Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: dreed on October 14, 2018, 09:05:14 am
-
The latest story on this website is about "AI" image processing of Topaz plugins where their calibration and algorithms have been trained on millions of images. Elsewhere on the web is a story about Skylum Luminar with a new "Sky Enhancer" plugin that is again trained on AI.
Both of these developments are significant steps forward in what computers to do to make images "better" (not necessarily more realistic) and cut down on humans needing to know complex Photoshop techniques.
Who is the market? Well everyone that takes a photo and wants it to look like some pro's photo.
If this is anything to go by then the technique part of post processing by professionals is under serious assault and within a year or two, it will be a $2.99 app you can get on your phone.
Or will it?
-
No app will replace the intellect and the artistry that goes into a fine photograph; apps assist with the appropriate reproduction of the photographer's vision and can make a useful contribution if used judiciously, again an act of human intervention.
-
Both of these developments are significant steps forward in what computers to do to make images "better"
Are they ? Nothing in AIClear that convinces me of it. Just marketing using the latest buzz word.
-
No app will replace the intellect and the artistry that goes into a fine photograph; apps assist with the appropriate reproduction of the photographer's vision and can make a useful contribution if used judiciously, again an act of human intervention.
Ha. I worked in print media. A long time ago the 4 color strippers said we will never be replaced. :) The other day I told my friend that his 2 year old granddaughter may never touch a steering wheel. I do get what you saying. AI will be able to create a perfectly exposed, etc image. I will be also be creative (or is already) - depending on what data is pumped into it. I don't know if self awareness is even possible. Hope not.
-
Ha. I worked in print media. A long time ago the 4 color strippers said we will never be replaced. :) The other day I told my friend that his 2 year old granddaughter may never touch as steering wheel. I do get what you saying. AI will be able to create a perfectly exposed, etc image. I will be also be creative (or is already) - depending on what data is pumped into it. I don't know if self awareness is even possible. Hope not.
Until AI is telepathic, it won't replace creativity. That's not going to happen in what remains of my lifetime, so I am not concerned.
-
Some say we are a hologram or just a advanced version of the Sims :)
-
Some say we are a hologram or just a advanced version of the Sims :)
advanced?
Jeremy
-
I don't know. I just seen it on TV once. Not sure which network because some provide shady information. I don't if the presenters were accredited or not.
-
If you do a search there is quite a bit on it. This fellow appears to be legit.
https://bigthink.com/think-tank/theres-a-20-chance-were-all-sims
-
Until AI is telepathic, it won't replace creativity. That's not going to happen in what remains of my lifetime, so I am not concerned.
https://www.thedrum.com/news/2018/05/18/ai-artist-behind-art-latest-bloomberg-cover
-
I am not worried. (Also, I am not a professional photographer, but AI affects my profession too.) Roughly, AI processing seems to be "mimicking the most common processing choices, as indicated by observing how a large number of people handle a situation"; a glorified "average/typical choice".
- Why should anyone expect to be paid good money to reproduce "average" or "typical" processing choices?
- Why should one consider mere competent execution of routine actions as equivalent to talented, artistic decision making?
It sounds a bit like "pattern exposure metering": a sensible time saving way for anyone from beginner to professional to handle routine situations without complications like making multiple spot metering measurements, with the professional's craft involving knowing when and how to turn off the automation and make less routine, more skilled artistic decisions.
-
Oil and water are at times considered the biggest resources but data is right up there. My wife read that while Uber is not making money, venture capitalists still invest because they are collecting tons of data.
-
I honestly don't give the issue any serious thought. My reasons for making photos have little to do with processing techniques. Anyway there's already a fair amount of decision-making built into your Raw processor(s) or in-camera JPEG engine(s) of choice. I'll be surprised if software makers don't keep augmenting/enhancing that stuff.
-Dave-
-
I am not worried. (Also, I am not a professional photographer, but AI affects my profession too.) Roughly, AI processing seems to be "mimicking the most common processing choices, as indicated by observing how a large number of people handle a situation"; a glorified "average/typical choice".
- Why should anyone expect to be paid good money to reproduce "average" or "typical" processing choices?
- Why should one consider mere competent execution of routine actions as equivalent to talented, artistic decision making?
It sounds a bit like "pattern exposure metering": a sensible time saving way for anyone from beginner to professional to handle routine situations without complications like making multiple spot metering measurements, with the professional's craft involving knowing when and how to turn off the automation and make less routine, more skilled artistic decisions.
Well with the hugely vast majority of people thinking cellphone images are great and with AI in these cellphones advancing at amazing rates...will there be anyone left who really gives a damn if you created art the hard way?
-
...will there be anyone left who really gives a damn if you created art the hard way?
