Hi, everyone. Attached are the Word and text versions of the chat session on getting good scans. Linda Adams
June 12, 2007 GETTING GOOD SCANS ** Best Scanning Results for Blind People Wanting to Scan Use either OpenBook scanning software from Freedom Scientific (www.freedomscientific.com) or Kurzweil 1000. Most people prefer Kurzweil over OpenBook because it gives more flexibility in settings and editing features. Kurzweil has been upgraded more often then OpenBook. These two programs are especially designed for blind users. ** Types of Books, Their Good Points, and Their Problems 1. Try to get hard cover books or the nicer trade paperbacks. Try not to get the ones that say Mass Market. The hard cover and trade paperback books have nicer print. Paperbacks are easier in terms of the spines. 2. If you get the Mass Market books, they work if you do a bit of preparation work with your scanning program of choice. Increase the resolution or dots per inch. Try 400 dots per inch instead of the default setting of 300 dots per inch. Experiment with different reader engines if you have multiple choices. There are different options in Kurzweil and OpenBook. The scans will tend to come out pretty well, but Mass Market books tend to yellow more quickly and deteriorate more quickly because they are not as high quality, but you can usually get a good scan out of those. Grayscale gives you a better scan with Mass Market paperbacks or when the print is of poor quality. The 400 resolution setting is good for books with faded ink. 3. Photocopied Material: Usually photocopied material is of poor quality, and if you need this material for your job or school, you will need a sighted person to look it over to check its accuracy. ** Scanning Steps 1. Get a dry cloth and lightly dust the scanning surface off. If you determine that the scanner glass is dirty enough to need more than a dry cloth, use plain water. Use a cloth made of the material of a cloth diaper. It doesn't have lint or a muslin towel that doesn't shed. You could also use a microfilament cloth for cleaning the scanner. These are sold at computer stores. Preferably do not use any solution. Try not to use even water because the glass in your scanner is in layers, not solid, and water or solution can get stuck between the layers. A lint-free cloth will remove ink from the glass. 2. Open the book and fan the pages out. Run your fingers along the spine on the inside of the book to limber it up, especially for paperbacks, so that you can lay the book flat. This is especially important if you are going to do two-page scanning because you have to be able to get all of the text close to the spine scanned and recognized. If you own the book, don't be afraid to work the spine over until it lets you flatten the book completely. Bend the spine back all the way. Of course if the book is borrowed, you have to be more gentle. 3. Brush the pages off with your fingers for Mass Market paperbacks to get the ink dust off them so that this dust will not coat the scanning surface. This will give you a better scan. 4. Press the book down flat on the scanning surface. If it has some give or bounces back up away from the glass when you release it, hold the spine down as you scan to make sure that there is no space between the book and the scanning surface. 5. Open to the center of the book, and if you have Kurzweil, use the optimize setting for the scan for two or three pages of the book to see if any setting adjustments need to be made. Kurzweil does a fairly good job picking the optimal settings to scan a particular book unless the print quality is exceptionally bad. Once optimization is complete, go through maybe five or six scans to see if those particular settings are worth keeping because sometimes you may find that you are on a page that is particularly bad, you are scanning a page with a table, or you are scanning a page with a photograph. Rely on results from five or six different pages or scans to be sure that your scans and results are consistent. 6. Tinker with the settings such as brightness and resolution until you get what you feel is the best scan that you are going to get. 7. For Kurzweil users, Grayscale is the best thing to try when optimization did not produce the quality that you wanted. Grayscale will make your scans slower, though. Grayscale gives the best page representation as opposed to dynamic thresholding or automatic contrast. 8. There is no magic setting that works for everything. You can start in Kurzweil with the default settings and see if they work. If that doesn't work, optimize. If that doesn't work, change some of the settings yourself. 9. In Kurzweil, if the statistics say 95 percent confidence level or less, rescan the page to try for a better scan. Otherwise, you will have to struggle with many errors on the page. 10. Do ranked spelling or spell check. ** Scanner Settings: OpenBook Paperbacks 1. Turn off the despeckle feature. 2. Turn off the Language Analyst feature. 