[Announcement] Prizmo Go for iPhone & iPad. Instant Text Capture.
Dear AppleVis members,
I'm Raphael from Creaceed, we are an indie company from Belgium making apps for Apple devices. We make Prizmo, that you may have heard of before. I hope this forum is the appropriate place to discuss new app releases (if that's not the case, please let me know).
I wanted to let you know we are releasing a new app today (April 27th), Prizmo Go, whose purspose is to quickly capture & recognize text using iPhone or iPad camera. The app is fully compatible with VoiceOver, and provides text-to-speech and shooting instructions, among other accessibility features. It can be downloaded on the App Store and tested for free, but requires purchase to export text or use the higher accuracy Cloud OCR feature.
Below you'll find the app description and links, and I'm available in case you have any question about it.
Raphael Sebbe, creaceed.com
Prizmo Go - Instant Text Capture
Aim at text. Shoot. It's yours! Prizmo Go lets you quickly grab printed text with the camera. After text is recognized in a blink of an eye, you can interact with it in many useful ways. Or just send it to other apps. And if you've got a Mac, you can even copy/paste it there (*). No more retyping. Feels like magic, just at your finger tips.
Additionally, Prizmo Go provides rich interactions with the captured text. After shooting, text from the original picture is revealed, and swiping through it allows fast and accurate text selection directly from the image. Selected text can then be read aloud, or you can tap to browse to any printed website address, call phone numbers, trigger Mail app from an email address, or even reveal physical address in Apple Maps and initiate navigation to that place.
Prizmo Go comes with enhancements specifically built for VoiceOver, in addition to spoken guidance prior to shooting. That, combined with its text-to-speech capabilities, make it a great companion in case you need help reading printed documents.
Note: Prizmo Go won't let you create PDF documents. If that's what you need, please refer to the standard version of Prizmo.
- Camera-based text capture for iPhone & iPad
- Real-time text highlighting in camera preview
- Flash and device orientation detection
- Innovative image stabilization (sharpness tracking)
- Import from photo album or iOS document picker
- Advanced image preprocessing (cleanup, perspective correction, dynamic rescale)
- Reliable Built-in OCR (works without internet connection), 10 languages (see below)
- Accurate Cloud OCR (***) option (Internet connection required), 22 languages (see below)
- QR code recognition
- Text selection by swiping through the picture
- Copy / paste to other apps or to the Mac (*, **)
- Interaction with detected information: phone number, email addresses, physical addresses, websites (**)
- Audio (text-to-speech) reader with word highlighting, including background audio playback & sentence navigation
- Full support of VoiceOver and improved workflow for visually impaired and blind users
- Text-to-speech output & speaking rate control
- Specific image processing to handle unaligned text and reversed configurations
- Spoken guidance prior to actual shooting
- Deep integration with Dynamic Type for improved & customisable text appearance
OTHER FEATURES & AUTOMATION
- Optimized for iOS 10
- Copy to the Mac with Universal Clipboard (*)
- 3D Touch Quick Actions & haptic feedback (iPhone 7 or later)
- iPad Multitasking: support for Slide Over / Split View
- Supports Action Extension and 'Open In': other apps can send an image to Prizmo Go for processing
- URL scheme mechanism (x-callback-url) to trigger Prizmo from other apps like Workflow, and retrieve recognized text. Works with camera input or provided input image (clipboard).
AVAILABLE OCR LANGUAGES
- Built-in OCR (10): Danish, Dutch, English, French, German, Italian, Norwegian, Portuguese, Spanish, Swedish.
- Cloud OCR (22): Arabic, Chinese, Chinese, Czech, Danish, Dutch, English, Finnish, French, German, Greek, Hungarian, Italian, Japanese, Korean, Norwegian Bokmål, Polish, Portuguese, Russian, Spanish, Swedish, Turkish.
(*) Requires a Mac running macOS Sierra (or later) that is compatible with Universal Clipboard.
(**) Requires purchase of Export Pack.
(***) Requires purchase of Export Pack and purchase of Cloud OCR units.
Pricing and Availability:
Prizmo Go is available starting April 27th worldwide as a free app on the App Store, and is localized in 16 languages. The application can be tried out, but access to text is limited. Text access can be unlocked by purchasing Export Pack for $4.99 as a one-time purchase. Cloud OCR requires units to be operated, units that can be purchased directly within the app: 100-pack is $0.99, 1000-pack is $4.99. Cloud OCR can also be tried out prior to purchasing.
