"Photography is dead" (AI), Peter McKinnon's latest video. Is it??

I do photography for myself. My photos represent photons that I captured with my camera at a certain instance in time.

One genre that I enjoy is nighttime photography. I use a star tracker to minimize star trailing. This setup is fiddly and takes time to set up and calibrate, but I enjoy doing this. I am aware that these days you can point a smartphone at the sky (handheld) and get a beautiful image of the Milky Way with intricate detail in nebulas and such. An image such as this is a collage, with the image of the sky derived from the training dataset. The image may even be a reasonable representation of the night sky, but the image doesn't represent photons captured by the smartphone. The smartphone image merely serves to guide the AI algorithm.

Yes, AI can create all sorts of digital art where mountains and lakes appear out of nowhere. These images do not represent reality. But right now I see a lot of use of AI to create reasonably accurate images; for example, noise reduction and upscaling. Suppose I have a low-res landscape image and I upscale it. It may have a pine tree in it, and the AI algorithm latches on to this and replaces my pixels with more detailed pixels from a pine tree in its training dataset. It goes through my image in minuscule detail, copying and pasting from the training dataset, replacing my pixels with someone else's.

Is this photography? Is this my image? Should I be giving credit to the millions of photographers who have contributed (perhaps unknowingly) to the training dataset?

I think that when photography transitioned from analog to digital, we introduced a certain amount of laziness. Take the picture and fix it in post. I think AI is going to create more laziness - just point your smartphone and an amazing surreal image will be generated. The question is: is this photography? Or are we using our smartphones to search a database to find someone else's work?

As for forensic photography - I think this is one area where AI should be used, but with caution. If AI can be used up upscale a grainy surveillance camera image with a high probability of being reasonably accurate, then why not do this?

I'll get off my soapbox now. Thank you for listening.

Do you still recall the time when so-called creative filters were in fashion? I think we all gave them a try ot two.
Star filters, sunset filters, degraded ones etc...
Photographers - and painters- have always played with distorted (or should I say improved?) reality. If people want to add lakes or dinosaurs, why not! It's their photography, not mine. I don't like AI's excesses? Who cares...
As long as pictures remain pictures and not means of coaxing people into believing untruth.

I was scanning ( digtizing) ALL my old Photograph and Color Slides dating back 60 years. what I have found is Phtoshop ( AI??) is extremely valueable. It make quite a few not so good images to be very good. So this is a case and point that AI's is useful when it is needed. As for phtograpphy is dead, I may not agree. AI will help create better picture. Yes it is not "pure" phtography. But it is the end result that counts. Ansel Adam took a picture and spends hours in the darkroom (including choose the right phot paper) to come up with stunning picture. For me that is kind of AI (in his head). Now we can use photoshop and much much easier and faster.
If you want "pure phtography", you need to take color slides. no cropping, no adjustment. What you see is what you got.

Here's another simple trick with Photoshop AI - expanding the borders of an image. I have just posted in the BIF thread a Cormorant flying fast and close in front of me, and it was difficult to keep the bird in frame. The one I posted was successful, and I cropped it to my satisfaction https://www.canonrumors.com/forum/t...ur-bif-photos-here.19270/page-350#post-968268
Another one had the bird too close to the bottom left, but using generative fill, I could easily add extra background and then crop (can't complain about the sharpness of the R100-500 in the extreme corner of the R5!).

Before:


Expanded the left and bottom borders and then crop (both reduced in size):

AlanF said:

Here's another simple trick with Photoshop AI - expanding the borders of an image.

Click to expand...

Depending on the background, that was relatively simple in PS even before generative fill. When I got a new monitor several years ago, I had a few images I wanted to use as wallpapers (I typically have 5 desktops/spaces on my Macs). The display is 5K:2K, with an aspect ratio of 21:9. Here's an example:



The original is uncropped, and was shot with the 1D X and 70-200/2.8 II. The red tail decided to have lunch on our fence post.

neuroanatomist said:

Depending on the background, that was relatively simple in PS even before generative fill. When I got a new monitor several years ago, I had a few images I wanted to use as wallpapers (I typically have 5 desktops/spaces on my Macs). The display is 5K:2K, with an aspect ratio of 21:9. Here's an example:

View attachment 210688

The original is uncropped, and was shot with the 1D X and 70-200/2.8 II. The red tail decided to have lunch on our fence post.

Click to expand...

How do we know the Red Tail was really there and you did not use AI? I just told PS to conjure me up a fence plus post. The post was too high so I used generative fill to cut it down, then used AI to put a hawk on it.