3. Turn off light text on black background. ** OpenBook Settings For decreasing quality with successive scans, hold the pages down firmly, especially the spine. Your hands and fingers tend to get tired, and you tend to lose the grip on the pages. Sometimes the pages move. Once you have optimized those settings, things should be the same unless the print quality changes. You don't need to optimize every two pages. You will get the best scans for the widest variety of material if you use the default setting in OpenBook, which is Scan for Accuracy in the Windows menu. If you are using the classic menus, it is the Automatic Contrast setting. This setting is equivalent to the Kurzweil's dynamic thresholding. For some material, you will get better results if you use Custom Scan. Experiment in that setting. Most of the time with Custom Scan, the results will be worse or no better, so use the default settings and then only experiment with them if they don't work. Another reason to use the default settings is that if you use Custom Scan and the print changes, if the contrast changes in the book, then it may not match the custom settings that you scanned. Using Auto Contrast will automatically compensate for scanning differences in the text, and it will either eliminate or cut errors way down. If you scan two pages at once, use the two-page scanning setting so that each page will be recognized as a separate page. Using the two-page setting keeps material from one page being mixed with material from another page. If lumps start to form as you are scanning and accumulating pages, press down and in the direction of where the lump is, either on the left or the right. You may also need to pick up the book and smooth out the pages so that no creases start to form. These two problems will progressively decrease the quality of your scans. ** Kurzweil: Trade Paperbacks 1. Turn off speckle removal. 2. Many times, especially in a book series, you can optimize a paperback and keep those settings for other paperbacks. ** Kurzweil: Hard Cover Books 1. You can use the speckle removal more for hard cover books, or use it for thick pages. Speckles go behind the text as decorations. These make books and papers more distinctive and attractive. If you have the setting on, it confuses the OCR engine many times. Sometimes when it removes speckles, it can also remove text that you want to be left in place, so try to use this setting sparingly. 2. Whenever you get a new hard cover book, you will probably have to optimize the scanning; no two hard cover books seem to take the same settings. ** Kurzweil Settings In Kurzweil versions before Version 11, Kurzweil's default threshold setting is static thresholding. In Version 11, the default setting is dynamic thresholding. ** Settings for Mass Market Paperbacks 1. Use Grayscale instead of the default Dynamic threshold setting. 2. In terms of brightness, use settings between 60 and 70. The higher in that category, the better. 3. For most paperbacks, you can scan two pages at a time, so set the scanner to recognize two pages per scan. 4. For paperbacks, most of the time, you will not need to turn on the column setting. It is helpful to have columnization off for books that don't have any columns so that you don't get into trouble with tables not being recognized. 5. You can set your front margin for 4.7 in two-page scanning to make the scanning go faster. 6. You can sometimes remove the lid from the scanner and then just lay it down on the scanner over something thick. 7. Optimize the beginning of the book where the title page and copyright information is, then change your optimization for the rest of the book as these settings will usually be different. To keep from having to do two optimizations, however, in most cases, you can choose Grayscale for title and copyright pages, and they will come out fine. For Bookshare, the preliminary pages are not as important; just be sure that the copyright information is correct. ** Fine Reader, OmniPage, and Premiere Systems For people who cannot afford OpenBook or Kurzweil, the plain commercial version of FineReader works with JAWS. You can use FineReader 7 and 8 with JAWS. You can use OmniPage 9, 10, and 12. Obviously there is a learning curve, and you get a little spoiled with the blind friendly features of both OpenBook and Kurzweil, but if money is an issue, FineReader is certainly an excellent choice. It is the engine that most of us use in Kurzweil and OpenBook. The FineReader people have done a lot to make their software very flexible and versatile. You can have it scan and put your books directly into Microsoft Word. It speaks well. The latest version of OmniPage is 15, and the latest version of FineReader is Version 8. FineReader's interface with screenreaders is a little more simple to navigate, whether you use JAWS, Window Eyes, System Access in VDA or Thunder. OmniPage also has a decent interface. There are a few tweaks and options that you need to remember. Pratik Patel will tell you about these if you are interested. When it comes to settings, they are fairly similar to what you might expect in Kurzweil. Only a few of Kurzweil's settings are proprietary. Most of them rely on the OCR (optical character recognition) engines provide. The settings may be called something slightly different; for example, Automatic Contrast and Dynamic Thresholding, but they mean very much the same thing. Both of the products have very good documentation. These will teach you a lot about their settings. Premiere Text Programming Cloner is very accessible and affordable, from $49 to $149. It is a very simple program, but the results that you get are very inferior to any of the other programs that it is not recommended for scanning books for Bookshare. They have upgraded with a few different pieces of scanning software that could be tested for quality. They have demos, so try them out yourself to see what you think. Premiere Assistive Technology's products are mainly designed for those who are learning disabled. There are a few products from this company designed for blind people that are accessible to screenreaders, but their main focus is on learning disabilities. There is heavy focus on tracking, lighting, thesaurus, so you may find that a lot of scanning features do not necessarily work