If you're looking for a Flickr-style photography platform on the Fediverse, there's one in beta testing called Vernissage:
-
If you're looking for a Flickr-style photography platform on the Fediverse, there's one in beta testing called Vernissage:
Bear in mind it's still in testing, so it may not all work properly yet. You can follow the latest news on its official account:
Vernissage is part of the Fediverse so you can also use it to view photos and follow photographers from Mastodon, Pixelfed and other Fedi platforms.
@FediTips @vernissage It‘s an already stable solution and the development is going on there with reason and accuracy - and when a feature is there, it‘s there: No continuous „coming soon“-announcements or similar stuff - Vernissage simply delivers!
-
@FediTips @vernissage the AI generation of alt texts is a big red flag tbh
@zenmaya @FediTips @vernissage if you want to try Vernissage with that function intentionally disabled, you can at https://vernissage.pnpde.social
-
@zenmaya @FediTips @vernissage
It's a tool, and nobody has to use it. I personally love it. Many of my pictures on Vernissage.photos would otherwise have much worse alt text.
@ruanjoma @FediTips @vernissage well alt text isn't for you, its for the visually impaired, and from what ive heard they don't like auto-generated alt texts, because they can generate them as well by their screen readers, the point of alt texts is the human actually typing the text there
-
@zenmaya @FediTips @vernissage if you want to try Vernissage with that function intentionally disabled, you can at https://vernissage.pnpde.social
@dnddeutsch @FediTips @vernissage I'm just not liking them intentionally boasting about it, I do not think it's a good feature to provide to users, because it goes against the wishes of the visually impaired community.
-
@ruanjoma @FediTips @vernissage well alt text isn't for you, its for the visually impaired, and from what ive heard they don't like auto-generated alt texts, because they can generate them as well by their screen readers, the point of alt texts is the human actually typing the text there
@zenmaya @FediTips @vernissage
Okay, then I didn't understand alt text and wouldn't know what else to write there. I meant it should say what's in the picture
️ -
@zenmaya @FediTips @vernissage
Okay, then I didn't understand alt text and wouldn't know what else to write there. I meant it should say what's in the picture
️@ruanjoma @FediTips @vernissage It is, but AI is not 100% correct and users of screen readers can always just generate the description themselves. The point of writing alt text manually is the fact that the author or some other human wrote it to describe the picture. It can add context a machine cannot etc. etc.
-
@ruanjoma @FediTips @vernissage It is, but AI is not 100% correct and users of screen readers can always just generate the description themselves. The point of writing alt text manually is the fact that the author or some other human wrote it to describe the picture. It can add context a machine cannot etc. etc.
@zenmaya @FediTips @vernissage
Of course, I read the generated text and correct or add to it if necessary. Most of the time, this isn't necessary, and honestly, the texts are better in most cases than what I could write myself.
-
Paris was the place to be for artists at the time!
@FediTips And musicians!
-
@dnddeutsch @FediTips @vernissage I'm just not liking them intentionally boasting about it, I do not think it's a good feature to provide to users, because it goes against the wishes of the visually impaired community.
@zenmaya @FediTips @vernissage understandably! But oth they put a lot of thought (read: more than similar software) into accessibility, so we decided they we're a better match for our users - ymmv ofc
-
p.p.s. People are going to inevitably ask how this compares to Pixelfed. This is my take:
- Pixelfed is an Instagram-style social network platform based around images and short videos
- Vernissage is a Flickr-style photography platform for exploring still photos as an art form
Please feel free to correct if this is wrong.
I'm not in a position to talk from experience of usage, but Vernissage seems to be the one to use if EXIF meta data matters to you and Pixelfed if you only want images
-
@zenmaya @FediTips @vernissage understandably! But oth they put a lot of thought (read: more than similar software) into accessibility, so we decided they we're a better match for our users - ymmv ofc
@dnddeutsch @zenmaya @vernissage
To be clear, I loathe AI and consider it an enormous con that is corroding society and burning the planet.
