AI description apps installable on iOS that have an explore by touch?
By
OldBear, 27 May, 2026
Forum
iOS and iPadOS
I've lost track, do any of the other AI description apps have an explore by touch feature to show where things are on the screen yet. This would need to be an app that installs from the app store etc.
Thanks Paige. Seeing AI has that feature, but for the last few days, the photo analysis has been putting up a, sorry it didn't work, message up almost all the time. I've tried everything, except deleting and reinstalling the app, including a different network. It's never done this for so long before and I'm thinking I need to work out a backup method for my work because Seeing AI can't describe any new photos, or if I edit any of the ones I've already had described.
That explore by touch is kind of a final check because AI can get things wrong in the positions of pictures. But being able to use the share sheet to get a description after each change is really important too.
I decided to try the explore by touch on Seeing AI for one of the photos it is not analyzing, and it worked. I have know idea what features are sent off to a server and which are done locally. I can't get a description of a picture, but I can explore it on the screen. If this works with all photos, I can still use it for my work. It just adds new layers to the tediousness.
I haven't used Seeing AI for describing images for quite a long time now, but after reading this I just fired up Seeing AI, took a random picture of my TV screen and had it described. Everything went smoothly: the processing sound was played and after about 3 seconds I got the description which was okay. I also checked explore by touch on this image and it worked well too.
Two things come to my mind: maybe there was a longer standing server problem (e.g. Piccybot had that sort of problem according to its dev for some weeks, but got fixed recently) and that got sorted out just now or my second guess is that maybe the images you tried to analyse were somehow in the scope of the censorship of the model. I don't know what model Seeing AI currently uses, but for certain models an image needn't be explicit or "harmful" at all to be flagged by some censorship. Certain models are becoming extremely picky nowadays. I know from experience that Seeing AI has never told the user when the censorship blocked an image description,it always just errored out in that case.
By the way I think explore by touch is also an online feature of Seeing AI. I haven't checked it, but I make an educated guess on this as a processing sound is also played before presenting the image for exploration, and the processing takes about the same amount of time as for descriptions. I think what's going on there is that Seeing AI sends the image to the same model, but with a different prompt and requests a list of objects and their locations instead of a description. As the state of AI currently stands, I can easily imagine that the model refuses to describe a certain image verbally, but "gladly" gives a list of object - location pairs.
So all in all I suggest a recheck, as things may be back in working order now.
I'm having difficulty with Seeing AI analyzing photos I've taken with the Camera app that are in the Photo app library.
It looks like you are talking about the in-app picture feature of Seeing AI, and that does fine for some strange reason. If you swipe around there is a Brows Photo option on that screen. That takes you to the Camera Roll or what ever they call it now.
Seeing AI has, sense Monday, not been able to process those photos most of the time.
It's probably their servers, and will eventually be fixed, but it's an unreliability that worries me.
As I posted before, the explore by touch seems to work whether the photo can be analyzed and described or not. So not quite as desperate a situation. I'm trying not to start blaming myself for using the service too much or other madness.
I have just tried analysing two pictures from my camera roll through that "Browse photos" interface and both were successful and explore by touch also worked on them as expected.
It has just come to my mind that the appearance of this problem might depend on geographical location. Maybe something was misconfigured for the model server, but this issue might affect only tthose who reside in certain geographical zones. This can happen indeed, e.g. Cloudflare-based server solutions tend to trigger location-dependent problems sometimes.
Thanks Laszlo. I was starting to think that. I guess it will eventually clear up, but it is a troubling unreliability.
I also guess that Seeing AI is the only app that offers the explore by touch. Luckily, that processing seems to be more local to the phone than the actual descriptions, and does still work, even if the analysis fails.
I have just ascertained that with a simple test. Without any net connection (e.g. in airplane mode) the explore by touch facility in Seeing AI doesn't work. However the error message Seeing AI gives in this scenario is misleading, because it says "no info found in image".
So explore by touch in Seeing AI is definitely a cloud-based feature. But the AI infrastructure behind it seems to beconfigured better, and that is why it seems to work for those for whom image descriptions are not generated currently.