AJ said:

As for forensic photography - I think this is one area where AI should be used, but with caution. If AI can be used up upscale a grainy surveillance camera image with a high probability of being reasonably accurate, then why not do this?

Click to expand...

Then you wouldn't want to be a person whose image was used for training of both upscaling and face recognition AIs.

AlanF said:

How do we know the Red Tail was really there and you did not use AI? I just told PS to conjure me up a fence plus post. The post was too high so I used generative fill to cut it down, then used AI to put a hawk on it.

View attachment 210727

Click to expand...

I'm sure one could do a much better job here with "classic" (non-AI) photoshopping.

Kit. said:

I'm sure one could do a much better job here with "classic" (non-AI) photoshopping.

Click to expand...

Do you need a sense of humour implant, or do I need to put a smiley to indicate I am having a joke?

AlanF said:

Dead easy. Used Photoshop select tool from the menu; it detected the bird and I added the branch it was on close to it, did inverse selection and then Field Blur. It has indeed changed my attitude to ultra-expensive lenses. One major reason reason for buying a wide lens is subject isolation. However, simple subject isolation is easy using Photoshop as above and is already standard automatically on smart phones for portraits. With dualpixel/quadpixel AF, it may be possible to estimate distances over the image sufficiently for software than to do sophisticated out-of-focus effects to mimick different apertures. I think using digital methods in this way to change aperture rather than using the current analog methods with aperture on the lens is no more different than is using digital vs film.

Click to expand...

You should try using the 'Depth Blur' option in the Neural filters. You can play with that or just export a depth field mask and apply it to a gaussian blur for another option to focus blur hacks. I find it more realistic.

Has everyone forgotten the word 'compositing'. That's all certain AI images are. Images culled together to produce a work of art. People have been doing this for years. Compositing is not photography. There are certain tools and uses for AI that I find quite nice and help my workflow go smoother and faster. For example the select subject feature in photoshop (it's much more accurate), the new remove tool that does a much better job at removing spots and blemishes that the old spot healing brush simply messed up sometimes.

Photography is what YOU make it, not what others think. If you are so concerned about what others think that it has this large of an impact on your creativity then you are limiting yourself and your creative freedom. Don't be afraid to experiment with your own photography. It's about the journey, not the end result. The end result is a creation of all of your hard work, time and energy. The joy should be in that process. The end result is like getting your piece of cake (your reward) after eating dinner.

Also, what are you photographing that you are so concerned about AI? AI can't be at a sporting event to take a photo. I've taken up sports photography for local Jiu Jitsu events. AI isn't going to be there to take the photos and capture the moment. The people competing aren't going to want fake AI generated versions of their matches. One thing I've come to find is that people still like to purchase images of themselves at their respective sporting events. So while you no longer work for a sport magazine or online publishing platform, it's time to think outside the box. Honestly I wouldn't worry about what Peter McKinnon thinks. The guy is just a Youtube influencer. You know a salesman, a pitchman. He has an agenda to sell clicks and likes so he can continue his lifestyle. That doesn't make what he has to say, truth or totality.

Don't give up and good l-u-c-k (learning under correct knowledge).

macrunning said:

You should try using the 'Depth Blur' option in the Neural filters. You can play with that or just export a depth field mask and apply it to a gaussian blur for another option to focus blur hacks. I find it more realistic.

Click to expand...

Thanks for the tips - I have had a try with them. I am an amateur with PS, and don't find it intuitive, unlike most Mac programs so I appreciate advice.

Photography is to capture emotions with light. Any technology like KI/AI will fail here. Machines do not feel, nor are they creative. They can change what is already there and show a different perspective that did not exist before. Only the photographer can take pictures - just the way he likes it. I would consider AI/AI for noise reduction or editing quite useful. Those who are said to be dead often live the longest...

The sales figures, sales prices and profits of Canon and Nikon, who bet on more modern systems very late, leave little doubt that photography will continue in the future. Many young people who took their first photos with their smartphones are striving for something better and I see a future for photography there.

Juppeck said:

Photography is to capture emotions with light. Any technology like KI/AI will fail here. Machines do not feel, nor are they creative. They can change what is already there and show a different perspective that did not exist before. Only the photographer can take pictures - just the way he likes it. I would consider AI/AI for noise reduction or editing quite useful. Those who are said to be dead often live the longest...

The sales figures, sales prices and profits of Canon and Nikon, who bet on more modern systems very late, leave little doubt that photography will continue in the future. Many young people who took their first photos with their smartphones are striving for something better and I see a future for photography there.

Click to expand...

Was this text written by ChatGPT?