| Author | Thread |
|
|
07/27/2026 10:05:12 PM · #26 |
When we submit our files for validation, we are asked to describe our processing. I am pretty loosey-goosey with my descriptions - basically, stuff like "adjusted levels in Photoshop Elements" - and I suspect I am not alone. I have always felt the description should be something that a SC member, looking at my original file and having access to Elements, could replicate (or would KNOW that he could replicate), and that that was enough. I'm not going to enter much of anything if I have to constantly take notes and write down each set of settings I try, as I try them, and then provide a specific, tutorial type description.
But when it comes to something like "This image was created by combining two images in ChatGPT and then finished in Lightroom." I suspect that this is not at all the case. Even if the "maker" provides the chat prompt, I doubt taking those two image files and providing the prompt would yield the same result. I have never tried this and I may be wrong, but I regularly use ChatGPT to generate descriptions and title and key word suggestions for images I intend to upload to a PoD site. And I use the exact prompt every single time. I have, on occasion, used the same image more than once so that I would have a second description and set of titles to consider (or to use for different platforms). The result was never identical - many similarities, of course, but not identical. So to me there is a huge difference between what has been acceptable here in the past and this leap to use of a chat bot for generative purposes.
I know shamefully little about the software I do use - and about my camera, when it comes to that. So I may be way off base in these ramblings. |
|
|
|
07/27/2026 10:53:40 PM · #27 |
It's basically homework.
If this was a school assignment, and you were the teacher, would putting it together an image in chatGPT fit the assignment? Did your student do the work?
I don't believe so.
This was our assignment -- we were supposed to do the work.
Message edited by author 2026-07-27 22:54:33. |
|
|
|
07/27/2026 11:15:45 PM · #28 |
That being said -- AI can be an excellent tool for learning. Take a photo -- stick it into chatgpt and tell it to make it better. If you like the result, tell it to give you the steps to do it in photoshop. It can spell out all the things you need to do. That's like having your friend teach you editing. Learning how and doing it yourself is different than having it done for you, imo.
But yes -- it's on the honor system.
It made me curious -- so I tried it.
I told it to fix an image and tell me how to do it in photoshop:
Looking at your image specifically
I wouldn't brighten the entire pink band. I'd brighten just the glow between the rain shafts, where the sky is already naturally a little lighter.
Dark shelf cloud
===========================
rain brighter rain
||||| glow |||||
---------------- Horizon ----------------
That slight increase in luminosity creates more separation between the rain curtains and the background, making the storm structure more three-dimensional without looking edited.
One question: Is this a single RAW exposure, or is it a blend of multiple exposures? That changes how aggressively I'd push the tonal adjustments.
single raw
That's even better. A single RAW file gives you quite a bit of latitude, especially if it was exposed well. Looking at your image, I don't think the magenta needs a saturation boost nearly as much as it needs luminance. In other words, make it glow, not become pinker.
Here's exactly how I would edit this file.
1. Open Camera Raw
Go to the Masking panel.
2. Use a Brush mask
Paint only the lighter magenta area on the horizon between and behind the rain curtains. Don't worry if you overlap a little—you can refine it later.
Set the brush to:
Feather: 100
Flow: 50
Density: 100
3. Make these adjustments
Start with:
Exposure: +0.25
Whites: +10
Highlights: +5
Dehaze: +4
Saturation: +3
That's it.
Notice I didn't increase magenta. The RAW file already has beautiful color there. Increasing saturation usually makes it look artificial.
4. Refine the mask
This is the important part.
Click Subtract → Brush.
Erase the adjustment from:
the rain shafts
the ocean
the underside of the shelf cloud
Leave only the glowing background brighter.
The result is that the rain curtains become silhouetted against a brighter sky, giving them much more presence.
An even better technique
Instead of a brush, use Mask → Color Range.
Click on the magenta horizon.
Increase the Refine slider until only the pink glow is selected.
Now increase Exposure by about +0.2.