Creaceed is a Belgian company that has been making apps for 9 years, focusing on image processing technologies in the fields of document scanning, note taking, photo and video processing. Other apps like Hydra, Carbo, and Prizmo have been awarded by the press & featured by Apple on multiple occasions.
well I tried this app, I never had much luck with other versions of prizmo and this is the same thing, I am not quite sure how the app gives shooting instructions I get haptic feedback if I tilt the phone or flick through the options with VoiceOver. But that's about it, after a few minutes of using it I never got anything recognizable, its a shame as the cloud OCR might have been interesting. But I would say that at least on first go knfb reader seems to still be the gold standard as far as an accessible OCR solution goes. Hopefully this app can be improved. Its great that the company is considering the needs of blind users, but there is far more to getting good results than just good VoiceOver accessibility, but full marks to you for trying and maybe it might improve in the future. I mean obviously this can't be a clone or copy of knfb reader but I would say feedback and helping to line up with text needs a lot of work, before I remember when prizmo used to try to give feedback to help line up with text it was two chatty and unreliable, knfb readers field of view report solves this problem quite nicely. Just for reference I got my first usable results from knfb reader within a minute or two of trying it for the first time, with this app tried about 5 or 6 pictures and got rubbish back. I tell you what I would really love would be a good realtime OCR solution that is a gap in the market knfb reader hasn't filled yet, and hopefully if this app could be improved to help with lining up it might be equally as good as knfb reader in time.
Hi Alex, first of all, thank you for your feedback.
The scene description button is actually less chatty than the other Prizmo. I can't tell if you could try it in this case. It gives instructions about the text angle and clipped document border, among other things.
About OCR accuracy, in our own testing, we found Prizmo Go's Cloud OCR option to be highly competitive against other apps. I won't comment with numbers or other kinds of evidence here, as I'm involved in the development of this product. This can be easily tested as apps do accept image files as input, which makes it possible to compare pure OCR performance on the same images. But I want to make sure you actually try the Cloud OCR option, as opposed to the built-in one, as this option makes a big difference here, especially with more difficult images.
About VoiceOver support, well, we had blind users who helped us improve the experience during development. It's not just VO labels, we had deeper thoughts about the workflow.
I would like to suggest that you allow some time to get used to it. It's a new app, and it may need some time to use it at its full potential.
We are taking all feedback though, so be sure to let us know about your experience at email@example.com if you want to.
I had better luck, but have only tried a few simple tests.
I downloaded the app, took a chance and paid for the functionality that would allow me to share the text, and pointed the phone at two different checks that needed cashing. I did not do anything to line up the camera, so I have no idea what kind of feedback the app might give. However, the text on the first check was read successfully and then I tried the second one. That translated to gibberish, so I tried turning the check 180 degrees and took another picture. It worked. There is an cloud-based OCR service available, but I have not tried that yet.
My first impression is favorable, but I will be testing it out much more thoroughly. As was noted above, I deeply appreciate the effort to make this accessible for VoiceOver users and look forward to the app becoming another one of those programs that we can all find very useful. Thanks for your current and future efforts.
Hi, thanks for the response, OK maybe I am doing something wrong I don't see any button called scene description when I launch the prizmo go app, going from left to right I see the following options, application settings, flash which can be on or off, stabilization same as flash either on or off, English, built in OCR, take picture, import image, and that's it. Just wondering why is the take picture button also being announced as a heading by VoiceOver? sounds like maybe that's a bug? I would say given that lining up is so important, this describe scene button shouldn't be tucked away somewhere, I hope its not a paid feature now that would be very disappointing, given that lining up is kind of important.
The "Scene Description" button is not a paid feature and it's available when VoiceOver is enabled.
The button is located just on the right side of the Take Picture button.
The button has two VoiceOver features.
1. The label of the button: it's dynamic and it's an overview of the current camera settings (stabilization on/off, flash on/off, ocr language, …).
2. Tapping on the button will shoot a picture and then VoiceOver will speak to give camera positionning instructions: if the text is detected, how much you should eventually rotate your device to improve the picture, ….
To find this button: if I browse the elements on the camera screen using finger swipe from left to right: I have: Application Settings button, Flash button, Stabilization button, OCR button,Take picture button, Import Image button, and the last one: the scene description button (button hint is: Provides scene description and camera positionning instructions).
Hi, I have now found the button, thanks for telling me where to find it, the problem with it is that button is a serious mouthful, I mean given that we can look at the things like flash stabilization etc, do we really need that big long summary jammed into a button label, it was so long I didn't wait for the label to finish, why not just call it scene description or something? its two verbose in my view. also I have hints disabled as I am an experienced VoiceOver user so 9 times out of 10 for me they tell something very obvious I ready know, like for example double tap to open on anything that's why I don't bother with hints. The button does work just needs usability redesign. from having tested the button I think I prefer the way knfb reader gives its field of view report, much more straight forward for me to understand what I need to do to improve my accuracy, but even so I will try to test this app further, its always nice to have another tool available.
