New AI app for describing images and video: PiccyBot

If I am in Facebook and tap an image, there is a More icon. Under that is an option to copy the image.

I presume this is going onto the clipboard but I could be wrong. Knowing me and my lack of social media skills, I'll find I've been posting them to my timeline or something stupid.

Ah I was wondering why everything was silent. Then it suddenly started working just now.

Thanks for such a quick response.

If the image is in your camera roll on your iPhone, when you go down to the share sheet, there is an option that says, copy. Not copy iCloud link, but, copy. I also know that it's possible to copy images directly on Google Chrome on the iPhone. When you find the image, if you long press on it, it will give you an option to copy the image. If you don't see the option there, if you go down to the share sheet, you should see copy image as an option. You should be able to paste the image to your notepad and interact with it from there. Sometimes, it's not always possible to paste the image directly into an artificial intelligence app for descriptions, such as this one, for some strange reason, even though the image has been copied to your clipboard. Letting you know this information, so that you can keep it in mind, in case it doesn't work. The copy image option isn't available on the Safari browser. Hope this helps.

The personality thing is just superb, especially if you're looking through your old pictures or are just killing time. Does anyone know if there's any way to use these voices from Open AI to read, say, an ebook?

I love the option of the personality but I personally couldn't stomach it for long. Literally every single photo is the most wondrous and magical thing that's ever been shared by anyone ever. The little laughs, breaths or gasps are realistic but doing that for every photo it just becomes a bit silly.

One time I had a photo and asked it to tell me which animals were in it. The voice waffled on and on about how amazing and wondrous this scene was and how special and magic the animals were. I swiped to the description and VoiceOver said "dog and a cat" which made me laugh quite a lot.

Anyway it's a good option but I can't see myself using it again but good if others are enjoying it.

I like the app, though. The AI voices are a really nice touch and being able to swap between the AI models is great.

I did have a couple of instances where I asked for a detailed description of my dog (yes that's more or less all my photos are of) and it refused saying it couldn't do that for known people or animals or something. Which is a shame. I recently lost one of my dogs and would love it to be able to describe her in a way that really brought her to life. I miss being able to look at photos myself and good though the AI is, it doesn't come anywhere near close to replacing that.

If it could make it easier for me to describe Facebook images then I would get a lot of use out of it as my sister is always posting images without saying what they are about.

One other very minor thing - I found the settings where you choose the voice a bit confusing in VoiceOver. Each voice is listed twice, the second one says button after it. It's not clear that the first time is where you select it and the second time is the preview. Well, it wasn't to me anyway. But that's only a small thing as I probably won't change voice much. I'd also love some British voices but I don't think OpenAI has any yet.

Does this app use the front or back facing camera on the iPhone? I took a picture of something laying on my bed, and it gave me a description of my bedroom.

When you take a photo, there is a switch camera button at the top which presumably toggles between front and back.

However, it doesn't seem to tell you which one it is using with VoiceOver unless I am missing something?

well, we just can't please everybody, can we? the waiting sound is probably a good idea, but personally i'd rather wait in silence. if it's not too much trouble, an option to disable the waiting sound would be great.

Thanks - that works great and I can now tell which camera is active.

I also noticed that you added a speaking rate option which is also really appreciated. I tried it at the max of 150% and it still sounds very convincing and human.

It still seems a little random as to how much detail I'm going to get out of it. For example I used the default prompt and was told there was a sofa and a couple of other things. I then asked if there were any animals and it gave me a more detailed answer about the room as well as telling me the answer. Whereas I think I find Be My AI a bit more consistent in the level of detail I get.

I like the fact you have sliders to increase the verbosity and I went straight in and set them all to max for now, but I can see some situations where less is more.

I think having all the customisations does set it apart. I can almost see a time when it might be handy to be able to set up different profiles (which could include the settings plus custom prompts) and then I can swipe up and down on the main screen to cycle between them. For example, let's say I always lose my keys, I might want a fast speaking rate, low verbosity and a prompt "where are my keys?". If I'm just wanting a description I might want it slower, with more detail. Or I might want a couple of different personality options. Anyway, that might be a terrible idea so please ignore if it makes thins too complicated.

hello the descriptions in my language are very brief eventhou i set the option for 100 procent. i do not appreaciate the opinion of the application. i appreaciate 3 engines althou the hird one does not work for me. can you add option descri,bing pictures that are instantly changing such in a movie or youtube? anyway keep up a good work and yes i, bought life time subscription to support you

i don't mind if the ai voice can't read the whole thing, but the text also got cut off, so there isn't other option, such as review it with voice over. some example here"

The image shows a person's neck with several red spots. There are a number of possible explanations for red spots on the skin, and without a proper medical diagnosis, it's hard to ascertain the exact cause. However, I can provide some general possibilities:

1. Acne or pimples: Red spots can be the result of inflammation due to acne, particularly if the person has been wearing something that rubs against the neck, causing irritation and breakouts.

