========================================================================= (C) 1993 by Atari Corporation, GEnie, and the Atari Roundtables. May be reprinted only with this notice intact. The Atari Roundtables on GEnie are *official* information services of Atari Corporation. To sign up for GEnie service, call (with modem) 800-638-8369. Upon connection type HHH (RETURN after that). Wait for the U#= prompt.Type XTX99437,GENIE and press [RETURN]. The system will prompt you for your information. ========================================================================== ************ Topic 10 Sat Dec 14, 1991 SANDY.W [RT SysOp] at 17:25 EST Sub: OCR from Migraph Optical Character Recognition from Migraph. 201 message(s) total. ************ ------------ Category 5, Topic 10 Message 1 Mon Apr 20, 1992 J.MEEHAN3 [>> Joe M << ] at 06:09 EDT All, It just hit me! Some times it takes a while. Some users are reporting recognization rates of 30 % and others 90%. I think we are looking at different things. I report a higher rate. My observations are based on what I get when I leave OCR. I chatch most of the errors in WordPerfect using the spell check. Others seem to be using the last % readout of the Character recoginizatin phase of the OCR processing. I really don't thing this is a good %. OCR is not done at that time, there is more processing to do. Yet others may be useing the % of recoginzation using Automatic OCR or pre-Intereactive learning stage. This is a good number (Althogh I don't think OCR computed it for you.). If you are not useing interactive learning you are loosing one of OCR's best most powerful features. >> Joe Meehan << ------------ Category 5, Topic 10 Message 2 Wed Apr 22, 1992 G.FUHRMAN [gnox] at 21:29 EDT Joe M, I suspect you're right about the recognition rates reported. The figures I've reported are always estimated, because the percentage readout that shows during OCR processing goes by so fast on the TT that it's almost impossible to read, let alone remember! Anyway I haven't noticed that it has much correlation to the accuracy of the output text at the end of the process. gnox ------------ Category 5, Topic 10 Message 4 Fri May 01, 1992 H.WOLFE1 at 00:40 EDT Joe and gnox, thanks for the information. Harvey ------------ Category 5, Topic 10 Message 5 Sat May 02, 1992 J.E.DONOHUE at 20:23 EDT Need Help - I purchased this package for a 300 page (8.5 x 11 - 2 column) book I contracted. Hardware in use is: Mega STe 4, Canon flatbed scanner with Navarone interface. Version 1.03 (shipped to me) gave me a FEW problems and Migraph Fed Ex'd 1.05 to me (nice - thanks folks). I've scanned and saved .IMG files for about 1/4 of the book and started to create ASCII files from the scans. I've built up a pretty good dictionary but still only get about a 40% character recognition rate. (That's not the problem tho .... I was zipping along thru 26 pages with no problems - and then started getting a 2 bomb crash, back to the desktop. I am appending to the existing dictionary, TEXT I have set for small (it's about 9-10 point Times Roman), PROPORTIONAL is checked, INTERACTIVE LEARNING is checked, PARAGRAPH is checked. I import the image ... clip the area I want to process ... the program goes thru the AUTO LEARNING PHASE ... gets to about 47% ... and 2 bombs - back to the desktop. Any ideas? As I said, the first 26 pages worked fine. I've tried to cold boot (unsuccessfully), I've tried to reinstall the program (on the same partition and a different partition). Help. john ------------ Category 5, Topic 10 Message 6 Sun May 03, 1992 G.FUHRMAN [gnox] at 08:22 EDT I suspect that problems like this result from the dictionary getting corrupted. I have no idea what we can do about it though - until Migraph gives us some way to edit the dictionaries. Wish I could help more! gnox ------------ Category 5, Topic 10 Message 7 Sun May 03, 1992 J.E.DONOHUE at 14:17 EDT I "suspected" a corrupt dictionary and that will be my next move. Sure would like to know (if this turns out to be the problem) how it gets "corrupted" and a fix would be nice. %^) john ------------ Category 5, Topic 10 Message 8 Sun May 03, 1992 J.MEEHAN3 [>> Joe M << ] at 14:32 EDT John, First if you are only getting about 40% character recognition, that would mean you are re-typing 60% of the document. I would guess you are looking at the read during the processing phase and that does not represent the final recognition rate. As for the crashing I also feel it is the dictionary. After you reach the plateau where you are not seem improvment in real recognition rates, you should stop the learning phase and use read only for the dictionary. The dictionary seems to have a max size and it also seems to get corupted easy. >> Joe Meehan << ------------ Category 5, Topic 10 Message 9 Sun May 03, 1992 J.E.DONOHUE at 16:40 EDT Slowly losing hope with this program folks. I deleted my dictionary -- started a NEW one and successfully converted one (1) page. Any other page I try to convert I either get 2 bombs and back to the desktop, or "Error! Ran Out Of Memory During OCR" . I checked (with Maxifile) my memory AFTER my .IMG file was loaded in and cliped and had 34k of RAM left. I then deleted EVERY Auto .PRG and .ACC .... same results. Bottom line is I can only convert an .IMG clip with a NEW dictionary. Oh ... even tried to APPEND .. READ, etc. Like I said ... slowly losing hope. Any ideas? john ------------ Category 5, Topic 10 Message 10 Mon May 04, 1992 J.MEEHAN3 [>> Joe M << ] at 05:55 EDT John, What version are you using? >> Joe Meehan << ------------ Category 5, Topic 10 Message 11 Mon May 04, 1992 J.E.DONOHUE at 07:44 EDT Joe ... version 1.05. I have now reached a point with this program where the ONLY way I can process a clip is to start a NEW dictionary. Anything else and I either get an out of memory error (Mega STe 4 .. and I thot the cache was supposed to handle that?), or 2 bombs and back to the desktop. What baffles me is I USED (quite successfully) the program for almost 4 days -- converted approx. 24 - 2 column pages (46-48 clips) with NO PROBLEMS. As I sit this morning ... the program is unuseable. Go figure john ------------ Category 5, Topic 10 Message 12 Mon May 04, 1992 K.VANDELLEN [Ken Van] at 18:47 EDT John, You seem to be having the same kind of problem I've had. It works fine for a while, and then starts bombing. Later versions are faster and more stable, but I'm still getting bombs. They've started, recently, after as few as 1 or 2 passes, but I got 6 or 8 in once. I think I had switched to read by then, but maybe not. I've gotten the memory error often, too, and that should give Migraph a clue. What's using up the memory on a Mega STe4? I get 40% recognition on the processing graph, but someone here says there's another indicator of that, and it gives a higher percentage. I forgot where he said it was, but it made sense at the time. (I guess it must have been Joe.) I'm sure the problems will eventually be solved. I hope soon. As gnox says, the dictionaries seem to be getting corrupted somehow. I wonder if it has something to do with too many scanned characters that may be significantly different being identified as the same "real" character. That is, the program is being told that this, that, and the other thing are all "P", until it can't handle it. Maybe I getting too anthropomorphic or whatever. Whatever it is, it's a pain to have to start the training process (dictionary-building all over again. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 13 Tue May 05, 1992 J.MEEHAN3 [>> Joe M << ] at 06:02 EDT John, You might try reloading your program, it might have gotten corupted. I am not sure, but if I remember right I got two version 1.05's. Maybe they were the same, I still have the disk, I will check see if they were the same or if there were really two versions. Good luck, >> Joe Meehan << ------------ Category 5, Topic 10 Message 14 Tue May 05, 1992 J.E.DONOHUE at 08:00 EDT Thanks for the suggestion Joe .. I DID reinstall three times (different partitions, etc.) .. same results. After a couple of FAX's to Kevin and a phone call .. here's what we came up with which seems to have cured all: (BIG knock on BIG piece of wood) The version originally shipped to me was 1.03 which was (from reading this topic - "buggy". I had the folks at Migraph Fed Ex the update to 1.05 which they most graciously did. Seems when I was upgrading (running the install.prg from the upgrade disk) I would go into the sub-directory where the MI_OCR.PRG was and "install". Kevin said "nope" -- that IS the logical way but all you need to do is go to the partition where the .PRG resides (E:\ whatever) and install. If anyone installed from 1.03 to 1.05 and hasn't had any problems .. well, I guess that's another mystery. Another "thot" ... If/when a dictionary gets corrupted - DON'T leave it on the same path (or partition for that matter) as the NEW dictionary you start. In other words get it as far away from your computer as possible! (grin> Hope I'm not letting out any "rumors", but Kevin told me (Joe Customer) so I guess there are no secrets ... a version 1.09 is VERY close which handles the "large" and corrupted dictionary problems. Also in the works this week is an update to the Canon Accessory for Touch-Up that will allow you to scan directly into the OCR package with the Canon/Navarone interface. Nice folks at Migraph ... Thanks! john ------------ Category 5, Topic 10 Message 16 Wed May 06, 1992 C.MONTGOMERY at 08:10 EDT Will they notify us here as to how we can get the new Canon/Navarone accessory when it's ready, or will we have to wait for USsnail? Charlie @ Design for Print ------------ Category 5, Topic 10 Message 17 Wed May 06, 1992 J.E.DONOHUE at 21:36 EDT Good question Charlie .... If you're a registered user (raising hand) they MIGHT notify you. (heck if I know). Anyway .. might be worth a phone call to them and "get in line" or let them know you're interested in the .acc when it's available. (soon I hope) john ------------ Category 5, Topic 10 Message 19 Thu May 07, 1992 C.MONTGOMERY at 08:15 EDT Good idea John. I'm a registered user, but I'll give 'em a call anyway. Just to let the know I'm paying attention. ;-) Charlie @ Design for Print ------------ Category 5, Topic 10 Message 20 Fri May 08, 1992 K.VANDELLEN [Ken Van] at 18:06 EDT I get the impression that Migraph doesn't check in here, and if we want to communicate with them we need to do it directly. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 21 Sat May 09, 1992 B.MENAGH [Bill] at 07:53 EDT On occasion when the processing is taking place quite OK, all of a sudden two lines of text appear in the #2 window (Manual: p12) with a number of characters highlighted. Can anyone explain what is happening and what a solution might be? Bill ------------ Category 5, Topic 10 Message 22 Sat May 09, 1992 J.MEEHAN3 [>> Joe M << ] at 09:16 EDT Migraph does check in here; but only on occasion. If you want to contact them it would be wise to do so by phone. I have found them responsive. >> Joe Meehan << ------------ Category 5, Topic 10 Message 23 Sun May 10, 1992 CBARRON at 02:06 EDT Bill If you mean the text in the interactive dialog box shows two lines of text (The graphic image in the middle.) I ussually find that when I have a poor scan and the point size of the characters drastically changes. If this is a poor scan rescan it. If it is because of 'fine print' Ocr all before the fine print. Ocr the fine print and ocr the rest in pieces so as to retain relative position of the text. Reording a bunch of rectangles may do it all in one OCR pass but I generally find when two lines are treated as one there is a problem with the scanned image. ------------ Category 5, Topic 10 Message 24 Sun May 10, 1992 J.HAYDEN [JOHN] at 02:12 EDT Bill, It sounds like a stacking problem. Version 1.05 has reduced this kind of problem. It is caused by not enough leading space between lines and also be caused by oversized letters. Sometimes fancy fonts dip too far below the line. ------------ Category 5, Topic 10 Message 25 Sun May 10, 1992 B.MENAGH [Bill] at 09:17 EDT Recently, I was using version 1.05 and started receiving two bombs during the processing of the scanned text. I have reverted to V1.03 and have had no difficulties. In fact, I had no difficulties with the earlier versions and was using V1.05 successfully until this latest difficulty. I have not changed my setup except with the addition of WARP 9 and MULTIDESK. Is there something I should be aware of? Any assistance appreciated. Bill. ------------ Category 5, Topic 10 Message 26 Thu May 14, 1992 P.MORALES1 [Pat] at 21:34 EDT In the last few days, I've been trading faxes with Kevin Mitchell of Migraph, Inc. I've used the program a lot and experienced some of the same problems reported here resulting in my correspondence. Kevin convinced me that they are close and with the next release will complete the bug fixes and add several enhancements to boot. Following in the final fax in its entirety, courtesy of MIOCR: > Dear Pat, > Thank you for your fax. > Corrupted Dictionaries: > 32k Dictionaries are probably ones that have already been > truncated. Usually beginning dictionaries will quickly become 40k > as OCR builds an initial dictionary from it's built-in training. > Low Memory/Dictionary problem > If there are any 2-3mb users on GEnie having this problem, > would you let them know to free up as much memory as possible. If > they are still having real problems they can call and request a > v1.09. > OCR Enhancements...... > If you would like to spread the word > that there are enhancements in the works for v1.1 on GEnie > that would be appreciated. We will be formally announcing a > complete breakdown in our next newsletter which is being > mailed the first week of June to all registered Migraph > owners. > Misc,,.. > There will be other Major NEW ST product announcements made > at the same time as the newsletter. (Thought I, drop the > teaser early.) > Sincerely, > Kevin Mitchell > Migraph, Inc. So enough of the negatism. Migraph DOES still care! Pat ------------ Category 5, Topic 10 Message 27 Fri May 15, 1992 G.FUHRMAN [gnox] at 06:49 EDT Migraph is announcing a complete breakdown, eh? Shucks, and they were doing so well! :-) Thanks, Pat. gnox ------------ Category 5, Topic 10 Message 28 Sat May 16, 1992 K.VANDELLEN [Ken Van] at 11:01 EDT Thanks, Pat. I can't wait for the new version. I wonder if they're going to have a TouchUp with faster load/save. That's my main dislike about it. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 29 Tue May 19, 1992 K.VANDELLEN [Ken Van] at 18:21 EDT I just talked with Migraph. My version is 1.05, and I've been having intermittent problems. They are sending me v. 1.09, and will do so for anyone who is having problems. They tell me that some people are not having problems with 1.05. Version 1.09 is stable. There may be another version a bit later. They are doing a v. 1.1 for the Amiga. There is a question whether that will be free and automatic or if there will be a charge, when it comes out for the ST line. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 30 Sat May 23, 1992 J.E.DONOHUE at 14:19 EDT Got version 1.09 today in the mail. Haven't had a chance to "play" with it yet, but it's supposed to handle memory management problems and help out with large dictionaries that became corrupt. Another feature for anyone owning a Canon flatbed or sheetfed scanner AND has the .acc for Touch-Up. If you have version 1.5 of this .acc you can now scan directly into the OCR package. Won't have to create an .IMG file then load THAT into the OCR package. My version is 1.0 (never got any kind of an upgrade notice, etc.) so I guess I'll have to wait a week until I can get an upgrade to test this feature out. john ------------ Category 5, Topic 10 Message 31 Sun May 24, 1992 K.VANDELLEN [Ken Van] at 14:07 EDT I got v.1.09 yesterday, too. Haven't tried it yet, either. MiGraph told me it was a "stable" version. Good! Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 32 Mon May 25, 1992 CBARRON at 05:13 EDT It is faster, and the function keys in the interactive dialog help by keeping my hand away from the mouse while typing. Have not tried it too much yet, but it is better than 1.03, ------------ Category 5, Topic 10 Message 33 Thu Jun 04, 1992 K.VANDELLEN [Ken Van] at 18:54 EDT Version 1.09 is very stable. No more bombs and lockups!! I had good results using it yesterday, but today a lot of words and groups of words were repeated. Kevin, the main guy at Migraph, suggested disabling Abbreviator and Data Diet, which didn't help. On the next call he said to try using line mode, and it worked VERY well. When I called to report that, I got Paul, the author, who wants me to disable most of my AUTO programs and DAs as a test. (Isn't it neat in the Atari world that we get to talk directly with these people?) I'm only using Abbreviator and MultiDesk Deluxe as DAs. (I'll have to ask the CodeHeads if anything _within_ MDD can possibly cause interference in a program. I didn't think that did anything but allocate memory for when you call up a resident DA.) I am using a few AUTO programs, but I'm not sure why any of them would start causing a problem out of the blue. Maybe the Mega STE needs a diagnostic test. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 34 Fri Jun 05, 1992 K.VANDELLEN [Ken Van] at 06:43 EDT I'm not sure which topic this fits under, so I'll leave it in a couple of likely spots. If I use a flatbed scanner on an IBM or a Mac, is there a way to get the file into IMG format, so I can use the Migraph OCR to do its thing? I have too many pages to scan easily with a hand scanner, but I will if I have to. As a last resort, I could see if I could scan at a dealer, but I'm afraid it would cost a fortune. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 35 Fri Jun 05, 1992 ISD [Nathan] at 08:44 EDT Digital Research's GEM file format is well-known on the DOS platform. You should not have a problem in this regard. ------------ Category 5, Topic 10 Message 36 Fri Jun 05, 1992 K.VANDELLEN [Ken Van] at 18:48 EDT Nathan, Were you addressing me in msg 35? If so, does the software for the IBM scanners give a choice of format, such as GEM? Can GEM be converted to IMG? Am I butting into some else's conversation? Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 37 Sat Jun 06, 1992 CBARRON at 05:25 EDT What file formats do you want to convert to IMG, for starters?? Depending on the original file format and the software at hand a conversion to IMG might be possible. GEM IMG files are possible no PC's but much software use TIF, or PCX. What file format the file is saved in depends on the Mac/PC software itself. ------------ Category 5, Topic 10 Message 38 Sat Jun 06, 1992 M.DRYSDALE [Drys] at 08:11 EDT Touch-Up and/or MVG will load a variety of file formats and save as .IMG (and many more). There are also programs that convert from one to the other on both the ST MS-DOS. Some are commercial, some are on GEnie. Mike, TEAM COMPUTERS ------------ Category 5, Topic 10 Message 39 Sat Jun 06, 1992 K.VANDELLEN [Ken Van] at 08:12 EDT CBARRON, Thanks. I've received help in another topic. Apparently the format I'm likely to get could be converted via TouchUp. Ken ------------ Category 5, Topic 10 Message 40 Sat Jun 06, 1992 K.VANDELLEN [Ken Van] at 12:36 EDT Thanks, Mike. I have both TouchUp and MVG, but I didn't know if they could work with what I would get with the IBM scanner that I have now discovered they have at the college...as part of a DTP setup in the Business Dept. (Too bad they don't know about Atari!) The two programs seem to accept just about everything, between them, but I hope I can use MVG, because it's so much faster. Ken ------------ Category 5, Topic 10 Message 41 Fri Jun 26, 1992 E.STUCKEY at 19:30 EDT I need some help when training OCR. I just purchased the software V. 1.1 1. If a letter appears in a word say "a" and OCR says it should be "o" , when you correct the spelling do you also click on 'train'? 2. If 2 or more characters need to be corrected do you also need to click on 'train'? 3. If a character is highlighted and it is the correct character do you just click 'accept' or click 'train'? Can anyone give me information about flatbed scanners availble for the ST and ones that will work with OCR. Do you need an interface? What are the costs? Where can they be purchased? Thanks.....Ed ------------ Category 5, Topic 10 Message 42 Sat Jun 27, 1992 CBARRON at 03:40 EDT E.STUCKEY 1) It depends if the letter is a 'clear' a press train, if it has a bit of scan garbage, slanted by scanner, etc. press accept. Accept will place the proper character in the text file WITHOUT modifying the dictionary. Train modifies the dictionary as well. 2). Only if they are a litagure(sp?).. 3). See 1). >flat bed scanner input??? If it is a canon compatible there is a DA avail for ocr. If the software must save a full page IMG FILE. (These can get huge!). ------------ Category 5, Topic 10 Message 43 Sat Jun 27, 1992 K.VANDELLEN [Ken Van] at 23:43 EDT Ed, I use train in all three cases, but you will have to type in the correct character(s) in both case 1 and 2. I think I'm doing it correctly. I use accept when what it sees is not a very good image, and I don't want to train it to recognize that, so I correct it and then use accept. An example might be a y with the tail chopped off that it reads as a v. You wouldn't want to train it to read the character that way, so you correct it and click on accept. I see that CBARRON mentioned something about ligatures in regard to item 2. I've used it for any case when it reads two characters together incorrectly. Maybe I shouldn't be training it to read characters two at a time like that. I thought it would learn them separately, rather than as a pair. Maybe I was wrong, and should correct and use accept in that case. If it reads text from an IMG file, I can't see that it would matter how the IMG was made, even by another computer. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 44 Sun Jun 28, 1992 E.STUCKEY at 10:44 EDT Ken Thanks for the help. I will call Migraph to see how they handle ligatures. I will post the feedback. Thanks again. Ed....... ------------ Category 5, Topic 10 Message 45 Mon Jun 29, 1992 K.VANDELLEN [Ken Van] at 06:23 EDT Ed, Then you'll have a chance to return the favor. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 46 Mon Jun 29, 1992 ISD [Nathan] at 10:11 EDT Ed...when you are speaking with Migraph, why don't you ask them to drop by and say hi to us all. :-) They do have a GEnie account and used to be here fairly regularly. ------------ Category 5, Topic 10 Message 47 Mon Jun 29, 1992 L.ROCHA1 [Lou Rocha] at 19:54 EDT I tried to send email to MIGRAPH.KCM and I got "Inactive". Are you sure they are still on GEnie? What is the GEnie address? ------------ Category 5, Topic 10 Message 48 Mon Jun 29, 1992 JEFF.W [ST Sysop] at 21:10 EDT Lou - I think Migraph's GEmail address is MIGRAPH-TECH or MIGRAPH.TECH. ------------ Category 5, Topic 10 Message 49 Tue Jun 30, 1992 K.VANDELLEN [Ken Van] at 07:56 EDT Someone there told me they were changing their account because of a change in personnel or something. I wanted to GEmail a sample to them a few weeks ago, and they said they wouldn't have an account for a week or so, at least that's the way I remember it. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 50 Thu Jul 02, 1992 G.FUHRMAN [gnox] at 07:45 EDT Did anybody get notified that v. 1.1 of the OCR was available? I've heard nothing from Migraph. Still doing fine with 1.05 though. Those who use hand scanners will be familiar with the headaches involved in using Migraph's Merge-It or other such utilities to prepare page-width IMGs for OCR. I've found the ultimate solution for this: Megapaint. With Megapaint you can scan in the right side, pop it into the buffer, scan the left side, and then insert the right side _exactly_ where you want it with the mouse. And if you find (as I often do) that when the top of the page aligns perfectly the bottom is way off, you can move any part of the combined image (even rotate it) with the greatest of ease. Merging two scans this way takes no more time than the physical act of scanning itself. At least until the OCR prg itself includes a good merging function, this is the way to do it! Of course Megapaint is a bit expensive for this use alone, but if you have other uses for a paint program, you can't go wrong with this one. Especially if you do OCR on one-column pages with a hand scanner. gnox ------------ Category 5, Topic 10 Message 51 Thu Jul 02, 1992 L.ROCHA1 [Lou Rocha] at 08:02 EDT GNOX, I know that they have shipped OCR 1.1 for the Amiga.... I have received two copies of the wrong disk! ;-) On the scanning/pasting features of Merge It, I am now using Coalesce from Wiz Works and I am very happy with that program - especially the price! ------------ Category 5, Topic 10 Message 52 Thu Jul 02, 1992 SANDY.W [RT SysOp] at 15:57 EDT Wizworks scanning tray and Coalesce are nice! I believe there is a demo and press release in the Software Library. ------------ Category 5, Topic 10 Message 53 Thu Jul 02, 1992 J.KUDRON [JIM] at 19:06 EDT GNOX, I agree with Lou. All Wiz Works software does more than their Migraph counterpart--for a lot less! Now if they would come out with a low cost OCR package--I'd be a very happy man!! Jim Kudron 2Jul92 18:58:56 ------------ Category 5, Topic 10 Message 54 Fri Jul 03, 1992 H.WOOTAN [Harry] at 02:34 EDT Jim, Maybe THAT's why they've been so quiet lately .... OCR! -- Harry ------------ Category 5, Topic 10 Message 55 Fri Jul 03, 1992 K.VANDELLEN [Ken Van] at 07:32 EDT Dr. Bob has Coalesce, a fine merging program, too, but I've found that just doing my scans in landscape mode does the job for me. I don't mess around with trying to prepare page-width IMGs. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 56 Fri Jul 03, 1992 C.MONTGOMERY at 08:59 EDT 1.1 is shipping now! Just give 'em a call and they'll mail you a copy. Charlie @ Design for Print ------------ Category 5, Topic 10 Message 57 Fri Jul 03, 1992 L.ROCHA1 [Lou Rocha] at 09:40 EDT Thanks Charlie. I would love to get a 1.1 for my ST since I don't have an Amiga ;-) ------------ Category 5, Topic 10 Message 58 Mon Aug 03, 1992 HAINES [Chuck] at 18:30 EDT I hope all the 1.1 upgrade disks they are shipping out aren't like mine. Clicked on floppy install.prg as directed, it asked if installing or updating. Clicked update, it copied files over after asking for path via file selector. But, when I went to run it, it would not go, Neodesk reports it as a non- program. When I looked at the file size, it was only 160K. The file size on floppy is 500K or so. If you copy only the file itself, as the instructions say, when it writes the file to disk, it will be only 160k again. The disk shows as 720K format, 556K used, 542K free! So, there is obviously a problem on the disk. File size reports as correct, but only partly copies, no matter what you do. Guess its back in the mail. ------------ Category 5, Topic 10 Message 59 Tue Aug 04, 1992 J.HAYDEN [JOHN] at 02:35 EDT Chuck, I have the same problem also. I call Migraph and they will send me another disk. ------------ Category 5, Topic 10 Message 60 Tue Aug 04, 1992 J.E.DONOHUE at 07:34 EDT Is 1.1 being shipped out automatically to registered owners or should you get in touch with Migraph and request it? john ------------ Category 5, Topic 10 Message 61 Wed Aug 05, 1992 D.FARRINGTO1 [David <>< .] at 01:35 EDT Well, I just got my upgrade to v1.1. I first tried to install it with the install program and the new OCR program would not run. 2 to 4 bombs. I then tried to install by just coping the two programs from the folder on the floppy (like the read me.now file said). Still 2 to 4 bombs. My OCR program on the upgrade floppy is 536,274 bytes long. Can one of you folks