Should people care that something was created "the hard way", or just about whether the hard work resulted is something artistically better than — or just interestingly different from — what they can get themselves from a small camera with big processing power? Not that there is anything wrong with some people wanting "hand-crafted" and paying extra for it, but I do not see that artists or artisans are "entitled" to be paid for hard work unless it improves over what can now be attained more easily and at far lower cost.
But perhaps I am just one of those soleless people who prefers a $50 digital watch over a $5000 hand-crafted mechanical masterpiece that keeps time less accurately.
-
I am not worried. (Also, I am not a professional photographer, but AI affects my profession too.) Roughly, AI processing seems to be "mimicking the most common processing choices, as indicated by observing how a large number of people handle a situation"; a glorified "average/typical choice".
- Why should anyone expect to be paid good money to reproduce "average" or "typical" processing choices?
- Why should one consider mere competent execution of routine actions as equivalent to talented, artistic decision making?
It sounds a bit like "pattern exposure metering": a sensible time saving way for anyone from beginner to professional to handle routine situations without complications like making multiple spot metering measurements, with the professional's craft involving knowing when and how to turn off the automation and make less routine, more skilled artistic decisions.
Perhaps you should be more worried. Game theory is the hotbed of AI development. AI has gone far beyond "competent" for chess and go. This is a specific AI "intelligence", but a AI general intelligence is potentially not too far away (1-3 human generations). We should also be asking ourselves what creativity is. I am not convinced it is uniquely the ability of human beings.
Given that the development of AI is likely the most significant development in human history we should be spending time understanding the possibilities and implications.
-
AI in its current state is merely an efficient way of encoding patterns, which still requires a skilled designer upfront to guarantee any reasonable rate of success. Creativity requires a deeply felt desire to communicate, and AI, no matter how advanced, will never "feel" a need to communicate, and especially not via an art expression because of for example an inability to articulate this desire via other means.
-
AI in its current state is merely an efficient way of encoding patterns, which still requires a skilled designer upfront to guarantee any reasonable rate of success. Creativity requires a deeply felt desire to communicate, and AI, no matter how advanced, will never "feel" a need to communicate, and especially not via an art expression because of for example an inability to articulate this desire via other means.
I think you are underestimating human creativity in the AI realm. It’s not here now in AI, but I think it will be in the future. I get nervous when I hear people using “never” in discussions. I suspect human creativity is just pattern recognition and extension.
-
Interesting. Aren't we just a result of being filled with data over our lifespan?
-
I suspect human creativity is just pattern recognition and extension.
In that case, you have reason to be scared yes! ;-)
The question is what pattern do you recognise or extend when we communicate the concept "infinity"? The remarkable part there is not that we recognise or extend a pattern, but rather that we can negate a pattern and remain completely confident about our sanity.
-
Would a machine be able to work at the patent office and wonder - "what would I see if I rode on a light beam"
-
Would a machine be able to work at the patent office and wonder - "what would I see if I rode on a light beam"
I hear you and not to in any way denigrate Einstein's achievements, but if he had not made the breakthroughs someone else would have soon enough. Each insight is achieved on the shoulders of past work. I see no reason an AI cannot do this in the future. It is possible that future some breakthroughs will only be achieved by AI.
-
I don't expect AI to replicate human intelligence. It'll likely be its own thing, maybe "smarter" than us in some or many respects and less so in others. There are aspects of the world even a turkey perceives and intuits better than we humans do. I'd expect no less from AI.
-Dave-
-
I don't expect AI to replicate human intelligence. It'll likely be its own thing, maybe "smarter" than us in some or many respects and less so in others. There are aspects of the world even a turkey perceives and intuits better than we humans do. I'd expect no less from AI.
-Dave-
+1
-
Interesting stuff.
-
I don't expect AI to replicate human intelligence. It'll likely be its own thing, maybe "smarter" than us in some or many respects and less so in others. There are aspects of the world even a turkey perceives and intuits better than we humans do. I'd expect no less from AI.
-Dave-
You are so exactly right that we humans are not always the most intelligent beings and many animals have intelligence and senses well beyond our means.
-
My dog is way smarter than I am. He doesn't follow me around and pick up my krap.
-
There are aspects of photography which have become less difficult. We don't need to bring a cart, mix explosive chemicals with dangerous fumes, paint them on a glass plate, try to keep dust off of them while we load them in a camera to make an exposure before it dries but after we have carefully moved the lens back and forth to get it in focus, being careful to have the aperture shut down and to avoid light getting onto the plate before or after exposure.
The technology is constantly evolving to make our tasks "easier." Sometimes, too much so. "You press the button, we do the rest" was an old Kodak add for the Brownie.
