NYCat Posted September 10 Share Posted September 10 It got a red arrow. Take that you machine! Paulette 1 1 1 Link to comment Share on other sites More sharing options...
geogphotos Posted September 10 Share Posted September 10 While integrating AI for generating image descriptions and keywords can streamline the process for POD (Print on Demand) sites, it also presents several potential drawbacks. Firstly, the reliance on AI-generated descriptions might result in a loss of nuance and depth that human contributors provide. AI systems, while improving, can still struggle with capturing intricate details or the contextual significance of certain elements in an image, leading to generic or inaccurate descriptions. Moreover, AI-generated keywords might not always align with the specific search behaviors or needs of users, potentially impacting the discoverability of images in a less precise manner. The process might also inadvertently reinforce biases present in the AI training data, affecting the diversity and inclusivity of the keywords generated. Furthermore, over-reliance on AI tools could reduce opportunities for creative input from human contributors, potentially stifling innovation and the personal touch that adds value to content. While AI has the potential to enhance efficiency, it’s important to balance this with the benefits of human oversight and creativity. In essence, while AI can be a powerful tool in streamlining certain aspects of image description and keyword generation, it’s crucial to be mindful of its limitations and maintain a role for human input to ensure quality and relevance. 1 Link to comment Share on other sites More sharing options...
M.Chapman Posted September 10 Share Posted September 10 32 minutes ago, geogphotos said: While integrating AI for generating image descriptions and keywords can streamline the process for POD (Print on Demand) sites, it also presents several potential drawbacks. Firstly, the reliance on AI-generated descriptions might result in a loss of nuance and depth that human contributors provide. AI systems, while improving, can still struggle with capturing intricate details or the contextual significance of certain elements in an image, leading to generic or inaccurate descriptions. Moreover, AI-generated keywords might not always align with the specific search behaviors or needs of users, potentially impacting the discoverability of images in a less precise manner. The process might also inadvertently reinforce biases present in the AI training data, affecting the diversity and inclusivity of the keywords generated. Furthermore, over-reliance on AI tools could reduce opportunities for creative input from human contributors, potentially stifling innovation and the personal touch that adds value to content. While AI has the potential to enhance efficiency, it’s important to balance this with the benefits of human oversight and creativity. In essence, while AI can be a powerful tool in streamlining certain aspects of image description and keyword generation, it’s crucial to be mindful of its limitations and maintain a role for human input to ensure quality and relevance. Also written by AI? Mark Link to comment Share on other sites More sharing options...
Jeffrey Isaac Greenberg Posted September 10 Share Posted September 10 28 minutes ago, M.Chapman said: Also written by AI? ask AI** **not written by AI Link to comment Share on other sites More sharing options...
Alex Ramsay Posted September 11 Share Posted September 11 14 hours ago, geogphotos said: While integrating AI for generating image descriptions and keywords can streamline the process for POD (Print on Demand) sites, it also presents several potential drawbacks. Firstly, the reliance on AI-generated descriptions might result in a loss of nuance and depth that human contributors provide. AI systems, while improving, can still struggle with capturing intricate details or the contextual significance of certain elements in an image, leading to generic or inaccurate descriptions. Moreover, AI-generated keywords might not always align with the specific search behaviors or needs of users, potentially impacting the discoverability of images in a less precise manner. The process might also inadvertently reinforce biases present in the AI training data, affecting the diversity and inclusivity of the keywords generated. Furthermore, over-reliance on AI tools could reduce opportunities for creative input from human contributors, potentially stifling innovation and the personal touch that adds value to content. While AI has the potential to enhance efficiency, it’s important to balance this with the benefits of human oversight and creativity. In essence, while AI can be a powerful tool in streamlining certain aspects of image description and keyword generation, it’s crucial to be mindful of its limitations and maintain a role for human input to ensure quality and relevance. 😆 Link to comment Share on other sites More sharing options...
StokeCreative Posted September 11 Share Posted September 11 I've been using jbphotos extension for a couple of weeks and I have to say I love it. There have been a few images where it's been sufficiently accurate (even nuanced) where I haven't need to edit. Mostly though it doesn't get the place/ vehicle etc unless iconic, but that's ok. I usually delete the last sentence which is often "with a background of beautiful trees" or similar. Whatever though it's a great time saver especially with images with a lot of text as it "reads" it and populates title and keywords. (Eg gravestones, plaques, labels). Also jb fixed an issue for me when it stopped working. If you use it, buy him a virtual coffee or two, all creatives/developers should be paid! (I'm guessing he's a photographer since he's on Alamy) I now actually look forward to keywording, it's less of a chore, give it a whirl! Hugh 1 1 Link to comment Share on other sites More sharing options...