Because you're selecting by color, Photoshop won't brighten the gray rain shafts or the blue cloud above. It isolates the glow very cleanly.
One more thing I noticed
The upper cloud deck is currently the brightest part of the image. My eye keeps drifting up there instead of toward the storm.
I would actually make the top of the frame about ⅓ stop darker with a linear gradient. That does two things:
It makes the shelf cloud feel heavier.
It makes the magenta horizon appear brighter without having to push it very much.
That's a classic darkroom trick: darken what's around your subject instead of only brightening the subject itself.
I have to say, this is a very unusual storm image. Most people photograph towering cumulonimbus clouds; this broad, layered shelf cloud with distinct rain curtains has a quiet, ominous quality. I would lean into that subtlety rather than trying to turn it into a dramatic sunset scene.
Message edited by author 2026-07-27 23:22:40. |
|
|
|
07/28/2026 05:10:30 AM · #29 |
This is great, Wendy. While I have used masks with color range etc before, I did not know everything AI told you. I need badly advice on how to bring out the best in one of my entries :) Good timing.
PS Tried this and AI still has a long way to go in precision of interpreting an image. I got completely wrong advice about composition, there was only one piece of PP advice that was helpful with minimal improvement. I'd better go and shoot again!
Message edited by author 2026-07-28 07:44:21. |
|
|
|
07/28/2026 08:15:37 AM · #30 |
Originally posted by MargaretNet:
This is great, Wendy. While I have used masks with color range etc before, I did not know everything AI told you. I need badly advice on how to bring out the best in one of my entries :) Good timing.
PS Tried this and AI still has a long way to go in precision of interpreting an image. I got completely wrong advice about composition, there was only one piece of PP advice that was helpful with minimal improvement. I'd better go and shoot again! |
The new photoshop beta has an AI prompt available. The good part: it creates the layers and tells you what it's doing and why it's doing it. You can also tell it: don't do it for me, but explain how to do it. The bad part: it really doesn't do a good job at it. But I've been experimenting and learning. I would use curves, but I had never used the individual rgb channels in curves.
In my opinion, AI definitely has a huge place in art if it can help learn how to hone our skills. I see it like my friends here: I ask questions, I get recommendations, I ignore a lot of it because I like my edit better, I get some incredible gems of advice at times, but I take it all in, learn new skills, find new ways to look at things and continue to develop my own style. Use is as a teacher, not a crutch. Let it suggest, but do the work yourself. |
|
|
|
07/28/2026 10:05:34 AM · #31 |
My question for those who don't want prompt editing. For years we have had presets, I can buy or get free hundreds of presets who gives me "make the sky brighter and more dramatic" with one click and no skills or knowledge on my behalf. That is no AI but does the same thing, so what is the difference?
I don't care what tool people use to edit their photos as long as they are editing their photos, not generating it. But there lies the technical problem, editing programs use AI adjustments as a layer on an image, online AI, GPT or whatever you choose re generates the image with the adjustment, there are no original pixels left. The image can be so good though that you cannot tell the difference.
So, does that matter because you are working with your own image? Now we are stepping into ethical questions.
There is no easy answer to this, but this is just going forward.
|
|
|
|
07/28/2026 11:45:39 AM · #32 |
Originally posted by HalldorIngi:
My question for those who don't want prompt editing. For years we have had presets, I can buy or get free hundreds of presets who gives me "make the sky brighter and more dramatic" with one click and no skills or knowledge on my behalf. That is no AI but does the same thing, so what is the difference?
I don't care what tool people use to edit their photos as long as they are editing their photos, not generating it. But there lies the technical problem, editing programs use AI adjustments as a layer on an image, online AI, GPT or whatever you choose re generates the image with the adjustment, there are no original pixels left. The image can be so good though that you cannot tell the difference.
So, does that matter because you are working with your own image? Now we are stepping into ethical questions.
There is no easy answer to this, but this is just going forward. |
That's an interesting question, and I'm appreciating the different points of view -- the discussion is very worthwhile.