2.

well, you see, there is definitely number 2, and possibly 3 or 4, but it got cut off at number 2. is this some kind of limitation or a bug?

Many of the images we get come as scanned pdf files. But PiccyBot doesn't appear in the sharing sheet of PDFs. I don't know if it's me or if it's that way. In case it's designed to not work with PDFs, including that option would be great.

i forgot to try it with other models. also, panorama photo can't be submitted using the share sheet. i can submit pano using the photo picker in the app, but not from the share sheet.

I really am loving this APP allot and I find this is better than Be My Eyes especially on the Claude 3 setting. I collect action figures and I do a bit of action figure photography, and with Be My Eyes, it has issues describing allot of the most... shall we say, "adult" looking figures and statues. (Try running photos of characters from Lady Death or Hellwitch, and you'd know what I mean LOL) This is where Piccybot on Claude 3 setting really comes through. It still has snags, some times it's acting self righteous and refuses to describe some things, but I find that if I keep at it it will generally work after rephrasing the questions a few times.
I wish that there's a website version of this APP we can access through a windows PC since most of my photos are stored on my PC and labeling/renaming them is way faster on the PC than having to swipe and click 6 or 7 things on the Iphone.
Another feedback is the "personalities" of these voices are really rather annoying at time :) I eventually just turned personality off cause they came across as condescending or snarky and tended to piss me off when all I wanted was detailed descriptions of a character.
I ahve also encountered the voiced descriptions cutting off half way quite often. Even the text shown on the screen would not show the intire description at times.
Overall, this is one of the best photo describing APPs I've used. Other than Be My Eyes and, to a lesser extent, Seeing AI, nothing else can compare at this point. Thank you for creating this little wonder of an APP :)

Hi, is this on android?

Still really enjoying the hell out of this APP. As an artist and action figure collector, I interact with lots and lots of photos and have to sort through them. This APP has been a god-send. Yes, Be My Eyes was my go-to before, but it is so sensered that it's useless when dealing with some artistic subjects.
I want to put in a feature request and suggestion for pre-set prompts. Often I am on the phone or meetings while sorting photos using the app- kinda like my little fidget toy :) but I can't dictate anything while a phone call or zoom meeting is going on, and typing by hand is slow and tedius after awhile. I sort through lots of action figures, statues and artworks, and I do have a set of questions I usually ask "Hone in on face." "Rotate to the left" "Hone in on background, ignore the figure" etc etc. It would really be useful if we can pre-program say, a set of 10/20 prompts that we can just tap on for descriptions when dictation is not possible.
Also, not sure if this is possible at all, but while I am on the phone it's really hard to hear the APP cause during a phone call other things like voiceover and sounds are lowered in volume. Is it possible to have a "phone call" mode that I can toggle on and the APP can speak louder?
Thanks for such a wonderful APP!

Would be great if pictures could be taken with, say, a double click of the side button, especially if you're trying to click some written content or when you're using the front cam.

"Would be great if pictures could be taken with, say, a double click of the side button, especially if you're trying to click some written content or when you're using the front cam."
Totally agreed. I actually submitted that suggestion to Seeing AI but they've not put it in. On Iphone's own camera app, you can start/stop video, take photos, just by clicking the volume button. I find that makes thing so much easier cause then you can take photos using the bluetooth keyfob remotes that come with selfi sticks and some tripods. With seeing AI and other apps, I find that I may have the shot lined up, but when I double tap on "take picture," it shakes the camera and things get blurred.

Thanks for the update! Took me awhile to find the "history questions" lol, it's beside the keyboard after you double tap to go to text field. Now I'm going to spend the next few days building up the history prompts, then it'll be smooth sailing!

the click with volume buttons should be a blessing and a time-saver. Will try it out!

Thanks for the update.

The volume shortcut works well from a quick play last night. I like how it skips the intermediary step of orientating etc the image which always felt a bit unnecessary to me. Not sure i that is still there if you use the camera the normal way but it's nice to skip past it.

hi,
It is not possible for me to share a screenshot to the app? When I tried it and press send on the prompt I see a quick dialogue to add a picture?
Secondly, where can I find the option to add the most recent picture?

This is a very nice app. I'm glad I got it a little later into its development cycle.

