I’m not really interested in computers
So tend to avoid contact if possible
However yesterday I booted up my aging laptop and sat there for a while deciding what to type in, what did I want to know…??
I don’t have a photographic style, basically because I don’t get out enough with my camera.
The nights are drawing in so I’m wanting to “read up” over the dark winter evenings
I’ve always been drawn to the cinematic cine still look. I tried to emulate this using basic adjustments in camera. Note I’m only shooting JPEGs- have zero interest in opting for RAW.
Because I’m a Luddite I’m clearly behind the door with regards to progress. I’ve heard people mention chatGPT and that it opened up insights I might have not thought of
So, for the past hour this morning (woke up at 5!!!) I typed in ChatGPT and asked it to recommend settings for a canon R8 using only JPEG to create a cinematic look…
OMG!!, my eyes have been opened. I fully expected a response saying question not understood?? 🤣 instead it came back with picture style recipes along with ratios I should begin with, told me to start with either 35mm or 28mm, opt for 5.6 or 8, what light balance I should try… even advising on WB Shift A1, M1…
Wow, wow, WOW!!
I’m not even sure if these settings will achieve anything but I’m absolutely blown away that the computer replied to my original question.
I’m now asking how to emulate the Stephen Shore look using a canon R8 and using only JPEG
Brilliant,
I’m guessing all this is old hat to you chaps? But it’s all new and exciting for me.
Modern LLMs will answer any questions that you prompt it to answer. They work by probabilistically —which is why I and lots of others call the stochastic parrots— picking what is the next word/ sentence they need to generate. This means that there is always an answer, but not necessarily that the answer is correct, logical, or otherwise feasible. ;-)
However, I would trust their answers for such questions. The likes of OpenAI have spent a lot of time downloading and training their models on forum posts, YouTube videos, and various guidelines available online. The chances of getting it right are quite high.
Recklesscoding wrote:
Modern LLMs will answer any questions that you prompt it to answer. They work by probabilistically —which is why I and lots of others call the stochastic parrots— picking what is the next word/ sentence they need to generate. This means that there is always an answer, but not necessarily that the answer is correct, logical, or otherwise feasible. ;-)
However, I would trust their answers for such questions. The likes of OpenAI have spent a lot of time downloading and training their models on forum posts, YouTube videos, and various guidelines available online. The chances of getting it right are quite high. ...Show more →
Yes, that's one of the easiest use cases since there are only a limited number of picture style combinations.
But when you ask Al about the "best" borkeh you will get differing answers due to human subjectivity scraped from the webs.
I've noticed that a lot of the answers one gets from artificial "intelligence" online is like asking for directions in cultures where people feel compelled to give an answer even if they have no clue where the destination is because they're too embarrassed to admit they don't know. You'll always get an answer, no matter how wrong it is.
The main problem is that it will answer your question even if it doesn’t really know or its sources were wrong. It can be helpful, but you will still have to see if it’s correct.
johnvanr wrote:
The main problem is that it will answer your question even if it doesn’t really know or its sources were wrong. It can be helpful, but you will still have to see if it’s correct.
