ChatGPT as a darkroom tool

[PrairiePhotographer]

Has anybody else found value in using ChatGPT as a darkrrom tool? I am a novice photographer that picked up film photography during the early days of the COVID pandemic. I've never had formal photography training and have pieced together whatever knowledge I've gained largely through trial and error supported by reading the Ansel Adams trio of books, this forum, and Large Format Photography's forum.

Over the last month I've used ChatGPT as a darkroom aid in two ways: 1) to help dial-in my development process for 4x5 B/W film, and 2) to print a B/W photo album I am giving as a Christmas gift. In each case I've uploaded images and other information (especially densitometer measurements) to ChatGPT. It interprets the inputs I've provided and suggests next steps . For example, when printing from a new negative an iteracton might go something like this: 1) I upload a test strip with 3 second increments made with a 2.5 filter. 2) ChatGPT recommends trying an intial print with a base exposure of 8 seconds using the 2.5 filter, 3 second burn on a particular object that needs to be lightened, and a 5 second burn using a 5 filter on a different object that needs to be darkened, 3) I execute that sequence and upload the results, 4) ChatGPT suggests tweaks. Etc.

I've found it to be hugely helpful. While it is not infallible, I have not uncovered any "hallucinations" or other clearly crazy ideas. I think of it as being a knowledgeable partner, one that might forget things that I'm surprised it will forget (it definitely "thinks" differently than other computer systems I've used over the years), but one that is still quite knowledgble and can explain its reasoning. I'm glad to follow its advice in the relatively low risk environment of the darkroom, where errors are measured in the range of a failed print or blown practice negative (I haven't trusted it with precious ones yet).

BTW, I have found ChatGPT to be an incredible aid in a wide range of DIY realms (plumbing, auto repair). I am optinistic that I'm on a path to solve a problem that my plumber couldn't solve -- and I'm a worse plumber than I am photographer, but that is a different subject!

 

[bags27]

Interesting.

I use both ChatGPT and Gemini professionally for research (never for writing), and, because I'm an experienced researcher and know how to drill down, I find many layers of hallucinations. I persist with both programs, because it's possible to "train" them and reduce--not yet (maybe never) eliminate--errors. But maybe most of those errors occur because I'm asking it really complex questions.

I imagine if you task it as you are, it can be very useful. I think I'll try it. Thanks!

 

[Horia]

Hello,

At the beginning I was impressed by this chatbot, for a few weeks, after that I've noticed how it works and what was it purpose, I guess we all know now what it's all about with the so called AI technology, is a business meant to make lots of money from different sources, first the users then the company, the hardware manufacturers and the others companies that sell this kind of services or rent them. I have friends in the industry and they are nothing more then "gold diggers" and the gpt is the shovel. The app "per se'' is a data mining platform, like facebook, discord, X, google and so on. There is nothing innovative to it the technology was used in the '50'for and put aside for the right time, it's just a Whatsapp predictor on steroids, nothing more. In reality this kinda technology should be considered illegal if the laws were to be applyed as they are right know in most countries, it has been trained with data from illegal sources, without permission and has been trained by almost enslaved people that hat to teach the bot to discriminate and prompts answers for the user. There is nothing intelligent about this apps, just pre-configurated answers. They killed the internet that was already killed by Google and YouTube, also I've read that by the end of 2026 almost 90% of the internet will be generated artificially by bots and people will lose interest in creating man made content for the web. As a web designer I am an activist, and fight the WordPress monopoly on the CMS scene, people don't even know that the internet died because of the lack of interest of the users in learning how to code and design them self, and use the so called ''free-CMS" that early this year told the world the truth that they are collecting data and user behavior stats to train the AI... I guess will have to deal with it and go along, the companies are far to powerful to fight against, and they all have the governments approval and protection to do as they will.


I don't know if you agree but it does not matter anymore what people think or want to do... see you around.

 

[loccdor]

I have not uncovered any "hallucinations" or other clearly crazy ideas.

I've found if you give it too much information or ask it about topics it doesn't have good data on it completely loses the plot.

For example, I tried some retirement planning calculations on it. It was doing okay for a few replies, then it completely forgot about an important piece of information, making its result inaccurate. Probably because it had to juggle about 10 pieces of info simultaneously.

