computational statistics at the University of Nebraska way back in the1970's. But many of the fundamental concepts and principles remain the foundation of artificial intelligence and it's many component spin-offs since then.
When it's all said and done, here are some "smell test" criteria you should look for, beyond advertisers' claims and in truly legitimate, hands-on product reviews, etc. when asking the question: "How will the quality and appeal of my own photographs be improved by artificial intelligence in my smartphone camera?":
Now, for the third and final question to ask about what you should know about your camera's artificial intelligence: "How can I make sure I'm taking maximum advantage of my camera's artificial intelligence features?":
Do you have additional thoughts on this topic? How has artificial intelligence algorithms in your smartphone camera app improved your photography? Which camera features would you like to see introduced or improved by AI?
This ariticle "Everything You Need to Know About Artificial Intelligence and Machine Learning" first appeared at ProMobile.Photo.
WHAT IS "PIXEL BINNING", and WHY YOU CARE
Pixel Binning is a collection of techniques for combining multiple (nearby) pixels into a single "super pixel." This results in a pixel that, in effect, derives from the light input of a larger pixel. What this accomplishes, in effect, is pixels whose each individual size (area) is closer to the size of a much larger camera sensor. For example, a smartphone sensor that simulates the larger pixels found in DSLR cameras. However, the trade-off is that a 12.4 megapixel sensor on a smartphone that combines 4 native pixels into a single super-pixel now has a resolution closer to 3.1 megapixels (1/4th of 12.4). (If that's all the explanation you want, then check toward the end of this post for why you even care about all this...and I think you would!)
There are many variations of how to select which 4 (or 2 or 6 or any number of) pixels to combine into a super pixel. In the diagram above, the red pixels on the outer corners of a 3x3-pixel grid are combined into one quadrupal-sized super pixel. Some of these algorithms can beneficially reduce some of the random color-dots of noise characteristic in pushing small sensors to handle lower luminance (night, indoor) scenes. However, new noise artifacts can result from some of these combinatorial algorithms. Many scientists have investigated methods for pixel selection, combining into super pixels, and even post processing in specialized editors/filters. Kodak's Pixelux is one such approach.
Why do we care? It re-teaches us something that is counter-intuitive to most photographers, yet it's quite important on a very practical level of selecting our next camera.
In the ad-driven marketing arms race among smartphone camera producers, the total megapixels of a given phone's camera has emerged as the defacto popular bragging point. But, just imagine, subdividing that teeny little sensor that already is divided into 12,000,000 (12megapixels), or even 24mp, then the ad guys tell the phone-design engineers: "Hey, guys, you gotta give us a phone with a 30mp(!) camera, so our ad's can beat all the other phone producers....Maybe even up our DxO score a notch."
Frankly, I would much rather have a smartphone camera that has the option to shoot great 4mp night shots natively, rather than a 16mp camera that works best only in bright daylight.
So, you see, we really do care what pixel-binning is, and now it's clear why it's important to us. And that's not just some abstract dream, because next month LG will be releasing their new LG V30S and LG V30s phones with (get this...) a Bright Mode that shoots real nice night photos at 4 megapixel "pixel-bin" enhanced alternatives to its regular 16 megapixel camera.
12 BEST WAYS to LEARN and GET STARTED in PHOTOGRAPHY
If you, or someone you care about, would like to learn my 12 Best Ways to Learn and Get Started in Photography, then read on. There's even some added bonus tips toward the end. People, mostly young people, occasionally ask me "how to get started as a photographer." Perhaps the best single answer is "The best way to be a successful photographer today, is to be almost successful as a photograher yesterday." What that means (to me) is, don't try to figure it all out ahead of time. Just get started early, and do it often. Practice, practice, practice. And one of the best tips for me was, get expert helpful advice on what's not just right with your photos, and not just praise from supporters. Before posting an easy "Top 12 Ways to Become a Photographer," here's another couple dimensions that I have found universally important. First, how "good" I am in the eyes of someone else is more important than how good I alone think I am. Sure, there's the several photos I take just for personal enjoyment. But as a commercial photographer, it's not what _I_ like, so much as, get this, it's important...it's what's important in the eyes of the consumer/buyer whom my business customer is trying to engage through an advertisement or branding banner or such. The point is, "shoot with a purpose." And decide before you ever point the camera, who is the ultimate most important judge of whether this photo will be "good" (i.e., successful, valuable, enjoyed), and try to begin the shot's journey from capture to editing, to production to delivery to receipt and, finally, to some desired action or mind-state by the true end-consumer. A second dimension...respect the display and receiving visual systems...their size, strengths and quirks. Will it be displayed on paper? On a desktop screen? Mobile phone screen? Billboard or banner? Framed on a gallery wall? There's a whole series of books that could be written (and learned) on this subject alone. And don't get me started on the neuro-physics (whatever that is) of the human visual system. It's truly exciting (to me, at least) that with all this complex jigamabob's, artificial IQ algorithms, add-on doodad's and light sources, it all winds up pretty much useless until some human's eyeball glances upon our work, and that ball bounces the received photon impression off to some cortex-whatever enroute to a human brain of some personality, sensitivities and capabilities. Outside of an occasional venture into direct-to-dog bottom-shelf pet shop display, ultimately no photograph is of real value until some human's eye sees our work, and that human's brain and personality processes it. So, my rule is, start early to learn as much as possible about display screens, printing technology, human-eye dynamic range, color sensation, etc. I suppose we have yet to discover how this whole realm of aritificial intelligence, machine learning, tensor flow, convolutional neural networks, neural processing units, and even Google's Visual Core competing with Secret Apple sauce will assert itself between the human brain and us. In fact, one of the famous scientists, whom I admire (albeit, perhaps now question a bit), recently predicted that during this century "photography" will bypass the human eye enroute to the brain. (Wowser, add that to the fact I've seen some cases where my photography more than once, even bypassed the "brain" part.) Anyway, in response to a recent request for my advice on "12 BEST WAYS to LEARN and GET STARTED in PHOTOGRAPHY," here is my response.
With the “Digitization” of photography and concomitant significance of in-camera software, breakout emergence of the many developing realms of in-camera artificial intelligence algorithms, and increasingly rapid evolution of consumer expectations, “photography” has become one of the most wildly evolving technologies ever invented. Therefore, always keep in mind: “You don't ‘learn’ photography... You chase after it and try to keep up.”
This article, "12 BEST WAYS to LEARN and GET STARTED in PHOTOGRAPHY" first appeared on the blog "Latest News and Next" at ProMobile.Photo.