I am now just reading a review of this app done by one of the applevis editorial team, this looks very interesting, so if I understand it right, prizmo go does actually already have realtime OCR recognition? I must admit if this is correct I hadn't understood that from the description, I thought the point of cloud OCR was simply to take a still photo then go and process it for better text recognition I hadn't appreciated it worked in realtime is that the point of cloud OCR to work in realtime?
I already have a use case for this app which on its own is likely to earn it a permanent place on my Home screen.
If your Twitter timeline is anything like mine, there will be numerous times daily when somebody has attached a tweet as an image or included a screenshot of a web page: leaving you to guess at what that tweet or web page said.
Okay, okay ... there are already ways to grab the text from those attachments.
However, this app provides what (for me) is probably the most seamless, user-friendly and satisfying solution.
On my timeline in Twitteriffic, I went to an attached image; opened the native iOS share sheet; double-tapped on "Recognize Text"; and had that text read to me almost instantly.
As I said, it made for a very satisfying user experience, and made Twitter a little less frustrating.
So, a big thumbs up for the app from me.
What's the difference between this app and the regular Prizmo app? Having never tried Prizmo, I wonder which would be better to check out? Is Prizmo Go more VoiceOver friendly than the standard Prizmo? As others have said, I appreciate all the work that's been done concerning VoiceOver support. I'll go check it out.
well I tested the cloud based OCR and its actually as I imagined it would be, basically a still photo sent off to the cloud, will be interesting to compare it with knfb reader. One thing to note is that I have the trial pack, and I found that prizmo ate a few of my units even when they didn't have successful results, so that's a little worrying if I should decide to fork out cash for units. I will of course do more tests, need to find it something very challenging like a ready meal box, as knfb reader definitely doesn't do them well. But when the review talked about realtime OCR its not what I was hoping for when I read the term realtime OCR, I was hoping for something where it basically captures text from a video feed in realtime for processing, now I know that could get very expensive, but I would pay good money for that especially for things like navigating visual on screen menus etc.
Thank you all for feedback and the kind words.
About the scene description button: we are getting the feedback from this forum and from other users that it's too verbose. We wanted to have one place where all information would be summarized and let the user get shooting instruction (VO is supposed to move to that button initially). We will simplify the label in a future release to make it more straightforward as you suggest.
When VO is enabled, the app tries more orientations than normally and it should be fine in most situations, including some perspective distortion, as long as document and its text lines are seen in their entirety. Angles (gravity and left-right) are reported, as it can improve results a bit if user shoots again with better alignment. Another point of interest: clipped borders (clipped text lines) and associated instructions (move left/right) are reported only when angle is straight. This is because it is too tricky to report 2 things, angle and direction, at the same time. It could create confusion. I would suggest doing it in 2 steps: first align against angle, then against position.
About real-time OCR, no, Prizmo Go does not do that. It highlights the text while shooting, that's all. The challenge here would be the lower quality images with OCR being processed entirely on the device. We have experimented with that a bit in standard Prizmo, where resizing of text boxes executes OCR in a continuous fashion, but that is still based on a static image, and that is not the same use case that you describe here.
About Cloud OCR units: they are decremented when an image is successfully processed by the cloud servers, and result returned. Note that results can be empty if there is no textual contents in the image. That could explain what happened here. Note that the scene description button does *not* consume units, so it can be used as a first step to avoid wasting units.
About the Twitter image stream: yes, this is a use-case we wanted to have, that is, process image from other apps with Prizmo Go app extension.
Hi, OK so a few more questions about scene description, what should an optimum scene description sound like? for example with knfb reader its field of view report says all 4 edges visible, or the left edge, right edge and top edge visible, so how do we know what the best description is we can get.
also, what exactly does the gravity figure tell us? that really isn't obvious to me, as gravity is what keeps us on the ground haha so I couldn't begin to guess what it means in the context of a scene description. I have also done further tests with the scene description and it seems like it needs work doing to it. as an experiment I tried tapping the button when I had my phone pointing at the floor deliberately not at any text at all, and the scene description still reported it could see text haha.
I will agree with the commenter above on the recognize text feature added to the data sheet being an already has me sold feature. There was another app that could do this, but you had to open that app, import the photo, etc. Not a simple solution. This however, is right in the sharing options for easy use. I have not yet tested it for actual documents, but I am glad there are options out there, as not everyone can afford KNFB reader.
Will it read handwritten text?
I'm answering new questions above.
Angle with gravity means the angle between the gravity axis (vertical) and the camera optical axis (camera direction if you prefer). 0 means that they are aligned and that the iPhone is perfectly flat. This is important to get as close to 0 when you shoot a document that lays flat on a table for instance.
Another angle is reported, it's the angle of lines of text against the horizontal line within the image. 0 means lines of text appear horizontal in the image, and this is the case that gives the best OCR results. If that angle is small, it's not reported. Otherwise, the app tells to rotate 13 degrees right for instance.