These are mathematical calculations similar to darkroom ones, which are one of the main things it's good at, but even then, you have to watch that it doesn't confidently make a mistake.

In my opinion it's good as a starting point to build your understanding, but after you get the concept, it's often better to solve the problems directly. That way your mind will also get exercise and be able to repeat the process without aid.

It's helpful with common machine problems, for example it was able to figure out what was wrong with my riding lawnmower.

I have experienced its hallucinations, for example it told me that a very grainy film developer was less grainy than a relatively fine grained one. A lot of the data it's training on regarding darkroom stuff is coming from internet sources of enthusiastic amateurs who may be completely wrong about what they're saying. Since it trains on that, it will repeat their errors.

 

[koraks]

There is nothing innovative to it the technology was used in the '50'for and put aside for the right time, it's just a Whatsapp predictor on steroids, nothing more.

This is factually incorrect as is most of your post. It's also irrelevant to the thread at hand so let's end this diversion right here.

Let's get back to the topic how AI can help (or hinder) darkroom work. I presently don't use it in that particular capacity, but follow with interest how others leverage it.

 

[Carnie Bob]

A good friend of mine, has been travelling extensively to many darkroom and workshops possibly to some of you here on Photrio, He is extremely interested in all aspects of
photography, and he video tapes records workshops as well on all different topics creates records from books that he then runs through something that is AI based to analyze the topic. For example right now he is scanning my books on Gum Bichromate and Palladium processes and we are looking at Calvins book , both Christina Andersons book , Stephen Livik and David Scopic books as well as pdfs I have done on the subject and the notes on the topic my students who have studied with me have made. I have also given him William A Jolly notes and my notes on solarization as well he will be videotaping me doing a round of negative to negative solarizations hoping to produce high quality positives that I can scan and then print. I find this all interesting and the first results
I see are quite interesting , it seems my friend has to ask the right questions as well prod the AI to look deeper into aspects of its response.
Hopefully he will garner a very complex and complete analysis of two of the processes I work with.

 

[Carnie Bob]

This is factually incorrect as is most of your post. It's also irrelevant to the thread at hand so let's end this diversion right here.

Let's get back to the topic how AI can help (or hinder) darkroom work. I presently don't use it in that particular capacity, but follow with interest how others leverage it.

Can you imagine the knowledge and lack of knowledge that is archived on this site. It would take a major effort to glean out. the good stuff but like having a moderator team, Sean could
put together a team of inhouse consultants of people that contribute here and look at topics discussed and create a world class resourse for future young photographers. Or does this sound too dystopian in theory or would it be illegal to take words of others ?

 

[koraks]

Sean could
put together a team of inhouse consultants of people that contribute here and look at topics discussed and create a world class resourse for future young photographers.

That's essentially what Photrio is. But we accept the 'noise' of people interjecting wrong and offtopic comments to an extent because it's a forum and in principle you don't need a license to play. Sometimes as a moderator I might nudge (gently or a little less so) into the right direction of people veer off course too much and this distracts from the question someone has asked. That's still quite some distance from some kind of consultancy solution, in which case the question of the business model would arise pretty early on. I don't think that really computes; you could assemble a team of experts but the 'customers' in question are characterized by a very limited (or entirely absent) willingness to pay. So I think AI with its currently low transaction cost fills that gap fairly nicely, insofar as it's not filled by forums like Photrio.

 

[Jan de Jong]

I use CoPilot, mainly because it is state-full, so the more you communicate about the particular subject the more it becomes a co-worker. It is very helpful in getting for the Alt processes some information or give me possible explanations on why some things occur. That is not direct in the darkroom but it is about explaining the results and look for improvements before and after going in the darkroom. I did use it for example to calculate the sensitivity approximately of some cyanotype emulsions.
Another use but more fun is that it is becoming very good in visual interpretations. So one use case was to let it re-render a double exposure in the single exposures.
I usually upload prints i have made in CoPilot for its evaluation, to allow it to see my direction and goals, to assist in further new adventures with the alt-processes.
So yes I use it and I find it increasingly useful. It does already often source from Photrio but also can go back to old books at the same time. Clue is what and how you ask, and to use a state-full AI.

that is my thought for the moment.
Jan.