Phil Robinson Posted September 11 Share Posted September 11 (edited) On 10/09/2024 at 09:47, Paul Lawrenson said: How bizzare, I ask a question about an incredibly good keywording system and get a lecture on how to sell images on Alamy!!! fwiw I know exactly how to keyword to achieve sales on this platform, whether it's worth it for the $ involved I don't know 🙄 I hope you don't mean me. I was responding to the original idea of the post - AI keywording may be coming soon to Alamy. My comments weren't aimed at you. I saw this topic pop up and hadn't seen it before (I don't come here much) and read some posts at the beginning. I was just expressing the hope that AI keywording wouldn't become the norm and photographers wouldn't be relying on it to do the work for them, as in my experience it's even less reliable than Google Translate. Edited September 11 by Phil Robinson Link to comment Share on other sites More sharing options...
Martin L Posted September 11 Share Posted September 11 5 hours ago, StokeCreative said: I've been using jbphotos extension for a couple of weeks and I have to say I love it. There have been a few images where it's been sufficiently accurate (even nuanced) where I haven't need to edit. Mostly though it doesn't get the place/ vehicle etc unless iconic, but that's ok. I usually delete the last sentence which is often "with a background of beautiful trees" or similar. Whatever though it's a great time saver especially with images with a lot of text as it "reads" it and populates title and keywords. (Eg gravestones, plaques, labels). Also jb fixed an issue for me when it stopped working. If you use it, buy him a virtual coffee or two, all creatives/developers should be paid! (I'm guessing he's a photographer since he's on Alamy) I now actually look forward to keywording, it's less of a chore, give it a whirl! Hugh It was very useful for generating keywords from a 'different perspective' to augment mine. Unfortunately since the update on Sept 3rd it no longer works in Chromium on Ubuntu 22.04. It was good while it lasted. I'm probably the only user on Linux so I doubt it will be worth the effort for @jbphoto to debug and fix. Link to comment Share on other sites More sharing options...
StokeCreative Posted September 12 Share Posted September 12 20 hours ago, Martin L said: It was very useful for generating keywords from a 'different perspective' to augment mine. Unfortunately since the update on Sept 3rd it no longer works in Chromium on Ubuntu 22.04. It was good while it lasted. I'm probably the only user on Linux so I doubt it will be worth the effort for @jbphoto to debug and fix. Also worth noting that it's only free for 30 images a month (just realised!). Subs are too steep for me...back to using (my) brain power! Link to comment Share on other sites More sharing options...
jbphoto Posted September 12 Share Posted September 12 He he, I see lots of AI comments on this thread, it's still quite easy to spot! On 11/09/2024 at 13:35, StokeCreative said: I've been using jbphotos extension for a couple of weeks and I have to say I love it. There have been a few images where it's been sufficiently accurate (even nuanced) where I haven't need to edit. Mostly though it doesn't get the place/ vehicle etc unless iconic, but that's ok. I usually delete the last sentence which is often "with a background of beautiful trees" or similar. Whatever though it's a great time saver especially with images with a lot of text as it "reads" it and populates title and keywords. (Eg gravestones, plaques, labels). Also jb fixed an issue for me when it stopped working. If you use it, buy him a virtual coffee or two, all creatives/developers should be paid! (I'm guessing he's a photographer since he's on Alamy) I now actually look forward to keywording, it's less of a chore, give it a whirl! Hugh Thanks! I've not noticed the ending with beautiful trees. I also use the very same extension, maybe I just have less trees on my photos (or that mine are not beautiful?) For now, I populate my location-specific keywords manually, I could make it read the EXIF metadata and/or existing keywords to give it more context when generating keywords. As usual, ideas are plenty, time to execute them is more constrained ... On 11/09/2024 at 19:15, Martin L said: It was very useful for generating keywords from a 'different perspective' to augment mine. Unfortunately since the update on Sept 3rd it no longer works in Chromium on Ubuntu 22.04. It was good while it lasted. I'm probably the only user on Linux so I doubt it will be worth the effort for @jbphoto to debug and fix. Hi, feel free to send me an email with details or screenshots on how it's broken for you and I'll see if that's something I can understand. Since it's a Chrome-based extension it should work on Chrome (even Edge maybe) regardless of the device OS. Linux users are able to debug their own issues, right, otherwise how would they even be able to boot? Just kidding, I am on Linux on some of my machines, I think Linux users deserve more love and care! 5 hours ago, StokeCreative said: Also worth noting that it's only free for 30 images a month (just realised!). Subs are too steep for me...back to using (my) brain power! Ah yes, I've had to set up a limit this week, I wrote 30 but it's actually allowing 100 over 30 rolling days, have you reached that already? It seems my extension is getting popular, and some days I get the same Chrome user generating keywords for over 100 images in a row, and the truth of the internet is that everything that looks free actually has a cost, so it's either paywall or I turn it off. If/when I have time, I could try offering an alternative with another model provider, for example the Vertex API from Google is cheaper, and/or downsizing the image before analyzing it. However, I think there's less value in a solution that produces half-baked keywords, as it would require as much time fixing keywords as keywording manually from scratch. What I like with GPT-4o-vision (and everything OpenAI releases, it keeps getting better, it recognizes more animals, locations, plants, etc) is that the quality is high enough and I can really save 30-60 seconds per image. And hopefully make more sales thanks to keywords that I wouldn't have thought of myself, but it's hard to do 1 change and observe results immediately in stock, at least for me. One sale would pay for the sub (if it's not one of these tiny amounts I see more and more recently), so I wish you at least one 😉 Link to comment Share on other sites More sharing options...