However, if people aren't on the Fedi then the centralised platforms they're using instead are definitely going to be using AI in much worse ways that cannot be opted out of.
If Vernissage is keeping the AI strictly optional so that it can be turned off by an instance, then the best path forward seems to be using Vernissage on such instances.
-
@FediTips And musicians!
And writers etc. Amazing to hear about that period.
-
@ruanjoma @FediTips @vernissage It is, but AI is not 100% correct and users of screen readers can always just generate the description themselves. The point of writing alt text manually is the fact that the author or some other human wrote it to describe the picture. It can add context a machine cannot etc. etc.
@zenmaya @ruanjoma @vernissage
I'm not visually impaired, but going by advice from @WeirdWriter and others apparently the main problem is AI doesn't have any understanding of anything, it just thoughtlessly describes everything even if it's totally irrelevant, and can hit people with a wall of irrelevant description. It's easier to understand an alt text if a human has written what's actually relevant and significant about the image.
-
@zenmaya @ruanjoma @vernissage
I'm not visually impaired, but going by advice from @WeirdWriter and others apparently the main problem is AI doesn't have any understanding of anything, it just thoughtlessly describes everything even if it's totally irrelevant, and can hit people with a wall of irrelevant description. It's easier to understand an alt text if a human has written what's actually relevant and significant about the image.
@FediTips @zenmaya @ruanjoma @vernissage Bingo. Do you know if they can use hashtags such as #AltForMe? But you nailed it. We have apps and tools with AI built into the apps and or our screen readers but from my perspective AI alt text is pretty useless. especially since what people usually generate when they generate AI alt text is mostly worse than our apps but in most cases it is exactly what FediTips describes. Yes, it’s giving us information, but is it useful information?
One example I can think of right now is a comic strip described by alt bot. Not only was incorrect by saying there were three panels instead of four panels, it just mindlessly described the panels. It would describe who was in the panels, and it would describe what they were saying, but it described it in such a way to where every panel seem to be very disconnected from the other. The panels were described in verbose detail, but everything was disconnected from each other. Not only that, but I learned later that it didn’t even describe what the characters were doing/wearing, killing the joke. Alt text is just as much an experience as when you took the picture. The alt bot couldn’t make connections from one panel to the other. Each description of the panels was very verbose, yet it left out a lot of crucial information for me to understand the joke, the setting, a lot more things. Personal alt text will always be better because it forces people to live in the moment. It forces people to bring us along with them. Even if they think they’re writing is mediocre or not good enough, I will always get more out of personal alt text
-
@FediTips @zenmaya @ruanjoma @vernissage Bingo. Do you know if they can use hashtags such as #AltForMe? But you nailed it. We have apps and tools with AI built into the apps and or our screen readers but from my perspective AI alt text is pretty useless. especially since what people usually generate when they generate AI alt text is mostly worse than our apps but in most cases it is exactly what FediTips describes. Yes, it’s giving us information, but is it useful information?
One example I can think of right now is a comic strip described by alt bot. Not only was incorrect by saying there were three panels instead of four panels, it just mindlessly described the panels. It would describe who was in the panels, and it would describe what they were saying, but it described it in such a way to where every panel seem to be very disconnected from the other. The panels were described in verbose detail, but everything was disconnected from each other. Not only that, but I learned later that it didn’t even describe what the characters were doing/wearing, killing the joke. Alt text is just as much an experience as when you took the picture. The alt bot couldn’t make connections from one panel to the other. Each description of the panels was very verbose, yet it left out a lot of crucial information for me to understand the joke, the setting, a lot more things. Personal alt text will always be better because it forces people to live in the moment. It forces people to bring us along with them. Even if they think they’re writing is mediocre or not good enough, I will always get more out of personal alt text
@WeirdWriter @FediTips @zenmaya @vernissage
Hi Robert,
thank you so much for explaining the problem from an affected person's perspective.First, a quick note about Vernissage, which was the original topic of this thread. I consider it the best platform for photos in the Fediverse. The parent instance offers, among other things, the option to generate and customize alt text with AI support, and to my knowledge, other instances can also disable this option for themselves.