Technology which means I can do less work and not have to get dust on my plates--yes, please! A box which takes away control and frequently makes poor choices--I'll pass. AI technology that competently removes noise that was introduced by the sensor--sounds great if it does its job correctly.
But there is still a lot of room for creativity and my guess is there will continue to be a space for creativity as the technology proves itself capable of certain tasks, our efforts will move to the spaces where they serve our vision better using the best of the available technology.
Step one is determining which of the automated features help make better pictures and which are better avoided. Step two is decide how you wish to express your vision. Step three is do the work.
Looking back at older articles and letters to the editor in older photo publications, you can see that this is a (relatively) old conversation in photography.
:)
-
Somewhere in the early eighties I was at a big name tech conference. One session was on the future. In one example, a speaker said within 10 years we will see movies with young Marylin Monroe acting beside young Harrison Ford.
Stand-in actors would make the scene. Computers would replace the stand-ins, drawing from every movie ever made by the two stars. Voices would equally be replicated. It would be an elaborate cartoon but no one would be able to tell the difference.
Where is that? I'm still looking for it. The closest I've seen was when Forest Gump met President Kennedy at the White House. No speaker at that conference mentioned self driving cars, although flying cars was a hot topic.
-
Somewhere in the early eighties I was at a big name tech conference. One session was on the future. In one example, a speaker said within 10 years we will see movies with young Marylin Monroe acting beside young Harrison Ford.
Stand-in actors would make the scene. Computers would replace the stand-ins, drawing from every movie ever made by the two stars. Voices would equally be replicated. It would be an elaborate cartoon but no one would be able to tell the difference.
Where is that? I'm still looking for it. The closest I've seen was when Forest Gump met President Kennedy at the White House. No speaker at that conference mentioned self driving cars, although flying cars was a hot topic.
Not to be too snarky but... ever hear of CGI? It's pretty commonplace these days.
-
There are aspects of photography which have become less difficult. We don't need to bring a cart, mix explosive chemicals with dangerous fumes, paint them on a glass plate, try to keep dust off of them while we load them in a camera to make an exposure before it dries but after we have carefully moved the lens back and forth to get it in focus, being careful to have the aperture shut down and to avoid light getting onto the plate before or after exposure.
...
Technology which means I can do less work and not have to get dust on my plates--yes, please! A box which takes away control and frequently makes poor choices--I'll pass.
I doubt that the options for doing it ourselves are going away, except in some cameras aimed at casual photography. The early days of auto-focus had some cameras going over-board, not offering manual focus, but camera makers then learnt their lesson. For example, even camera-phones whose basic camera app is AE/AF only offer apps giving more control. It is amazing the number of times I have read in posts in forums that seem to believe that low-level DSLRs are so automated that they lack the option for manual exposure controls or manual metering, but it just ain't so.
And I welcome every new option to allow automation of routine tasks, for the (great majority of) cases where I know that I would make a "standard" choice, so long as manual override is a quick button push or dial twist away.
-
Not to be too snarky but... ever hear of CGI? It's pretty commonplace these days.
Certainly, I'm always amazed at the fake architecture, explosions, floods, and animals. especially the animals. The tiger in Life of Pi was outstanding (and the CGI company that made it went broke within a year). But where are Marylin and Harrison, Bacon and Bacall, Bogey and Bullock?
-
Certainly, I'm always amazed at the fake architecture, explosions, floods, and animals. especially the animals. The tiger in Life of Pi was outstanding (and the CGI company that made it went broke within a year). But where are Marylin and Harrison, Bacon and Bacall, Bogey and Bullock?
A tiger doesn't claim property for the use of his image ;)
Ever see "The Congress" with Robin Wright?
-
Certainly, I'm always amazed at the fake architecture, explosions, floods, and animals. especially the animals. The tiger in Life of Pi was outstanding (and the CGI company that made it went broke within a year). But where are Marylin and Harrison, Bacon and Bacall, Bogey and Bullock?
How about Peter Cushing (and briefly a younger Cary Fisher) in "Star Wars Rogue One": https://www.yahoo.com/entertainment/rogue-one-creator-defends-cgi-tarkin-says-carrie-fisher-loved-her-digital-self-203453637.html
-
Just to note that CGI and AI are distinctly different subjects that may merge further in the future.
CGI is an already established technique (basically using modeled objects with surface overlays). AI is starting to become accessible to the general public in everyday applications (recognizing or replacing features/structures). Behind the scenes, AI and Neural networks typically depend on lots of (distributed) processing power. Recently AI has been optimized for use on lower processing powered hardware for a few dedicated use-cases, notably feature recognition (like faces/eyes). AI requires training of a system to recognize objects/features. Once trained, it becomes faster to apply actions (like focusing) to the recognized features. Not all AI is neural-network based.