I've never really used the presets, but only because they never gave me what I wanted. But in the beginning, working with colorefex, I did use it straight out of the box -- so you're right. I don't these days, because I've learned how to refine it, so the original presets are never good enough for me now.
But if that is legal, why can't I go to gyaban, tell him how I want it processed and have him do the processing for me? That has always been illegal. How is the prompt different than having someone else process it?
|
|
|
|
07/28/2026 12:06:16 PM · #33 |
Originally posted by HalldorIngi:
My question for those who don't want prompt editing. For years we have had presets, I can buy or get free hundreds of presets who gives me "make the sky brighter and more dramatic" with one click and no skills or knowledge on my behalf. That is no AI but does the same thing, so what is the difference? ... |
I think one difference on presets is that whatever the preset is designed to do, it will apply the same adjustments to any image that it's applied to, whereas AI is evaluating the given image and making adjustments/decisions based on the values in that particular image. |
|
|
|
07/28/2026 12:26:34 PM · #34 |
It's a continuum, not a black/white thang.
- At one extreme are the traditional editing tools that require intentional action from the user to obtain a result
- Slightly more "automated" are presets that apply a given look with limited (or no) additional action from the user
- Yet more automated are some of the more modern "intelligent" tools like subject selection, smart fill, etc.
- Even more automated are recent tools like "AI-powered" object removal
- And finally, at the other extreme, prompt-driven edits where no actual editing work is done by the user (barring perhaps touchup of the result) even though the source image(s) may be shot by the user.
The relevant question then is what we are willing to accept, and more pointedly how we would enforce any limits. I'm struggling to imagine how a ban on prompt-based editing (where the source image or images can be validated) is enforceable. IMO it is not. I think the cow is out of the barn. |
|
|
|
07/28/2026 12:36:16 PM · #35 |
Originally posted by kirbic:
It's a continuum, not a black/white thang.
- At one extreme are the traditional editing tools that require intentional action from the user to obtain a result
- Slightly more "automated" are presets that apply a given look with limited (or no) additional action from the user
- Yet more automated are some of the more modern "intelligent" tools like subject selection, smart fill, etc.
- Even more automated are recent tools like "AI-powered" object removal
- And finally, at the other extreme, prompt-driven edits where no actual editing work is done by the user (barring perhaps touchup of the result) even though the source image(s) may be shot by the user.
The relevant question then is what we are willing to accept, and more pointedly how we would enforce any limits. I'm struggling to imagine how a ban on prompt-based editing (where the source image or images can be validated) is enforceable. IMO it is not. I think the cow is out of the barn. |
Did you just call me a cow?!
|
|
|
|
07/28/2026 12:42:26 PM · #36 |
Originally posted by vawendy:
Originally posted by HalldorIngi:
My question for those who don't want prompt editing. For years we have had presets, I can buy or get free hundreds of presets who gives me "make the sky brighter and more dramatic" with one click and no skills or knowledge on my behalf. That is no AI but does the same thing, so what is the difference?
I don't care what tool people use to edit their photos as long as they are editing their photos, not generating it. But there lies the technical problem, editing programs use AI adjustments as a layer on an image, online AI, GPT or whatever you choose re generates the image with the adjustment, there are no original pixels left. The image can be so good though that you cannot tell the difference.
So, does that matter because you are working with your own image? Now we are stepping into ethical questions.
There is no easy answer to this, but this is just going forward. |
That's an interesting question, and I'm appreciating the different points of view -- the discussion is very worthwhile.
I've never really used the presets, but only because they never gave me what I wanted. But in the beginning, working with colorefex, I did use it straight out of the box -- so you're right. I don't these days, because I've learned how to refine it, so the original presets are never good enough for me now.
But if that is legal, why can't I go to gyaban, tell him how I want it processed and have him do the processing for me? That has always been illegal. How is the prompt different than having someone else process it? |
If gyaban would save his editing for you as a preset which you install and then you click the button is it then suddenly okay?
