Which image type is best for Android or iPhone?

When capturing images on an android or iphone that have text on them... I've been saving them as Text Images... does it matter on a mobile device or would Tolerant be better?

Thanks

Comments

  • EggplantMattEggplantMatt ForumAdmin admin
    Text should be fine unless you're seeing false positives in your image matches. The Text search is a little bit slower than a Tolerant search, but with screens the size of a smart phone's, the difference is going to be negligible. And if there is any dynamic antialiasing occurring, you're covered.

    BTW, we have released eggPlant v11 with true OCR capabilities. You may find that helps in situations where it's more difficult to script with images. It should enable you to script against the some of the same content on android and iphone without the need for collections or other workarounds.
  • I musta missed something. From what I read about OCR I can use it to read text and put that input into a variable or store it. How do you use OCR to Click on an image?
  • Some examples would be good for me...

    Thanks
  • EggplantMattEggplantMatt ForumAdmin admin
    You can set "Generic OCR" to be the default Text Platform (if you have not previously modified your Text preferences, this will be the new default setting) and use it just like the TIG, except without the need to worry about any text characteristics, so it is just used like this:
    Click (text:"192.168.1.12")
    
    or if OCR is not the default text platform:
    Click (text:"192.168.1.12",TextPlatform:"Generic OCR")
    

    The OCR search can be a little slow if it's not constrained, but again, on a screen the size of a smart phone's, it should be pretty quick. If you're running against a desktop app, it's better if you can narrow the search area. We've made this easier in eggPlant v11 by allowing you to simply specify two images representing opposite corners of a rectangle:
    put readText(("image1", "image2")) -- works in current v11 release
    click(text:"Rate", rectangle:("image1", "image2")) -- will be working in forthcoming v11.02 release
    

    The readText() function will also take a single point or an image representing a single point as an argument and will try to read text in a horizontal region around that point:
    put readText("lastNameField")
    
  • Hey, I finally got around to trying this on an iPhone. It works!

    I set Text Engine to OCR search and it found some text on a safari page but not text in an image on that page.

    Should I try to add a Platform or leave well enough alone? I'm not really sure where to start with Platforms... back to the Manual!

    Unless you have some handy tips?

    Thanks
  • EggplantMattEggplantMatt ForumAdmin admin
    At this point, it would be good to take this discussion to e-mail and send us some screenshots of the text that you're trying to identify. Bear in mind that if you're trying to identify an image with text in it, and that text isn't being generated dynamically, then you should probably just capture an image. There's not much reason to use OCR to read a known, fixed value or to use it to find that value to click on, unless either its content or its appearance changes dynamically. We like to say that OCR is a tool that should be used judiciously -- it's not meant to be the new way to do everything with eggPlant.
Sign In or Register to comment.