Cheers,
Bart
-
It will be interesting but I doubt I'll see it, if it happens. We were watching a mid 90's police show. No computers and they used a big map and pins to follow murder pattens. Look how much has changed in 20+ years. All the face recognition tools, etc. My wife said if that had been a show from the 70's it would have been pretty much the same. I remember playing the most sophisticated game of it's time at the pub - Pong.
-
In answer to the OP question:
Yes.
https://ai.googleblog.com/
-
From everything I've heard, AI will be a massive revolution in the coming years.
We can say for sure it's going to get very advanced. What is really hard to predict though, is how it will change our lives. Seems to me that being a creative is one of the slightly safer lines of work to be in.
The thing to remember is: Anything that becomes ubiquitous becomes un-special. People get bored of consuming the same thing all the time. If A.I. makes all pictures look too perfect, people will quickly become interested in real pictures again.
As another poster pointed out, there is already a huge trend for AI modified selfies. This will not last long. In just a few years I would predict it will go completely out of fashion, or become a more niche obsession.
-
Automatic transmissions have been around for decades; I will still choose a standard/manual transmission even if that means switching manufacturers to get one. I know automatic transmissions aren’t AI, but I prefer to be in control of what’s happening where the tires meet the road. Auto might be fine for the nice days, but when there’s snow under the wheels, I’m glad to have the skills to drive a standard.
Needless to say, the same goes for photography. Push button processing is rampant even without AI. It may or may not make the end result better. In fact it stifles creativity, just like digital filters have done. And, if all you’ve ever done is push button processing, then you won’t have the skills to deal with the more difficult situations.
-
This is interesting and is well beyond accenting a sky by automatically segmenting it and editing its appearance:
https://youtu.be/kSLJriaOumA
kirk
-
Here's something to consider:
An AI (manmade) sunset shot from mai bedroom window with an (a)iPhone lightly processed in AI Clear! ai ai ai!
-
On my Flickr site I've posted 2 images that show the power of AI in recovering images:
https://www.flickr.com/photos/20926615@N05/46264040954/in/album-72157678279709128/
https://www.flickr.com/photos/20926615@N05/46074561705/in/album-72157678279709128/
Bt whatever you feel about the advance of AI the undeniable fact is that it works and is getting even better every day.
-
"AI", as has been demonstrated above, seems more misunderstood than understood. It may well already be beyond actual comprehension for many, as it does require walking away from flat earth a bit and opening the mind. there are many quite comfortable with the view just to the edge where the "ships fall off".
An expanding discussion can be found on the generalities in two recent works of Yuval Noah Harari, "Homo Deus" and "21 Lessons for the 21st Century".
These are unbelievably exciting times, which seem light years away from the day I heard my grandmother marveling at the fact that the plane could not only take off and fly, but actually carry humans within. I tried to reference this extraordinary leap in another area here in the forum, but realized that the fact that the ability, once taught to perform a task, quickly lead to the ability to build better tasks and better abilities, far more quickly and completely than we are able, is already reality in AI. All the data we toss around freely is already being put to use in ways beyond most humans ability to comprehend.
Photographic use is the mere tip of the iceberg.
-
I hear you and not to in any way denigrate Einstein's achievements, but if he had not made the breakthroughs someone else would have soon enough. Each insight is achieved on the shoulders of past work. I see no reason an AI cannot do this in the future. It is possible that future some breakthroughs will only be achieved by AI.
Lorentz and Schwarzchild in fact had already published equations calculating time dilation for both constant speed near lightspeed as well as time in a gravity well before Einstein. Einstein did however make the leap to conclude that space was a malleable mesh that could be stretched and squashed.
As for AI, I think that it already has usurped human creativity. Deep dream for example is a piece of software that is a wonderful surrealistic tool that no human could ever create. And all it does is modify an image so that it approaches the training data used to create a particular convolutional neural net. The data that it was trained on, in fact determines the output.
And when our preferences have been sufficiently tabulated, a neural net will be able to create images, memes and products that most efficiently stimulate and gratify us. And because the granularity of such a system can be focused to EACH individual's tastes, an AI will be able to generate 'art', food, clothes. novels, news, toys or anything that is most appealing for YOU and you ALONE. And therein lies the tragedy. The isolation we are all experiencing by the cell phone stealing our face will become more acute as the AI steals our mind. Human interaction will become mediated by a machine buffer. We will lose the ability to interact. And if you apply the definition of matter and energy provided by quantum mechanics, we will in fact cease to exist by virtue of the fact that we will have ceased to interact.
-
Just had a lecture from a professor who studies AI.
His conclusion was: AI will be a tool in the hands of people ; like a calculator.
It will not be able to make ART, other than being a tool helping to get to the outcome more quickly / more precise.