Dear users and system administrators, I Am reading Yours discussion related to performing reliable OCR from screen. The existing solution, which can not be universal but can help is to use special NVDA addon, which is using Tesseract OCR engine. It is necessary to use NVDA object navigation keys to point The object pointer to The object, which You would want to perform OCR on it. Press INSERT+R. You must firstly install this addon. http://files.nvaccess.org/nvda-addons/ocr_0.20120529.01.nvda-addon The ideal solution would be, if somewhere would be AN universal video driver, which would be able to force all available video bioses on The world to use The resolution 300 X 300 pixels, but this is very probably not real and Windows GUI would start to be unusable, because build in Mathematical formulas would stop to enable users to click on somethink reliably. But fortunately, it is still possible to switch The on screen resolution to The 640 x480 points, 8 Bits colour. When Window is maximized, user can really read some texts from The screen. If somebody will make tesseract for some mobile phone platform, it would be may be universal OCR on screen solution. The instructions how to use NVDA addons are here. http://www.nvda-project.org/wiki/AddOns The ideal solution would be, if somebody, who is deeply familiar with mathematic formulas and algorithms would be able to develope new revolutionar algorithm, which would be able to convert image with high resolution to 300 DPI resolution and only after this image processing, this image would be send to Tesseract OCR or other OCR engine. By The way, does somebody of us know free image editor, which would convert image file with high resolution to user specified lov resolution, not only on The screen inside working graphical editor, but The changes would really had to build in to The resulted image file. I do not know this tool, and it is very sad.