-
@WeirdWriter @FediTips @zenmaya @vernissage
Hi Robert,
thank you so much for explaining the problem from an affected person's perspective.First, a quick note about Vernissage, which was the original topic of this thread. I consider it the best platform for photos in the Fediverse. The parent instance offers, among other things, the option to generate and customize alt text with AI support, and to my knowledge, other instances can also disable this option for themselves.
@WeirdWriter @FediTips @zenmaya @vernissage
The question that arises in this discussion is: What should such alt text look like to be useful? We're talking about a platform for photography. While the comics example makes sense to me, I can't yet translate it to a photo.
I don't like doing pointless things, so I need an example of what proper alt text should look like to be helpful.
Could someone please look at the alt text in this example and correct it?
-
@WeirdWriter @FediTips @zenmaya @vernissage
The question that arises in this discussion is: What should such alt text look like to be useful? We're talking about a platform for photography. While the comics example makes sense to me, I can't yet translate it to a photo.
I don't like doing pointless things, so I need an example of what proper alt text should look like to be helpful.
Could someone please look at the alt text in this example and correct it?
The trick? Make alt text personal, because you will never be able to make alt text useful for everyone, so make it personal. What draws your attention to the image? I actually could not find anything unhelpful about that alt text. It sounded very cute! Pretend someone is on a phone call and they can't access a computer. How would you describe the image? But you might want to take a look at https://alt-text-as-poetry.net/ @ruanjoma @FediTips @zenmaya @vernissage
-
p.p.s. People are going to inevitably ask how this compares to Pixelfed. This is my take:
- Pixelfed is an Instagram-style social network platform based around images and short videos
- Vernissage is a Flickr-style photography platform for exploring still photos as an art form
Please feel free to correct if this is wrong.
@FediTips And to further clarify (obfuscate?), you can use the Impressia app to browse Pixelfed, giving you a more Flickr-like, text-less experience than its interface.
-
@FediTips @zenmaya @ruanjoma @vernissage Bingo. Do you know if they can use hashtags such as #AltForMe? But you nailed it. We have apps and tools with AI built into the apps and or our screen readers but from my perspective AI alt text is pretty useless. especially since what people usually generate when they generate AI alt text is mostly worse than our apps but in most cases it is exactly what FediTips describes. Yes, it’s giving us information, but is it useful information?
One example I can think of right now is a comic strip described by alt bot. Not only was incorrect by saying there were three panels instead of four panels, it just mindlessly described the panels. It would describe who was in the panels, and it would describe what they were saying, but it described it in such a way to where every panel seem to be very disconnected from the other. The panels were described in verbose detail, but everything was disconnected from each other. Not only that, but I learned later that it didn’t even describe what the characters were doing/wearing, killing the joke. Alt text is just as much an experience as when you took the picture. The alt bot couldn’t make connections from one panel to the other. Each description of the panels was very verbose, yet it left out a lot of crucial information for me to understand the joke, the setting, a lot more things. Personal alt text will always be better because it forces people to live in the moment. It forces people to bring us along with them. Even if they think they’re writing is mediocre or not good enough, I will always get more out of personal alt text
@WeirdWriter @zenmaya @ruanjoma @vernissage
In a way this is perhaps something that techy sighted people can relate to:
A long time ago I used to handle a lot of tech support calls over the phone, so just audio. I needed people to describe to me what they saw, and the calls worked best when the people described just the relevant details.
AI/LLM alt text is the equivalent of difficult tech support callers who tell you the colour of their computer's mouse but not what is on their screen.
️ 