I posted elsewhere on the site about this but, what I think a real USP could be for this app is to create a new album that includes markup data and the description so, once parsed by the AI, we can quickly browse and even organise into folders then, if we wish, we can pull them out again and ask the AI more questions. Currently, browsing larger libraries is very slow indeed, so a batch means of labelling photographs with short descriptions and saving them would be awesome. I know that, if you include a short description through mark up in the apple photos album, voiceover will read it out, aside from that, if we're looking through a large library, we really have nothing to go on aside from the date an image was taken.

If you can create a way of saving photos with this meta data, you're onto a winner.

Thanks

Also, I do like the voices, it's nice to hear something more natural.

There are situations when we might need to take one picture, get the describtion, take the same picture from a different angle or may be take a new picture, and the actual thing we need might come from a comparison. so an add picture option or something like that will be nice to have.

Further request...

I would like to be able to select the voice independent of personality. Furthermore, I'd like to provide instructions, much like found in the GPT app as general prompts that influence all responses, EG, tone, specifics on certain aspects. An example might be:

You are an AI describing images for a blind fashion student. Pay attention to the styles worn, the colour combinations ane cut of the garment. Attempt to extrapolate the material used but, if uncertain, make this clear. Describe with precision rather than inturpritation.

And so on. I think this is fairly easy to implament. I do like the idea that you've already added personalities, and they should stay, but custom ones will make this much better for specific use cases.

the sharing of the photograph should also include a short description, either as a file name, or included in teh metadata as a description.

That's wonderful re the metadata, will make organising images so much easier.

Do you happen to know, does it use the 'description' field in the meta data markup as, as I understand it, this is what voiceover can read in the photos app on IOS. I put a short description in there for one of my photographs, as it's limtied to something like 120 characters.

Very interested to try this out.

I fed it a photo of a woman sitting on the bed, wearing a form fitting dress. I then asked it to "generate a 20 words descriptive title for this photo based on outfit, posture and body type." When I had the language set to Mandarin Chinese, it absolutely refused to generate the title. It kept returning something like "I am sorry, but I can't do that because it's not respectful of the person." I changed language back to English, and suddenly it's working every time LOL.
BTW, while having the prompt history is pretty good and acts as a good alternative to having pre-programable prompts, I noticed that once I rebooted my phone, it cleared all my prompt histories and that was a bit frustrating.

The labelling of the file is a game changer! it's amazing. If you do manage to make an app that can do this on batch, I'll certainly go pro on it. I'm just going through my photo album now and labelling everything up. Awesome, thanks so much for this addition.

I'm so glad I supported you, I don't need the app anymore but you're so commited to making this app the best it can be that I'm willing to donate again, is there a donate button in the app or is it a one time thing?

This app is turning out to be a game-changer in so many ways, particularly because you are continuously listening to the user feedback and getting user suggestions implemented. That also makes this practically the most useful app in day-to-day life. What would be so nice will be to see this app being integrated to something like the Meta Glasses as a subscribtion service or something.
One suggestion: it'll be nice to have a 'module' where you can ask the app to look for a specific thing or item at the first instance itself rather than first getting a general discribtion, and then asking follo-ups etc. There are instances where I perfectly know where I am, and what I need to find. For example, let's say I am in a department store. I need to just say something like "tell me my position with respect to X section" and then the app can click the picture and give the info.

I think there is a chance to spin out some different apps here, or make a suite of them similiarly to how seeing AI does it.

On the one hand, we've got a neat and simple app for browsing photographs, interrogating them which is really awesome, especially since the update.

then there is the side of it for real time photographs like Gokul says, when we're looking for items. Possibly there is an argument that there are already apps, be my eyes, for example, that might work as a better tool for this.

I guess it's what this app is and isn't. I recently got the meta glasses, which are cool for quick pieces of information, but kinda suck when it comes to deep detail or reading documents, but that's okay because I've got other tools for those. An all in one would, of course be nice, but one of the advantages of this app, to my mind, is its simplicity.

Gokul, you can ask a question of the image and another request is sent. ARe you asking for the ability to define the original question that is sent to the AI?

Personally, I'd not use this tool in public for such a task as it is slow to respond, but the quality of the responce is exclent, which makes it good for photograph browsing. I'm wondering if shortening the response character count decreases response time proportionally?

Anyway, great app. I'm excited to see where this goes. I would very much like the means of setting system prompts in the app that defines the general output, though I realise, as I've been saying that this might add a level of complexity the developer doesn't want to burden the user with.

@Ollie. Exactly that's what I wanted to have. instead of asking it for a general describtion, the user should be able to define the exact info they need in the first instance itself.
Also, while we do have apps like BeMyEyes and Seeing AI, it is always good to have alternatives out there; for instance, when you need to compare results and make sure things are exactly the way they are being described.

This is such a great new feature, thank you.