The is no notion of document edges unlike other apps or even regular Prizmo. We are trying a new approach, only text matters here. Sometimes you have a large page with small text in the middle, and we want to focus on that text, not the whole page, to get best OCR results. If text coverage is too small, you should move closer. If text lines are clipped on edge of the image, it will tell to move that way.
About taking random pictures, well don't expect it to work, it's not meant for that. I think most apps don't work well when you provide unexpected intput.
It does not handle handwritten text. That'd be great to have that, but I don't think it's possible to have a robust recognizer at this time.
Hi. I tried this app on my iPhone 7 plus and it scanned my mail properly and I think it's good as the KNFB reader I would like to thank the author for developing a such a wonderful and reliable app?
Can anyone tell me if this app can read memes? Thanks in advance
Most text memes should be fine, and probably better with Cloud OCR, as they vary in size and resolution. People are reporting using it to convert Twitter image attachments that contain text into text. It can handle different contrast modes (bright text against dark background or vice-versa)
@Saqib: thanks, appreciated!
So, I have used the KNFB reader, Fine scanner pro, Identifi and now Prismo Go. Honestly, I'm very impressed with Go. I haven't used any of the premium features like cloud recognition, but I probably will. Even the basic recognition seems to be more accurate and user friendly than any of the others. I primarily use OCR apps like this to read printed documents at work, and especially to read big blocks of text on my 32 inch LCD screen when gaming. This last has proven very difficult with nearly all apps except Identify, and now Prismo. I prefer prismo because it is faster and easier to use, and it also appears to be more accurate. I also appreciate its landscape capabilities. All in all I have good things to say about this app so far.
Prizmo Go 1.0.1 is now available on the App Store. It brings some VoiceOver enhancements:
- The scene description button now has a short label for easier discovery.
- The volume down (physical) button now triggers scene description while volume up button triggers shooting.
- Added support for VoiceOver two-finger scrub gesture (Z shaped gesture) to go back to the camera screen.
hi all, i am wondering as to why i need to buy the export pack, just to get 100 or 1000 cloud-based units? I do not need the export pack, for my use cases, at least, not yet.thank you, Will
We assume that if a user purchases cloud OCR units, he also expects to be able to export the recognized text. That's the reason of this requirement.
More generally, you can consider the Export pack In-app purchase as a way to try the app and then purchasing it if you like it.
Now and again, an app is developed that has what I will call the wow factor. In my estimation Prizmo Go is one such app. I have used it extensively both the cloud and built in OCR and have received accurate results on a variety of document types.
Coupled with the fact that the basic app is free you cannot go wrong in my view
I was wondering when using this app it might report something like 23 lines of text visible. So, would this mean that the more lines visible, the better the scanned result should be? Thanks for this very nice app.
I have been using this app for only a couple of days now, and so far I am really impressed with it. I went ahead and purchased the five dollars export pack, which is a reasonable price considering its a one time only purchase, and costs about as much as a cup of coffee from Starbucks. Lol. Anyway, I can't say enough about this app. It's so wonderful to have a scanner in my pocket. The results with the built in OCR recognition is very good, with the cloud recognition being very good as well, and for 100 units for a dollar or 1000 for five dollars, the price can't be beat. I've used a flatbed scanner and the kurzweil 1000 software for windows with just as good, if not sometimes worse, results than this app. While KNFB reader may be the app many prefer, this app comes in a close second, especially for folks like myself who can't afford the 100 dollars price tag for the KNFB reader. I hope this app continues to grow and become even more useful. I highly applaud the developers for making voiceover accessibility a built in part of this app. Congratulations on a job well done. I've scanned several document types, and even some food boxes with favorable results. Once again, thank you for the hard work that was evidently put into this app.
Hi, what is the diference between the Prizmo app and Prizmo go?
I'm answering the previous questions:
Jan: it is just a description of what is on the picture, within the field of view of the camera. If you expect lots of text but are only getting a few lines, it lets you know something may be wrong in the capture. Also, if Prizmo Go finds that lines of text are cropped by the picture (partially outside of the camera field of view), it willl let you know with a "move towards document left border"
Matt: thank you for the nice words, appreciated!
Ulises: Prizmo Go is meant to quickly capture some text and use that text or play it on the loudspeaker. Prizmo on the other hand is meant to capture and produce multi-page PDF documents. Prizmo also comes with text-to-speech capabilities, including high quality voices option. Prizmo has been there for a long time and has more power features (including iCloud sync'ed documents with the Mac version of Prizmo), but in it is more complex and in its current state, does not offer Cloud OCR option.
hi does either app's ocr feature handle scanned mS word Images in a multi page PDF doc?
Some people send me these and I am looking for an app to convert it back into something readable By voice over.
Does Prizmo do it?