Comments
seeing ai
Last I checked Seeing AI can do this but that was awhile ago so things could have changed
Yes
Thanks Paige. Seeing AI has that feature, but for the last few days, the photo analysis has been putting up a, sorry it didn't work, message up almost all the time. I've tried everything, except deleting and reinstalling the app, including a different network. It's never done this for so long before and I'm thinking I need to work out a backup method for my work because Seeing AI can't describe any new photos, or if I edit any of the ones I've already had described.
That explore by touch is kind of a final check because AI can get things wrong in the positions of pictures. But being able to use the share sheet to get a description after each change is really important too.
Very Strange...
I decided to try the explore by touch on Seeing AI for one of the photos it is not analyzing, and it worked. I have know idea what features are sent off to a server and which are done locally. I can't get a description of a picture, but I can explore it on the screen. If this works with all photos, I can still use it for my work. It just adds new layers to the tediousness.
No issues for me whatsoever
I haven't used Seeing AI for describing images for quite a long time now, but after reading this I just fired up Seeing AI, took a random picture of my TV screen and had it described. Everything went smoothly: the processing sound was played and after about 3 seconds I got the description which was okay. I also checked explore by touch on this image and it worked well too.
Two things come to my mind: maybe there was a longer standing server problem (e.g. Piccybot had that sort of problem according to its dev for some weeks, but got fixed recently) and that got sorted out just now or my second guess is that maybe the images you tried to analyse were somehow in the scope of the censorship of the model. I don't know what model Seeing AI currently uses, but for certain models an image needn't be explicit or "harmful" at all to be flagged by some censorship. Certain models are becoming extremely picky nowadays. I know from experience that Seeing AI has never told the user when the censorship blocked an image description,it always just errored out in that case.
By the way I think explore by touch is also an online feature of Seeing AI. I haven't checked it, but I make an educated guess on this as a processing sound is also played before presenting the image for exploration, and the processing takes about the same amount of time as for descriptions. I think what's going on there is that Seeing AI sends the image to the same model, but with a different prompt and requests a list of objects and their locations instead of a description. As the state of AI currently stands, I can easily imagine that the model refuses to describe a certain image verbally, but "gladly" gives a list of object - location pairs.
So all in all I suggest a recheck, as things may be back in working order now.
That is a different feature
I'm having difficulty with Seeing AI analyzing photos I've taken with the Camera app that are in the Photo app library.
It looks like you are talking about the in-app picture feature of Seeing AI, and that does fine for some strange reason. If you swipe around there is a Brows Photo option on that screen. That takes you to the Camera Roll or what ever they call it now.
Seeing AI has, sense Monday, not been able to process those photos most of the time.
It's probably their servers, and will eventually be fixed, but it's an unreliability that worries me.
As I posted before, the explore by touch seems to work whether the photo can be analyzed and described or not. So not quite as desperate a situation. I'm trying not to start blaming myself for using the service too much or other madness.
No issues for me with that either
I have just tried analysing two pictures from my camera roll through that "Browse photos" interface and both were successful and explore by touch also worked on them as expected.
It has just come to my mind that the appearance of this problem might depend on geographical location. Maybe something was misconfigured for the model server, but this issue might affect only tthose who reside in certain geographical zones. This can happen indeed, e.g. Cloudflare-based server solutions tend to trigger location-dependent problems sometimes.
Yep
Thanks Laszlo. I was starting to think that. I guess it will eventually clear up, but it is a troubling unreliability.
I also guess that Seeing AI is the only app that offers the explore by touch. Luckily, that processing seems to be more local to the phone than the actual descriptions, and does still work, even if the analysis fails.
Net is required for explore by touch
I have just ascertained that with a simple test. Without any net connection (e.g. in airplane mode) the explore by touch facility in Seeing AI doesn't work. However the error message Seeing AI gives in this scenario is misleading, because it says "no info found in image".
So explore by touch in Seeing AI is definitely a cloud-based feature. But the AI infrastructure behind it seems to beconfigured better, and that is why it seems to work for those for whom image descriptions are not generated currently.
Thanks Laszlo
It has improved on my app today. I still have to retry questions, but the initial analysis seems to go through and generate a description.