After upgrading to the latest release, Share was broken until I went in and selected an AI Model (probably because the list has changed), which confused me a bit. But now it works very well - the description is now just sat there under the photo in the Photos app.

Awesome work, thank you.

As far as I can tell, when you have personality on it will spin the description that is spoken so maybe turning personality off will make the written text the same as spoken.

It's a great app for a specific task. The fact that it can label pictures now so I can dip into my photo album, for me, is a game changer. It's also why I'd, personally, not want to see the scope of this app extend too far beyond analysing existing photographs. I see it as a calloguing app for the blind rather than a live action app like be my eyes, seeing AI and so on. It is slow, but I imagine this will improve over times, the bottleneck being the AI call and response which isn't going to be amazing as it is a free account, I'd guess, rather than something like a GPT4 subscription that has priority.

What would be nice, and the fact we can choose AIs is awesome, is for apps like this to allow us to not only connect to our prefered AI, but a paid AI. I believe we're headed in the direction of personal AI now, choose your flavour and then it becomes yours, knowing you, your friends, etc. It would hold a lot of data about you, but so does google/apple/microsoft anyway.

Sorry, I do think this app is great and the way the dev is continuing to improve it is admirable, it's a bit of my sweet heart app at the moment so I'll defend it to the death!!! :)

Hey developer, I have been using other AI models to get descriptions for the last couple of months and stumbled upon this a few months back so I decided to get a paid version of it and I'm loving it. There's a lot of good features and options so that made me want to pay for it. I do have one suggestion though, the share sheet situation is a bit aggravating because I choose to share photos from X which is formally known as Twitter and when I click on this app to get a description it basically just brings me to the app like I was opening it for the first time and I'm not able to get any descriptions unless I would have to save the photo to my photos app and then hit it from there to get the description. Not all the time I want to save photos to my album just to get descriptions. What's going on with that? And sometimes when I hit share from my Photos app it basically brings the app open but it does not get the description I would have to actually open the photo from the album through the app and select it from there. Is there a way for you to fix that?
You are doing an amazing job at this. I gave you kudos for all your work and then helping us by listening to our suggestions. Even if this doesn't get fixed I'm still going to continue to support you and the development of this app. I'm loving how I can save photos now with the description in my albums! Keep up the good work and mini blind/visually impaired people are loving what you do because it's unlike any other AI models.

I just reread my comment and am struggling to find any point that deserved that reply. I made it clear that everything I said was my own opinion and I did not attack your opinion in any way. I made a suggestion regarding the technical aspect of the voiced results not being the same as the written result, but that's just about it.

My opinion has no more validity than anyone else's and my intention was not to diminish yours.

I think the voices are one of my favourite things about the app. I am a sucker for that sort of thing.

I wonder if Martin's problem was the same one I had. I presumed I just needed to switch model but maybe that was a coincidence. I was finding that the app would start but not speak anything.

One related thing I've noticed - when I come out of Settings, sometimes it seems to then re-process the current image and sometimes it doesn't. If I wanted to try the same image with a different model, and I go to settings and switch, it doesn't always have another try. And if I go to the text prompt and tell it to go then it won't because I've not changed anything. Not a particularly big deal, just a tiny bit fiddly.

it identified my coffee roaster. no other ai model or any image recognition program have ever been able to recognize it as a coffee roaster, they just called it a machine, some contraption or something that was completely wrong, like a oven or a vacuum. opus also stated the name of the roaster, a detail description of the machine, so much so i suspect it searched the web to gather info about this particular roaster?

The APP is getting hung up on the previously described photo. For example, I had a photo described last night. This morning I shared another photo with it, and it gave me the same description of the photo last night. I thought it was me selecting the wrong photo, so I shared that photo with another app, it described correctly. I then shared it with Piccy again, and again, it described the photo from last night. I even prompted it with "That's not the correct photo" and it came back with "I am sorry for any confusion" and then described last night's photo again... I had to physically use the camera button in the app to select the new photo for it to finally describe it properly, but now it's stuck on that one and won't move on to the next shared photo LOL.
It's like the cache is not getting cleared. I even restarted the APP but still no good.
Also, not sure if this would be doable or not, but when we export/share from the APP to say, "save to dropbox" it would be nice if the App can put a descriptive title in the file name. Right now how I been doing it is, I'll ask the APP "generate a 30 words descriptive title for the photo." and once I got the title I want, I copy it, go export the photo, save to dropbox, and manually paste the title into file name. It's a bit lengthy and tedius, so if the APP has a function to do that it'd be really nice.

I tried this app, but the thing I don't like is that the description is too poetic and imaginative, that is, it doesn't describe what's in the image, but it makes all its personal considerations from the image which don't help.