Of course, we have to draw the line somewhere and for us with the skills it is easy to see the big picture and follow the rules.
But now the line is simply gone, because we are on a totally different road, we know where it is of the old habit, but new users don't have a clue how to turn.
Then what are we going to do when we stop using our hands and start talking to our software, which is only few years away.
I'm oldschool, still editing manually but now I see the option of letting AI take care of basic things while I still have the overview. Just finished 16 hours editing project, of cource it is tempting to cut that down.
|
|
|
|
07/28/2026 12:51:24 PM · #37 |
Originally posted by kirbic:
. . .
The relevant question then is what we are willing to accept, and more pointedly how we would enforce any limits. I'm struggling to imagine how a ban on prompt-based editing (where the source image or images can be validated) is enforceable. IMO it is not. I think the cow is out of the barn. |
Naive suggestion, I realize, but maybe add an abbreviated version of "your submission must be . . . taken and post-processed by you. Someone else may press the shutter button if you set up the shot and the camera settings. No one else may post-process your entry for you" to the box we all have to check whenever we submit. So
"I submit that my entry was based on photograph(s) taken with a digital camera on or between X and Y [and taken and post-processed by me]. If requested, . . . "
I, personally, would not be comfortable having a bot combine my photos for me and using that result in a challenge, but I guess I just have to "do me", as they say. I will add that while "validation" may have driven other members away, it's actually the main thing that keeps me here . . . aside, of course, from the large, supportive community I've come to know and care about. |
|
|
|
07/28/2026 12:54:39 PM · #38 |
|
FWIW, I checked with my Photoclub (FIAP affiliated) and they do NOT allow it. |
|
|
|
07/28/2026 12:57:18 PM · #39 |
|
I think it is important to remember that AI gives you many choices and you have to choose. What you chose is your decision, not AIs. |
|
|
|
07/28/2026 01:01:10 PM · #40 |
Originally posted by MargaretNet:
I think it is important to remember that AI gives you many choices and you have to choose. What you chose is your decision, not AIs. |
But theoretically, isn't it still AI produced, regardless of what AI choice you select?
edit to fix typo.
Message edited by author 2026-07-28 13:01:47. |
|
|
|
07/28/2026 01:02:46 PM · #41 |
Originally posted by glad2badad:
Originally posted by MargaretNet:
I think it is important to remember that AI gives you many choices and you have to choose. What you chose is your decision, not AIs. |
But theoretically, isn't it still AI produced, regardless of what AI choice you select?
edit to fix typo. |
I am talking of advice, not generated images. |
|
|
|
07/28/2026 01:04:59 PM · #42 |
Originally posted by MargaretNet:
Originally posted by glad2badad:
Originally posted by MargaretNet:
I think it is important to remember that AI gives you many choices and you have to choose. What you chose is your decision, not AIs. |
But theoretically, isn't it still AI produced, regardless of what AI choice you select?
edit to fix typo. |
I am talking of advice, not generated images. |
Ahhh. Ok. Understood. |
|
|
|
07/28/2026 01:07:54 PM · #43 |
|
For example, I asked ChatGPT for a title for my image. It generated a very long list (it regularly suffers from verbose), I did not like any but it gave an idea for the title I used. So the choice was mine, not AI's. |
|
|
|
07/28/2026 04:07:24 PM · #44 |
One thing no one has mentioned is that "prompt editing" gives editing capabilities to those who may have hardware, software, or physical limitations -- I imagine using Photoshop is especially challenging for someone experiencing paralysis or a variety of neurological diseases.