 

[loccdor]

I think the domain of well-compiled carefully scrutinized information is instructional books. These can be paper or electronic, made freely or charged for, but someone has to do the work of creating them in order for them to exist. The filtration of good from bad information is done through experience and experiment.

Shortcuts with AI often must be so closely watched for error that the time-saving bonus is minimal. Many workplaces that have incorporated AI spend most of their time error checking to the point that some realize its better go back to the old way.

I like AI as a way to give ideas on how to get an unfamiliar task started, or when you are stuck on something. You have to be able to understand its outputs are only as good as its inputs when judging its response.

 

[koraks]

Shortcuts with AI often must be so closely watched for error that the time-saving bonus is minimal. Many workplaces that have incorporated AI spend most of their time error checking to the point that some realize its better go back to the old way.

Sorry, I don't recognize this at all. I see people using AI around me all the time and boosting their productivity massively with it. This ranges from healthcare situations where diagnoses are performed quicker and especially more reliably than by human doctors, generation of reports and reference letters in psychiatry, customer support in various fields, technical troubleshooting, software engineering is HUGELY affected and accelerated by AI (which has been happening for a long time already but has now been boosted again), entrepreneurial activity using it for idea generation and initial market analysis, etc. etc. etc. The list is literally endless. As to the supposed reverting to non-AI ways of working - sorry, nope nope nope this is not the general direction at all. There may be isolated cases where they take a step back temporarily only to take two massive strides ahead tomorrow. That's not the representative pattern of this change, however.

And sure, like any tool, the quality of work depends on the competence of the person wielding it. "A fool with a tool is still a fool" as they used to say in university; it's still true. But give a bunch of competent people powerful tools and they'll find ways to improve the output as well as the quality of their work and that's exactly what's happening right now. Pretty much the only people I see being systematically skeptical are people who are afraid to give it a serious shot themselves or people who are for some reason resistant to any form of change so they'll automatically reject whatever comes their way. Mind you, I can sympathize to an extent and I feel society should accommodate the laggards as well as the innovators, but that's a different matter altogether.

To be clear, I'm not an AI evangelist, far from it and perhaps even the opposite. But the resistance I see in some people today is just holding on to a world that's just gone forever; sorry, but things change and this innovation is like many radical innovations that have gone before it: it's a game-changer.

PS: one of the reasons @Sean has been able to boost functionality, speed and reliability of this site in some areas (although perhaps not clearly visible to many yet) is because of...AI. The fact that we can continue to enjoy this place where we can discuss yesteryear's technology is largely thanks to AI.

 

[Alan Edward Klein]

Hello,

At the beginning I was impressed by this chatbot, for a few weeks, after that I've noticed how it works and what was it purpose, I guess we all know now what it's all about with the so called AI technology, is a business meant to make lots of money from different sources, first the users then the company, the hardware manufacturers and the others companies that sell this kind of services or rent them. I have friends in the industry and they are nothing more then "gold diggers" and the gpt is the shovel. The app "per se'' is a data mining platform, like facebook, discord, X, google and so on. There is nothing innovative to it the technology was used in the '50'for and put aside for the right time, it's just a Whatsapp predictor on steroids, nothing more. In reality this kinda technology should be considered illegal if the laws were to be applyed as they are right know in most countries, it has been trained with data from illegal sources, without permission and has been trained by almost enslaved people that hat to teach the bot to discriminate and prompts answers for the user. There is nothing intelligent about this apps, just pre-configurated answers. They killed the internet that was already killed by Google and YouTube, also I've read that by the end of 2026 almost 90% of the internet will be generated artificially by bots and people will lose interest in creating man made content for the web. As a web designer I am an activist, and fight the WordPress monopoly on the CMS scene, people don't even know that the internet died because of the lack of interest of the users in learning how to code and design them self, and use the so called ''free-CMS" that early this year told the world the truth that they are collecting data and user behavior stats to train the AI... I guess will have to deal with it and go along, the companies are far to powerful to fight against, and they all have the governments approval and protection to do as they will.


I don't know if you agree but it does not matter anymore what people think or want to do... see you around.

It will kill photography for many as well.