as well for blind and visually impaired people. ** Optic Book 3600 It is a book-edge scanner. It is fantastic for books that have large pages and you are not able to place two pages at a time on your scanner. It is good for books that you are getting from a library and you don't want to break the binding. It is the fastest scanner seen within the price range of between $200 and $250. The software is not really accessible. Unfortunately you have to install the Optic Book software in order to operate it. Kurzweil will bypass that software in order to access Optic Book. There are optimal settings that you can use in Kurzweil with Optic Book such as the WIA drivers that are Twain that will make your experience even better. The first time when you set it up, you will need a little sighted assistance if you want to set the lamp to go off after 15 minutes. The usual setting is for the lamp to go off after one minute. You will definitely want that setting changed because one big down side to this particular scanner is that once the lamp goes off, if you want to warm it up, it takes about 30 seconds to warm up. That is very important. With the 15-minute setting, it will go off after 15 minutes of inactivity. You get a good scan within about five to six seconds. Get the latest drivers from their site because they install easier. The latest ones are dated February of 2007. You want the 3600 scanner. You don't want the Plus. The other scanners have software that we can't use and don't need. Plug the scanner in first. XP will come up with the New Hardware Wizard, and I pointed it at the driver folder, and away it went. The Twain settings will work, but use the WIA settings because you can't pause the scanner with the Twain. It gives better results through Kurzweil than Epson. This scanner was specifically made for scanning books, not for graphics but for text. Larry Lumpkin has a set of instructions that he got from Nick Dotson at Kurzweil. Go into Windows Explorer, find Kurzweil Educational Systems, then find Diacs, and then there is a program in there called ScanConf. You choose the scanner. It has settings regarding document feeder, duplex, holding the Twain source between scans (set that to never). The scanners are either in their Twain version or their WIA version, and you can set the settings for the WIA version. Within Kurzweil itself, you go to the Scanner Settings and choose Optic Book 3600 WIA. The scanner itself does not come with a document feeder. This scanner should work with any version of Kurzweil back to at least version 8. Even version 7 works. Economically, you should upgrade to the latest version of Kurzweil and keep using your old scanner before you try to purchase this scanner. The only drawback to this scanner is that it takes about a minute to warm up when you turn it on, and it doesn't scan pictures or other things. Grayscale is noticeably slower with the Optic Book. If you use Grayscale, don't keep images in your files. Do not enable the Keep Images in Recognized file. ** Getting Your Books Approved and Not Rejected 1. Periodically check for scanning clarity. Every 15 or 20 pages, look at the last page of your file and see if the settings are still producing accurate results. Is the material still clear? When you are finished scanning the book, page down through all the pages and make sure that you have all of the pages in the book. 2. Prevention of Two Pages Sticking Together: Sometimes it is easy for two pages to stick together and you skip one by accident. Pay attention to the thickness of each page as you are scanning a specific book. If pages stick together, they may likely feel more thick, a little more rigid, or they may feel different in some way. As you turn the page, rub your finger and thumb of one hand on opposite sides of the page as you hold it so that if two pages are stuck together, you will increase the chance that they will separate. Read at the top of each page to see if the scanner picked up the number on that page. If the scanner is reading your book's page numbers consistently, you will know whether you skipped a page that way. Some page numbers are at the bottom of books, so check the top and bottom of every page for these page numbers. If you are using OpenBook, use the setting Scan and Read; don't scan ahead so that you can keep track of page numbers. If no page numbers in your book are announced, you can scan ahead because you won't hear page numbers anyway. In Kurzweil, there is an Operator Page setting to tell Kurzweil what page you are on, so when you find the page in your book that begins content, you can tell Kurzweil that that is page one. Use the setting that says Keep Blank Pages. This is especially important for books that have blank pages between chapters. This also helps you to keep track of page numbers. This helps Bookshare as well with page numbers. 3. Check for the reason that a page has garbled text. Maybe one of the pages becomes very garbled so that validators have to find book submitters to get the original page that goes in the book. If the validators or Bookshare staff cannot locate the submitter, Bookshare has to reject the book. If a page has gibberish, rescan it to see if it can be better or whether the garbled material is a picture. Once you get a good scan from many pages of your book, save those settings in Kurzweil for that book. ** Learn by Experimenting Everyone's opinion is different. Try the suggestions here and see what works best for you. By being wiling to experiment, you will develop a system of your own that will be efficient, accurate, and save you many hours of work and frustration in your scanning.