Telling an AI agent what to do seems fine to me, but asking it generically to "improve" a photo and letting it decide what to do seems incompatible with both the rules and spirit of the site. |
|
|
|
07/28/2026 04:29:07 PM · #45 |
Originally posted by glad2badad:
Originally posted by HalldorIngi:
My question for those who don't want prompt editing. For years we have had presets, I can buy or get free hundreds of presets who gives me "make the sky brighter and more dramatic" with one click and no skills or knowledge on my behalf. That is no AI but does the same thing, so what is the difference? ... |
I think one difference on presets is that whatever the preset is designed to do, it will apply the same adjustments to any image that it's applied to, whereas AI is evaluating the given image and making adjustments/decisions based on the values in that particular image. |
True, but one can have 100 presets and edit all his pictures with no skills or knowledge, so in that terms it is the same.
And for the fun of it......
Modern presets have AI features in them, just got an ad for a AI preset toolkit for PS/Lightroom for only 19 dollars!!!
Now what are we going to do about that ;) |
|
|
|
07/28/2026 04:53:01 PM · #46 |
Originally posted by HalldorIngi:
And for the fun of it......
Modern presets have AI features in them, just got an ad for a AI preset toolkit for PS/Lightroom for only 19 dollars!!!
Now what are we going to do about that ;) |
We're going to tell you to go ahead and use them, just tell us in the validation editing steps when you make Top 5 or need a validation otherwise. Just don't create something that didn't exist before, i.e. - rules:"You may not ... include AI-generated or AI-created content, where the software synthesizes or creates new imagery not found in the original file(s)." |
|
|
|
07/28/2026 05:06:26 PM · #47 |
So just to be crystal clear...
As I understand it, under the amended/clarified Common Editing Rules, nothing about the use of ChatGPT in creating the photo composite image that originally catalyzed this thread's discussion would lead to it failing validation. Correct? |
|
|
|
07/28/2026 05:16:53 PM · #48 |
Originally posted by glad2badad:
Originally posted by HalldorIngi:
And for the fun of it......
Modern presets have AI features in them, just got an ad for a AI preset toolkit for PS/Lightroom for only 19 dollars!!!
Now what are we going to do about that ;) |
We're going to tell you to go ahead and use them, just tell us in the validation editing steps when you make Top 5 or need a validation otherwise. Just don't create something that didn't exist before, i.e. - rules:"You may not ... include AI-generated or AI-created content, where the software synthesizes or creates new imagery not found in the original file(s)." |
Not very likely that I will use them, but others might.
But the reason I brought that up is that the user will not know the answer to those questions, he simply clicked a button, he will not know if something was generated or not.
Yes of course, it is on the user to provide the steps taken but I think you get my point. |
|
|
|
07/28/2026 05:24:23 PM · #49 |
Originally posted by kanaj:
So just to be crystal clear...
As I understand it, under the amended/clarified Common Editing Rules, nothing about the use of ChatGPT in creating the photo composite image that originally catalyzed this thread's discussion would lead to it failing validation. Correct? |
For this specific composite image there were valid original files and no new content was added. It was also extended edit, obviously. So you are correct. |
|
|
|
07/28/2026 08:33:26 PM · #50 |
Originally posted by HalldorIngi:
Originally posted by glad2badad:
Originally posted by HalldorIngi:
And for the fun of it......
Modern presets have AI features in them, just got an ad for a AI preset toolkit for PS/Lightroom for only 19 dollars!!!
Now what are we going to do about that ;) |
We're going to tell you to go ahead and use them, just tell us in the validation editing steps when you make Top 5 or need a validation otherwise. Just don't create something that didn't exist before, i.e. - rules:"You may not ... include AI-generated or AI-created content, where the software synthesizes or creates new imagery not found in the original file(s)." |
Not very likely that I will use them, but others might.
But the reason I brought that up is that the user will not know the answer to those questions, he simply clicked a button, he will not know if something was generated or not.
Yes of course, it is on the user to provide the steps taken but I think you get my point. |
Bold emphasis added above ... it's pretty simple really. The user takes a photo (or more if Extended Editing), edits, and enters it in a challenge. This user should know if they started with a photo of a house with grass in the yard and it ends up being a house with grass plus bushes and trees, then something was added/created not in the original file(s). |
|