Martin L Posted September 13 Share Posted September 13 23 hours ago, jbphoto said: He he, I see lots of AI comments on this thread, it's still quite easy to spot! Thanks! I've not noticed the ending with beautiful trees. I also use the very same extension, maybe I just have less trees on my photos (or that mine are not beautiful?) For now, I populate my location-specific keywords manually, I could make it read the EXIF metadata and/or existing keywords to give it more context when generating keywords. As usual, ideas are plenty, time to execute them is more constrained ... Hi, feel free to send me an email with details or screenshots on how it's broken for you and I'll see if that's something I can understand. Since it's a Chrome-based extension it should work on Chrome (even Edge maybe) regardless of the device OS. Linux users are able to debug their own issues, right, otherwise how would they even be able to boot? Just kidding, I am on Linux on some of my machines, I think Linux users deserve more love and care! Ah yes, I've had to set up a limit this week, I wrote 30 but it's actually allowing 100 over 30 rolling days, have you reached that already? It seems my extension is getting popular, and some days I get the same Chrome user generating keywords for over 100 images in a row, and the truth of the internet is that everything that looks free actually has a cost, so it's either paywall or I turn it off. If/when I have time, I could try offering an alternative with another model provider, for example the Vertex API from Google is cheaper, and/or downsizing the image before analyzing it. However, I think there's less value in a solution that produces half-baked keywords, as it would require as much time fixing keywords as keywording manually from scratch. What I like with GPT-4o-vision (and everything OpenAI releases, it keeps getting better, it recognizes more animals, locations, plants, etc) is that the quality is high enough and I can really save 30-60 seconds per image. And hopefully make more sales thanks to keywords that I wouldn't have thought of myself, but it's hard to do 1 change and observe results immediately in stock, at least for me. One sale would pay for the sub (if it's not one of these tiny amounts I see more and more recently), so I wish you at least one 😉 Linux users don't need love and care they just want longer ponytails, bigger beards, stronger deodorant and more real ale! It's easy to explain what happens.......nothing Right click and select 'Generate' Before you used to get a flashing 'Wait' icon and then a message box to say keywords complete. Now, nothing. No wait icon, no message and no keywords. This is on the latest version of Chromium on Ubuntu 22.04. I suppose an update could have booked it but I can't debug it so no idea. Good luck, if you need any other info or help testing or want me to try something, let me know. 1 Link to comment Share on other sites More sharing options...
M.Chapman Posted September 14 Share Posted September 14 20 hours ago, Martin L said: Linux users don't need love and care they just want longer ponytails, bigger beards, stronger deodorant and more real ale! It's easy to explain what happens.......nothing Right click and select 'Generate' Before you used to get a flashing 'Wait' icon and then a message box to say keywords complete. Now, nothing. No wait icon, no message and no keywords. This is on the latest version of Chromium on Ubuntu 22.04. I suppose an update could have booked it but I can't debug it so no idea. Good luck, if you need any other info or help testing or want me to try something, let me know. I'm seeing exactly the same behaviour in Brave Browser on Mac OS Ventura. I haven't used it for a while so a number of things may have changed. I tried clearing the cache and removing and reinstalling the extension, but still doesn't work. I tried Google Chrome and it didn't work there either. Mark Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now