check your length of the 1.1 OCR and/or offer any suggestions on what I'm doing wrong. David <>< P.S. I copied v1.09 back to the folder and the program runs fine. ------------ Category 5, Topic 10 Message 62 Wed Aug 05, 1992 J.HAYDEN [JOHN] at 02:30 EDT David, The problem with the upgrade I have is that there was a cross-link sector between the install and the OCR program. I was not able to repair the disk and Migraph will send me another disk. ------------ Category 5, Topic 10 Message 63 Sat Aug 08, 1992 D.FARRINGTO1 [David <>< .] at 21:51 EDT John: I checked my disk with ST Tool 1.92 for cross-link clusters. In the MI_OCR.prg clusters, number 0091 is linked with another file. I will see what Migraph wants me to do. Thanks for your input. David <>< ------------ Category 5, Topic 10 Message 64 Wed Aug 12, 1992 D.FARRINGTO1 [David <>< .] at 06:06 EDT I just got a Fax from Migraph (Thank you JOPPA) telling me that the upgrade disks to version 1.1 that were sent out between June 11 and Aug. 7 are bad. ONLY THE UPGRADE DISKS! The MASTER DISKS are FINE. They told me to put up this message and tell everyone that new v1.1 disks are being sent out Aug. 11. David <>< ------------ Category 5, Topic 10 Message 65 Wed Aug 12, 1992 L.BUDNICK [Lorne] at 22:17 EDT Question re: the 1.1 upgrade. Does anyone know whether Migraph is sending this upgrade out to all registered users, or must a call be placed to Migraph specifically requesting the update? BTW, I currently have version 1.09. Thanks, Lorne.... ------------ Category 5, Topic 10 Message 66 Thu Aug 13, 1992 LEXICOR2 [Ringo] at 08:07 EDT Lorne I receive the Migraph newsletter monday. It has upgrade prices and information about their new wand scanner and the new Touch-up 1.8 that does 256 greyscales. Ringo. ------------ Category 5, Topic 10 Message 67 Thu Aug 13, 1992 J.E.DONOHUE [Golf Pro] at 20:19 EDT I guess you have to get in touch with Migraph to get the 1.1 upgrade and newsletter. As a registered owner I've never received either. john ------------ Category 5, Topic 10 Message 68 Thu Aug 13, 1992 L.ROCHA1 [Lou (Sysop)] at 21:54 EDT I phoned Migraph and got 1.1 in three days. ------------ Category 5, Topic 10 Message 69 Thu Aug 13, 1992 L.BUDNICK [Lorne] at 22:16 EDT Ringo, Thanks for the response. Hopefully Migraph is sending their newsletter to ALL registered users. I haven't received mine yet, but now I have something else to look forward to checking the mail box for. Lorne.... ------------ Category 5, Topic 10 Message 70 Fri Aug 14, 1992 LYRE at 21:14 EDT Lorne, In the Touch-Up topic I mentioned that Migraph was intending to send the newsletter to registered owners. Of course, I've been registered for Touch-Up since v1.0 and have only received a newsletter twice. Make of it what you will. Lyre ------------ Category 5, Topic 10 Message 71 Sat Aug 15, 1992 B.ROBINSON5 [BRIANMATE] at 06:57 EDT Does one have to contact Migraph for the upgrade or will it be automatically sent to all registered owners. I am still battling with V 1.05 Brian Robinson ------------ Category 5, Topic 10 Message 72 Sat Aug 15, 1992 K.VANDELLEN [Ken Van] at 08:15 EDT I think you have to ask for the upgrade, but if you do they will send it promptly. I received Headlines, their newsletter, yesterday. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 73 Sat Aug 15, 1992 J.MEEHAN3 [>> Joe M << ] at 10:16 EDT My understanding is Migraph will send out the latest version they have to anyone haveing a problem. They will send out the final version to all registered users when they are satisfied they have fixed -all- the bugs. >> Joe Meehan << ------------ Category 5, Topic 10 Message 74 Tue Aug 18, 1992 D.WALTER7 [Doug Walter] at 00:49 EDT Joe, et all, I received my defective 1.1 update and newsletter on August 10th and the working 1.1 disk on the 14th. With the second disk was a letter of explanation and apology dated August 11. I had -not- called Migraph requesting the update, nor about the defective disk. I consider that kind of customer support to be nothing short of excellent! Looks like a little patience is in order here. Doug @ Interface ------------ Category 5, Topic 10 Message 75 Thu Aug 20, 1992 C.MONTGOMERY at 08:05 EDT Without checking my 1.1 disk, I call Migraph yesterday (August 19) and they knew nothing about defective disks. I'm going to check mine out today, and if it is defective call 'em back today. Charlie @ Design for Print ------------ Category 5, Topic 10 Message 76 Thu Aug 20, 1992 L.ROCHA1 [Lou (Sysop)] at 14:38 EDT Charlie, I just received in the mail a second 1.1 upgrade disk along with a letter stating that they had learned that there were some defective disks sent out. Since my original worked fine, I haven't tried the new one. ------------ Category 5, Topic 10 Message 77 Fri Aug 21, 1992 D.FARRINGTO1 [David <>< .] at 02:57 EDT Charlie: When you call talk to Paul. He is the guy who faxed me my info about the disk problems. I also got a new disk from them the other day. David <>< ------------ Category 5, Topic 10 Message 78 Fri Aug 21, 1992 C.MONTGOMERY at 08:26 EDT Thanks for the replys, I'll keep Pauls' name in mind if I have any problems . I tried my 1.1 disk and everything seems work. Mine was shipped on June 26. Charlie @ Design for Print ------------ Category 5, Topic 10 Message 79 Sat Aug 22, 1992 J.E.DONOHUE [Golf Pro] at 11:42 EDT Is everyone receiving the 1.1 update in the mail automatically or must you contact Migraph? I received the newsletter in the mail about a week ago. john ------------ Category 5, Topic 10 Message 80 Sun Aug 23, 1992 J.MEEHAN3 [>> Joe M << ] at 12:54 EDT john, >Is everyone receiving the 1.1 update in the mail automatically or >must you contact Migraph? Why not give them a call and find out? >> Joe Meehan << ------------ Category 5, Topic 10 Message 81 Sun Aug 23, 1992 J.E.DONOHUE [Golf Pro] at 14:43 EDT Gee Joe ... I would have NEVER thought of doing that! (sigh). Maybe I've got the wrong idea about getting information concerning the program from this topic. Pity the folks that don't own modems. But I'll take your "advice". Thanks for the help. john ------------ Category 5, Topic 10 Message 82 Sun Aug 23, 1992 K.VANDELLEN [Ken Van] at 16:08 EDT I thought I had the latest version, and just assumed that the 1.1 you all were talking about was what I had, but mine is 1.09b. I called for that a couple of months ago, but I guess that did not put me on the automatic mailing list. I got the newsletter about a week ago, and thought I might buy that new scanner (with the TouchUp upgrade), so I'll ask for the OCR upgrade at the same time. We do seem to be just pooling our ignorance here, don't we? Nobody here but us users. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 83 Mon Aug 24, 1992 J.MEEHAN3 [>> Joe M << ] at 05:59 EDT Migraph has told me they do check in on this topic from time to time, but due to time limitations they can't do it on a regular basis, therefor they do not want to infer or suggest they are providing online service here. It sounds like a good idea to me. Not providing a service prople are expecting is really a bad idea. I saw that new scanner and I thought I would take take a look at it in Oct at the Washington show. >> Joe Meehan << ------------ Category 5, Topic 10 Message 84 Mon Aug 24, 1992 K.VANDELLEN [Ken Van] at 20:41 EDT Joe, The scanner price is guaranteed until the end of Sept. Check the newsletter. I guess if they aren't going to be here regularly, they'd best maintain a low profile. I hope they benefit from the comments, anyway. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 86 Sat Sep 19, 1992 K.VANDELLEN [Ken Van] at 14:37 EDT Has anyone gotten the Migraph PS400 Wand, yet? I'm thinking about getting it, and wondered if anyone had any good results with it. How do you use it without the sheet feeder? The price list in their latest newsletter expires on 9/30, and I should upgrade TouchUp and OCR, which I can do with a Wand purchase. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 87 Tue Sep 29, 1992 K.VANDELLEN [Ken Van] at 23:26 EDT Tomorrow is the last day to get the Wand on the introductory offer. I understand that without the $199 cradle, the Wand is just a wide hand scanner. Has anyone here gotten it and tried it out? I'd like to get a full-page scanner, but think I might better get a used flatbed rather than this. Any opinion? Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 88 Wed Sep 30, 1992 SANDY.W [sysop] at 20:04 EDT I saw the Wand at Glendale. It does appear to be a wide hand scanner, but the addition of the cradle allows the paper to be drawn through for scanning similar to other non-flat bed scanners. In this regard, I think the option of also using it as a hand scanner, for things you can't send through the rollers, would be an advantage over the other non-flat bed scanners I have seen. The flat bed would probably be the most accurate scanner if you can afford one, but the Wand looked like a nice unit. ------------ Category 5, Topic 10 Message 89 Fri Oct 09, 1992 A.DEFFENDERF [ALAN] at 02:54 EDT Is Migraph doing anything about making the OCR work on Dave Small's SST board for the ST? I'd love to use the OCR on the SST. Alan ------------ Category 5, Topic 10 Message 90 Sun Oct 18, 1992 K.HULET at 15:28 EDT Hi, I just got MiGraph OCR at WAACE. I've used the demo and now have a quest question. Page 42 of the manual says you can start new user dictionaries, I assume to learn from different graphic sources. I tried to create a dictionary for the demo to see if it would process faster the next time using the "demo" dictionary. I'm sure I followed all instructions but no .ALD files exist anywhere, there's not even a standard or default dictionary. Also, even if you could create a new dictionary how would yyou direct the program to use it? THANKS ........ Ken Hulet ------------ Category 5, Topic 10 Message 91 Mon Oct 19, 1992 CBARRON at 02:45 EDT The location of the dictionary and output file, as well as new,append, etc to a dictionary are all in a dialog box on a configuration choice of one of the menus used to start the ocr process. It is fairly simple. ------------ Category 5, Topic 10 Message 92 Mon Oct 19, 1992 K.VANDELLEN [Ken Van] at 07:58 EDT Ken, I've generally had good success with OCR, but sometimes (always) it doesn't save a dictionary for me, either. That's not too bad if you're just doing one page, but if you have more pages, you have to do each one the long way. I know I've done multipage documents, though, without that hassle, so it must save the dictionary sometimes. I think there's no default dictionary because you have to make one (at least in memory) for your particular document. Then, if you get it to save, you'll have it later for that same type of document (point size, font, etc.). Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 93 Tue Oct 20, 1992 D.STMARTIN [Binary Ink] at 20:56 EDT I've got a project that requires me to scan in phone book listings and I've been having OCR throw in extra spaces especially in the number itself but in some other positions as well. Most phone books appear to be a mix of porportional names followed by monospaced numbers (for alignment proposes I'm sure). I can write some code that attempts to clean up the output, but I'm curious to know if anyone else has slain this dragon. I've played with the PROPORTIONAL/FIXED and SMALL TYPE/REGULAR settings on Migraph OCR to no avail. I've come to the conclusion that Migraph should consider an adjustment feature that would allow us to set a character spacing factor. I realize it wouldn't be 100% but I think I could get better OCR with it. My other desire (discovered during this phone book fiasco) is a switch for the linguistic processing. For the phone book project anyway, linguistic process was of no value but a CONSIDERABLE drain on processing time. I'd like to be able to turn it off unless it's doing some black magic mumbo jumbo that I'm unaware of. . . . Comments, suggestions, thoughts -- Please. . . ------------ Category 5, Topic 10 Message 94 Tue Oct 27, 1992 G.FUHRMAN [gnox] at 05:42 EST to the Kens, I'm not sure I follow what you're asking about but ... The `default dictionary' is actually built into the engine, so there's no ALD file for it. In my experience, a new ALD file will always be created and saved IF you configure the control panel to do that; and you can use it again by configuring to Append or Read the specified dictionary. Can you explain step by step a scenario where you expect to get an ALD saved and you don't? Binary ink, You probably get a lot of spaces after 1s, right? I do. I figure they're just a fact of life. I haven't come up with a better solution than putting the text through a search-and-replace with query in a word processor (searching for 1-space). gnox ------------ Category 5, Topic 10 Message 95 Tue Oct 27, 1992 LYRE at 18:24 EST Greetings, So how is this program? I'm considering it and wondered if any of you would offer your thoughts concerning it. Lyre ------------ Category 5, Topic 10 Message 96 Wed Oct 28, 1992 D.FARRINGTO1 [David <>< .] at 03:04 EST I've been converting from typewriting print to ASCII quite a bit. Once the program has learned the scanned images it works great. I can't wait to see how fast it goes with a tiny 030. After converting and the program learning a dozen songs, the dict. seems to be finding 99% of the letters correctly. I still seem to have problems with the punctuation. The program about 85% of the time puts a space between the comma and the word it follows. But I like the program. It helps and if you watch you can find it at a bargin. David <>< ------------ Category 5, Topic 10 Message 97 Wed Oct 28, 1992 D.STMARTIN [Binary Ink] at 06:56 EST gnox: You hit the nail on the head - extra spaces everywhere. I'm scanning telephone books and they're a unique combination of BOTH fixed and proportional type. It certainly would be nice to have a parameter that we could set (via a slider perhaps?!?) that would allow you to "fine tune" space recognition. Thanks for the tip on search and replace. That's what I've been doing. ------------ Category 5, Topic 10 Message 98 Wed Oct 28, 1992 LYRE at 18:39 EST David (or anyone else), Could you tell me the system requirements for OCR? Mono or Color? Can it learn various typefaces? Can you save the learned typefaces so that you don't have to do substantial manual editing each time you scan? Does it require that each font be recognized at every point or does it allow for some type of vectoring to compare say, Times Roman at 10 points versus Times Roman at 16 points and determine that they are the same font? Sorry to ask you (or anyone else who chooses to reply here) all of these questions. But I did not receive the OCR flyer/info and am not well acquanted with facts / hands-on info about the program from users first hand. Lyre ------------ Category 5, Topic 10 Message 99 Thu Oct 29, 1992 CBARRON at 03:35 EST Migraph's OCR requires two megs of RAM and a hard drive. It runs in STmed, SThigh and TTmed for sure. It can save separate dictionaries if desired. Works fine if the input is good. ------------ Category 5, Topic 10 Message 100 Thu Oct 29, 1992 D.STMARTIN [Binary Ink] at 06:09 EST I picked up a copy at L&Y Electronics in Woodbridge, VA for $150.00. They advertize in Current Notes and I recently noted that they had it in their ad for $199.00. You might check with them though - they do run "specials" frequently. As far as the program goes, you might want toi read some of my previous posts regarding extra spaces. Overall the program functions as advertized at a cost far than that of OCR on the "other" platforms. I would like a little more control over the OCR process, however. A parameter for spacing is just one example of what would improve the program. Playing with the present OCR control panel has made a significant difference in the things I scan -- both good and bad! ------------ Category 5, Topic 10 Message 101 Fri Oct 30, 1992 K.VANDELLEN [Ken Van] at 06:40 EST gnox, I believe that I set up a file name for the text, the dictionary, etc., and when I scanned and processed a document, I got a text file but no dictionary file. I have gotten an ALD file sometimes, though. Maybe I messed up. I'll check it out. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 102 Fri Oct 30, 1992 D.STMARTIN [Binary Ink] at 07:09 EST The Migraph OCR package is based on the OmniFont OCR engine which covers 20 typestyles to start with and as I understand it, is trainable on others. It trains on your documents and refines its recognition based on input from you. Once trained you can let it rip on its own and then search for the "@" character which it inserts where it gets confused. Although a previous message said it requires 2 megs of RAM, I understood it to require four. At any rate the more RAmthe happier the system is. It does use your hard drive for virtual processing and appears very smooth at that. For my money Migraph OCR is worth it. It just needs some minor fine tuning which Migraph has shown to do in the past. ------------ Category 5, Topic 10 Message 103 Sat Oct 31, 1992 CBARRON at 01:57 EST The packaging says two megs and a hard drive. I have only used it with 4 megs... It works most of the time. ------------ Category 5, Topic 10 Message 104 Sat Oct 31, 1992 R.BEATTY3 [Buffalo Bob] at 03:46 EST I use OCR in TT Medium, but I'm sure it will work in ST Mono. Unsure of ST Medium. It will create a "dictionary" of font styles that you can use anytime you scan. I often scan documents that come in two fonts faces, and two point sizes. The first few scan required a lot of editing, but it has since become a breeze. OCR will usually ask me to clarify on five or six characters now, and often those characters are punctuation or 1/l 0/O type of things... Bob Beatty ------------ Category 5, Topic 10 Message 105 Sat Oct 31, 1992 G.FUHRMAN [gnox] at 07:34 EST Lyre, It also works in TT High, and it does learn and allow for font scaling via a couple of parameters. I'm still using v. 1.05 (no longer current) and find that it generally saves time over retyping text, if the text scanned in is fairly uniform in darkness and not too tightly spaced between characters and lines. When many characters touch each other it slows everything down drastically. It's a pretty safe buy and has been well supported by Migraph (although not directly in this Topic). I'm glad I bought it. gnox ------------ Category 5, Topic 10 Message 106 Sat Oct 31, 1992 K.VANDELLEN [Ken Van] at 08:19 EST I've been scanning in chapter from a textbook that I hope to take over authorship of through a phase-out, phase-in deal. The editor wanted me to redo a chapter, and the author didn't use a computer, so it isn't on disk. OCR is working great. It only took about a column of text to set up its dictionary. I then switched to interactive, and it flies right through. Interactive is a lot faster in the long run than automatic, because it's easier to do the editing along the way than to have to look for the errors (@, &, etc.) and pounce on them. I'm thinking of getting the Wand, although I'm annoyed that the sale period was so short, and no reviews or articles came out before it ended, so now we have to pay more for it. Has anyone had experience with it? They tell me that most people get the sheet feeder (cradle) about a month after buying the Wand, so I guess we should expect that that's a necessity. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 107 Sun Nov 01, 1992 CBARRON at 02:54 EST I have used MI_OCR in TTmed,STmed, and STHigh. It is resolution independent. (requires 'at least 80 col' resolution I believe and that is all for res. requirements.) ------------ Category 5, Topic 10 Message 108 Mon Nov 02, 1992 L.BUDNICK [Lorne] at 19:51 EST Ken, I received my Wand and sheet feeder in the same shipment. My experience thus far is very positive. I really appreciate being able to read in full pages. I am using it with the latest release of OCR and TouchUp. ------------ Category 5, Topic 10 Message 109 Tue Nov 03, 1992 D.STMARTIN [Binary Ink] at 07:27 EST Lorne: >I received my Wand and sheet feeder in the same shipment. My experience thus >far is very positive. I really appreciate being able to read in full pages. >I am using it with the latest release of OCR and TouchUp. I'm considering a purchase of the Migraph Wand. Thus far my experience with Migraph products has been very positive. The Migraph OCR software works best when the scan is "straight and true". Are you using the cradle unit as well as the Wand? How straight is the end result? Have your results with Migraph OCR improved? If you would please comment on the QUALITY of the scan. Thanks. ------------ Category 5, Topic 10 Message 110 Tue Nov 03, 1992 LYRE at 20:08 EST Just a comment concerning OCR's system requirements / abilities according to the flyer that they sent me (on request): o Process any scanned text in IMG, IFF or TIFF format. o Save scanned text in ASCII format. o Direct support of Migraph, Alfadata and Golden Image hand scanners. o Allows use of other scanners - including flatbed, but not directly supported. o Omnifont technology (mathematical definitions, not raster images) used to recognize text. o Automatic recognition of 20 fonts (including Courier, Helvetica, Times, Bookman, Letter Gothic and Artisan). o Font sizes recognized for included fonts is between 8-18 points. o Reads 8-18 point proportional or monospaced fonts, whether typewritten, typeset, laser printed or NLQ dot matrix. o Linguistic databases for English, German and French make intelligent interpretations of unusual, defaced or questionable characters. o Use locator, multiple zoom levels and polylines to view and select text to save time and disk space. o Works on any ST, STe, Mega or TT computer with 2MB of memory and a hard disk. (4MB recommended for flatbed scanners.) All of the above is from the flyer. It has been paraphrased in places, but should be basically correct. Just thought that I would pass this info on to anyone else interested. Lyre ------------ Category 5, Topic 10 Message 111 Tue Nov 03, 1992 L.BUDNICK [Lorne] at 23:02 EST Binary Ink, I am using the cradle unit as well as the Wand. The need to have 'straight and true' scans was the very reason I ordered both the Wand and the Cradle unit. My operation of the hand scanner was not consistant enough to give me that type of scanning. I guess my hand is not steady enough. :^) Anyway, since using the Wand installed in the cradle, my OCR results have improved dramatically. The scans are truly 'straight and true'. I have been quite happy with the results, and with the support I have received from Migraph. LGB ------------ Category 5, Topic 10 Message 112 Wed Nov 04, 1992 K.VANDELLEN [Ken Van] at 21:00 EST Lorne, Thanks for the info. I'm sorry I missed out on the sale that ended a month ago, but I wasn't anxious to spend that much money without knowing more about the product. I ordered Wand, sheet feeder, and OCR and Touchup upgrades, but the price has gone up considerably, so I'm paying more for having gotten some assurance. Migraph sales said there was an article in ST Informer, but Rod said they had not used the product, so it was hardly a review, just an announcement. Ken Van Dellen ------------ Category 5, Topic 10 Message 113 Wed Nov 04, 1992 L.BUDNICK [Lorne] at 22:38 EST Ken Van, No problem. To bad about having to pay more now though. I guess that's why I jumped on the original offer 'site unseen'. The pricing was very attractive. Also, I felt that the quality of the products and service I had obtained from Migraph in the past was very good, thus minimizing the potential risk associated with the new product. Hope you find your newest acquisition to be as functional as I have. LGB ------------ Category 5, Topic 10 Message 114 Fri Nov 06, 1992 D.STMARTIN [Binary Ink] at 07:27 EST Lorne: Do you have any experience with any other (flat bed) scanners that you might compare to the Migraph Wand+Cradle unit? I'm especially interested in the Canon/Navarrone unit. It would seem to me that any time you're moving the paper during the process there's a likleihood for "skewing" to occur. Granted, some might do a better job than others. Also, have you seen any speed increase in the initial processing stages of OCR? I would think that a straighter scan would make the process of breaking down the lines and letters faster. ------------ Category 5, Topic 10 Message 115 Fri Nov 06, 1992 K.VANDELLEN [Ken Van] at 18:15 EST Lorne, They wanted $110 more than the special price, but I wangled them down to $50 over, which was still too much, under the circumstances, but they give good support (but not here), so I decided I'd just contribute the extra to their overhead expenses. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 116 Fri Nov 06, 1992 L.BUDNICK [Lorne] at 20:37 EST Binary Ink, I'm afraid that I have not had an opportunity to work with any flat bed scanners. While what you say about 'skewing' with paper movement would seem to be the case, I have not noticed this when using the Wand/Cradle unit. I suspect that such would be the case if the feed and rollers do not hold the paper properly. My experience has been that the feeder captures the paper firmly, and there has been no movement of the paper during scans other than the feed thru process. My perception regarding the initial processing speed is that if the copy being scanned is in good shape, and scans are not 'skewed', the processing speed is faster. This is my impression based on scanning I have done with the original hand scanner, (skewed scans), and the scanning I have done with the Wand/cradle combination, (non-skewed scans). In either case, if the copy being scanned is not clear, the learning process will take longer. Ken Van, Well it doesn't sound as if it was quite as bad as it could have been. :^) ------------ Category 5, Topic 10 Message 117 Sat Nov 07, 1992 K.VANDELLEN [Ken Van] at 12:48 EST Lorne, I appreciate your reassuring comments about skewing. That was a concern of mine, too. I would hope that for that price skewing would not occur. Otherwise there would be little benefit in getting the product. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 118 Fri Nov 13, 1992 K.VANDELLEN [Ken Van] at 07:45 EST The Wand and sheet feeder arrived this week, and work very well. If you get the feeder, read the instructions BEFORE unpacking it. I unpacked it, and then discovered that the instructions say you should let the feeder push through a little piece of plastic that's in the mechanism, rather than just pulling it out. Fortunately, I didn't damage anything, as far as I can tell, but a warning on the plastic would have been useful. Ordinarily I unpack an item, and then read the instructions to see how to make it work. It seems reasonable. Even with the scanner turned all the way to "DARK", some letters don't have their parts joined. An "n", for example, shows up in the OCR as an "l" and a "candy cane". I don't know how to train the program to recognize the two parts as an "n". It would be useful if one could combine parts like this. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 119 Sat Nov 28, 1992 R.PILATO [bob] at 14:28 EST I just received a Holiday offer from Migraph which includes a new Jr. version of OCR for $149.00 through Dec 31. (after $199). It has "all the power and features of the full Migraph OCR, except it does not load image files." Can someone tell me how important it is to have this missing feature? Thanks. PS. The mailpiece included other attractive offers such as OCR Jr., The handscanner, and Touch-Up for $299 Bob Pilato ------------ Category 5, Topic 10 Message 120 Sat Nov 28, 1992 D.FARRINGTO1 [David <>< .] at 19:19 EST bob: Migraph offers scanner software. I would think this is built into or comes with the OCR. My guess would be the OCR jr. will allow one to scan in only and export text only. Only a guess, mind you. :) David <>< ------------ Category 5, Topic 10 Message 121 Sat Nov 28, 1992 LYRE at 19:25 EST Bob, Thanks for the information. Hopefully I will receive this mailing so I can check out this information first hand. I've wanted OCR but the need for additional memory and the cost of the program was limiting me. Lyre ------------ Category 5, Topic 10 Message 122 Sun Nov 29, 1992 K.VANDELLEN [Ken Van] at 00:40 EST Bob, I never load image files in the Sr. version. I just work with what I scan. I wish they had had this deal when I bought. The Wand is great. It does excellent work, and is easy to use. I get good results every time, except, of course, for occasionally adjusting the contrast. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 123 Sun Nov 29, 1992 CBARRON at 02:58 EST I think it is very important to be able to import files into an ocr program. Unless scanner support has been vastly improved, the support for flatbed scanners is minimal with the program itself. I have taken the faxes of straight fax, converted to img in straight fax and ocr'ed with miocr them with no paper involved as another example. The resulting text files are a lot more useful than an img of the printed text. I'd say go for it all. Heck I'd like it to load mono pcx files as well, making the use of someone else's flatbed a lot easier. File import is important. The full thing also allows saving of graphic areas as separate img files, does ocrjr?? Memory and hardware requirements of ocr are due to the nature of the problem and not file importation, that might save 30K at best. I would guess that ocrjr has the same memory requirements as ocr. What is the difference in cost, its been a while since I checked prices of this stuff. ------------ Category 5, Topic 10 Message 124 Sun Nov 29, 1992 R.PILATO [bob] at 22:21 EST Thanks for sharing your opinions... Bob Pilato ------------ Category 5, Topic 10 Message 125 Mon Nov 30, 1992 K.VANDELLEN [Ken Van] at 07:51 EST CBARRON has a point. I have had no occasion to work with any files other than the ones I produce myself with the hand scanner or the Wand, so import wouldn't have affected me. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 126 Mon Nov 30, 1992 D.STMARTIN [Binary Ink] at 19:13 EST I'd say it depands on how you'll be using the OCR software. If your an occaisional user and happy with the supported hand scanners then OCR Jr. should fit the bill. In reality, this provides Migraph ab in for their Wand system as this is probably the only full page scanner that is supported by OCR Jr. This would spell future sales potential. My guess would be that if your not planning on getting a full page, flat bed scanner other than the Migraph Wand, the OCR Jr. package should support your needs. ------------ Category 5, Topic 10 Message 127 Mon Nov 30, 1992 LYRE at 19:54 EST I just got the notice from Migraph. Thought I would post if here in case anyone is lurking and wants to know. -=*=- -=*=- -=*=- NEW! Introducing Migraph OCR Jr.! This new version has all the power and features of the full migraph OCR, Except it does not load image files. Scan images directly using a Migraph, AlfaData or Golden Image hand scanner. Suggested retail $199. Requires 2MB RAM. *Upgradeable to Migraph OCR (full version). Holiday Scanner Sale! Hurry, Specials end Dec. 31st! Super Scanner Bundle* $299.00 includes Hand Scanner, Touch-Up, and Migraph OCR Jr. Hand Scanner & Touch-Up $199.00 Migraph OCR Jr. $149.00 Scanning Tray w/Merge-It & Scan & Save $ 69.00 PS-400 Wand Full Page Scanner $649.00 Scan full page images in a single pass with the PS-400 full Page Scanning Wand. 100-400 DPI scanning resolution, 4 other settings, 1 line art setting, adjustable contract, up to 256 grayscales via software conversion. Includes Wand, Touch-Up, Migraph OCR (full version), interface and power supply. PS-400 Sheet Feeder $199.00 Scans pages automatically. Requires PS-400 Wand. Call 1-800-223-3729 today! or FAX 1-206-838-4702 Sale ends December 31, 1992. Visa/MC/Checks accepted. Additional shipping charge. UPS Blue Available. Sales Tax where applicable. No cash refunds. Exchanges only. ------------ Category 5, Topic 10 Message 129 Tue Dec 01, 1992 LYRE at 00:31 EST Ringo, Thanks - I think. ;) It's just that I hate hearing of these things without actually being able to read what was provided. Sometimes things "get lost in the translation". For instance, having to create that message I noticed that it doesn't say Jr. supports data from other scanners - which can make a bid difference. This was not something I would have considered if it wasn't for posting the information here. Lyre ------------ Category 5, Topic 10 Message 131 Fri Dec 04, 1992 R.PILATO [bob] at 22:20 EST I was glad to learn that the hand scanner/OCR can handle pages of text that are wider than the hand scanner window. The OCR program allows you to scan a block of text horizontally. It then automatically flips the scan 90 degrees...So I ordered the OCR Jr. today. And those folks at Migraph are always so courteous and helpful on the phone! Glad they're on our side. Bob Pilato ------------ Category 5, Topic 10 Message 132 Tue Dec 08, 1992 LYRE at 19:16 EST Greetings Everyone! Having contacted Migraph the other day, here is what I found out about OCR Jr. Appearently it is the same exact program as the full version OCR except that it does not import/export IMG files. Therefore, I believe, that the suggestions people have made here regarding the best method(s) they've found for correcting scanned text are viable on Jr. OCR Jr is limited in regards to the scanners that it supports though. Only the hand scanners mentioned (Migraph, AlfaData or Golden Image) will function properly. It is *not* possible to use your existing scanner wih this program (assuming the scanner isn't one of the above)! The upgrade cost to the Full OCR package is $50.00. I did not ask if their was any time restraint on this upgrade though. Migraph will be sending me more information (Full and Jr). If their is anything new, I'll let you know. :) Lyre ------------ Category 5, Topic 10 Message 133 Mon Jan 04, 1993 M.SHANNON at 00:28 EST Ken, Did you get a reply to the 'broken letters syndrome'? Cause I hav it too and need the fix before I go crazy or somethin. Thanks MS Ken, I just found the switch next to the DPI switch that, when set to 'letter' gives me the contrast I needed in the prior message... you found it too? MS ------------ Category 5, Topic 10 Message 134 Wed Jan 06, 1993 K.VANDELLEN [Ken Van] at 00:10 EST MS, I guess I never noticed that switch. I'll give it a try. If it fixes the broken letters, it should make the program really fly. Ken ------------ Category 5, Topic 10 Message 135 Sat Jan 09, 1993 M.SHANNON at 21:17 EST I'm having a problem with (it looks like) 11pt. Times Proportional fully justified RIGHT: it would seem that there are countless combinations to deal with in this combination... too many in fact to begin to mention... so perhaps I've answered my own question here. thanks MS ------------ Category 5, Topic 10 Message 136 Sat Jan 09, 1993 K.VANDELLEN [Ken Van] at 23:21 EST I'm still having a problem with broken letters. I have the Wand set for letter and the contrast turned all of the way to dark. Letters such as b, m, h, r, and u consistently show up in two parts. When I magnify the screen in OCR, there is a connection where the break occurs, but the program doesn't see it, apparently. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 137 Sun Jan 10, 1993 G.FUHRMAN [gnox] at 07:18 EST Ken, Last night I started using 1.24 (after upgrading from 1.05) and I see what you mean about broken letters. It seems to happen much more now than it did with 1.05. Today I'll be doing some work with better copy (the stuff I was doing yesterday was so bad it would be faster to type it) and I'll see what happens. Sure is nice having the function keys though! That really speeds things up. gnox ------------ Category 5, Topic 10 Message 138 Mon Jan 11, 1993 K.VANDELLEN [Ken Van] at 07:31 EST gnox, Your previous version didn't have function keys? I thought they all did. I get broken letters with stuff I do on my computer with the laser printer. How much better can you get, unless it's put out by a print shop? After I make a dictionary, and try to do later pages with it in auto mode, I get two letters for one in many cases. I think that can be fixed. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 139 Tue Jan 12, 1993 G.FUHRMAN [gnox] at 06:46 EST Ken, I think the function keys were added shortly after 1.05, which is what I had for several months. Your description of `broken letters' sounds a bit different from what I'm seeing - I'll have to look into this further. gnox ------------ Category 5, Topic 10 Message 140 Wed Jan 13, 1993 K.VANDELLEN [Ken Van] at 07:10 EST gnox, Do you understand what I mean by broken letters? A b comes out as an l and a backwards c, or some such combination, and others behave similarly. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 141 Sat Jan 16, 1993 M.SHANNON at 04:11 EST Ken, I've had to photocopy some stuff to improve the contrast before running OCR... it helps. Too much contrast makes typeface 'fuzzy' and they [merge] which is the opposite, I guess of [broken] letters. ms gnox, What [function] keys are you referring to? Or should I RTFM? or do you mean the, for instance, th F-2 for Delete type keys? ms Ken, AUTO mode leaves me with 'blips' and 'marks' and garbage if I use it at all... I have to monitor all the imput to get a usable output file. I get a double ["]'s and double [h]'s aand [nn]'s for "m" a lot and I can't figure out how to retrain my .ALD file to correct it... can that be fixed??? I hate to think I have to start over building a new .ALD file for a simple but extremely aggravating correction-in-need! I also have to correct ALL THE TIME with broken [n]'s and [h]'s. What a pain... no macros!! What about you... MS Ken, Perhaps we can upload some 'exemplary' .ALD files here?! Like one that works with RIGHT-JUSTIFIED SERIF TEXT (8-11pt) for instance. ...or one that works flawlessly with courier 10pt. What-say-U? MS ------------ Category 5, Topic 10 Message 142 Sat Jan 16, 1993 G.FUHRMAN [gnox] at 07:32 EST MS, Yes, I'm referring to the five function keys that are indicated in the interactive-learning dialog. Until this month I was using an older version of OCR where you had to do all those things with the mouse. So far, what I've found is that the input to the OCR is the most critical factor - a clean scan will mean fairly clean text output. Still, it would be nice to have the capability to edit an ALD. I always run the resulting text through WordPerfect and often find that half of the problems can be fixed up with a simple search-and-replace. Also, the latest version (1.24) is supposed to replace '' and `` with " when it outputs the text. I can't imagine why right-justified text would cause special problems, though. gnox ------------ Category 5, Topic 10 Message 143 Sat Jan 16, 1993 K.VANDELLEN [Ken Van] at 08:53 EST Mike, I'll try photocopying, but that's an extra step and probably considerable lost time, by the time you go to a copy machine, and it shouldn't be required. I've run into the merging problem, too, with some documents. I think gnox had an early version that didn't use the function keys (F-9 = Learn, F-10 = Accept, etc.). Oh, yes. The double quotation mark (") almost always comes out as two singles (''). I don't know how to fix that or nn for m, etc. There should be a way to merge the broken letters, instead of manually correcting one part and deleting the other part. (Say an h gives ln. I correct the l to h, accept it, and delete the n.) There should be a function key that would allow the l and n to be merged, and then train the file to recognize it as h. Nice idea to upload .ALD files, but how do you get them in the first place? I've had some docs that did scan almost flawlessly, however, so it can happen. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 144 Sat Jan 16, 1993 K.VANDELLEN [Ken Van] at 19:18 EST gnox, V. 1.24 doesn't replace '' with " for me. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 145 Sun Jan 17, 1993 M.SHANNON at 23:39 EST gnox RightJustified proportional text creates an unbelievable number of variable spaces (chaotic kearning?) especially with serifed fonts like TIMES. Took me pages upon pages to create an .ALD file worth it's salt. I run my output through IdeaList and then STeno and use Thunder to spell check the OCR.TXT file. I'll try WP tho'. ms Ken, Is anybody out there besides you me and gnox? Maybe were victims of room at the top? You create your own .ALD files as required by document type; one for magazines, one for newsletters, one for typewritten letters, etc. ms gnox, Mine doesn't replace it either! And yes lets get them to teach OCR to make the correction: l-n = h then accept & learn. GREAT IDEA! ms ------------ Category 5, Topic 10 Message 146 Mon Jan 18, 1993 G.FUHRMAN [gnox] at 06:55 EST >V. 1.24 doesn't replace '' with " for me. The readme on the disk says that it does this _when it saves the TXT file_ (not before). I haven't done anything with quote marks since I got 1.24 so don't know if it works for me. ms, >RightJustified proportional text creates an unbelievable number of variable >spaces (chaotic kearning?) especially with serifed fonts like TIMES. Took me >pages upon pages to create an .ALD file worth it's salt. I see, you mean it's justified at BOTH margins. I can see how that would create extra spaces, but I didn't know it was possible to correct for this with the ALD file. Interesting. How do you do that? Yesterday I did a 9-page document that was done on a 24-pin printer, justified on both sides, and used the non-proportional spacing setting - worked great! gnox ------------ Category 5, Topic 10 Message 147 Mon Jan 18, 1993 K.VANDELLEN [Ken Van] at 07:12 EST Mike, I understand about making different .ALD files for different source docs, but why the problem with justified? Does OCR run the letters together? It certainly doesn't put spaces in or something like that. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 148 Wed Jan 20, 1993 K.VANDELLEN [Ken Van] at 06:25 EST It doesn't fix '' in the text file for me. :^( Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 149 Sat Jan 23, 1993 M.SHANNON at 03:11 EST gnox, Yes, it was justified BOTH ways and the only way I know of is to create an >ALD file that has learned 97% of the patterns involved! Whew..... Non-proportional text is great, and simple to do, the problem is that most publications (excerpts I want) have used proportional serif'd fonts and right-justified (both) them. A real pain since I cant teach OCR to make an "h"= l+n. I'm bummed out here... MS Ken, Don't make me say it again... :^[ For me R-Justified is BOTH. Is there another term? MS ------------ Category 5, Topic 10 Message 150 Sat Jan 23, 1993 G.FUHRMAN [gnox] at 06:45 EST MS, Ah, the light came on that time. You're trying to train the OCR to recognize _combinations_ of characters, right? And the serifs cause trouble because they run together. Now I'm more surprised that it works at all than that it's a tedious process! I have never even tried to do that; I never train it on more than one character at a time. When multiple characters pop up I just Accept them unless they're ligatures, like "fi" in some fonts. If I'm getting lots of characters joined, I re-scan at a lighter setting until most of them are separate - and if that doesn't work I exit the OCR and start typing, because I know I'll get the job done faster that way. I'm not saying that your method is wrong, just that _I_ wouldn't do it like that. :) According to my sources, simply "justified" means that the type is flush with both margins. I was thrown off by your use of the word "right" with it. Text flush with the right margin only is generally called "flush right," "block right" or "ragged left." Same with "flush left" etc. gnox ------------ Category 5, Topic 10 Message 151 Sat Jan 23, 1993 J.MEEHAN3 [>> Joe M << ] at 09:22 EST I think I just figured out what you all are talking about. There are -THREE- kinks of justification LEFT This is the one you will see here. RIGHT The right edge is lined up and the left edge is ragged FULL Both edges are lined up by adding spaces between words or letters I know a lot of word processors don't seem to know the difference but it would be nice if they could start using the right terms. BTW this paragraph is right justified. >> Joe M << ------------ Category 5, Topic 10 Message 152 Sun Jan 24, 1993 G.FUHRMAN [gnox] at 07:05 EST Joe, > I know a lot of word processors > don't seem to know the difference but > it would be nice if they could start > using the right terms. Unfortunately, there is not as much agreement about "the right terms" as you seem to think. Pagemaker and PageStream (for example) both use the term "justified" to mean lined up on _both_ edges. Shusan and Wright in _Desktop Publishing by Design_ use the term "alignment" where you use the term "justification". None of them use the term "right justified" because of the possible confusion. (This is hardly a crucial issue, but I can't resist debates about words. Hey, I'm an English teacher!) gnox ------------ Category 5, Topic 10 Message 153 Sun Jan 24, 1993 ST.LOU [Lou Rocha] at 10:28 EST GNOX... you are correct about the inconsistency of terminologies. In my experience I have seen these terms used synonymously: justification = block justfication = full justification My preference is for 'full' as it better complements 'center', 'left' and 'right' justification. ------------ Category 5, Topic 10 Message 154 Sun Jan 24, 1993 J.MEEHAN3 [>> Joe M << ] at 11:46 EST gnox, Unfortunately you are right. However the terms "justified" and "alignment" are just shortened forms of "full *" It's too bad that when a term that is as confused as this one is; the top names in the field would take shortcuts when they should be the ones trying to clarify the situation. >> Joe M << ------------ Category 5, Topic 10 Message 155 Sun Jan 24, 1993 K.VANDELLEN [Ken Van] at 14:17 EST Mike, I understand right-justified, and I understand that you prefer to use the word both, instead, which is OK, as long as you define "both" for people you communicate with. What I don't understand is why right-justification is a problem for OCR, since I thought it just looked at the characters and not the spaces, and the characters don't change when the format changes. I guess the same thing is true of proportional, though, and we have to set OCR for that, so apparently it wants to have an idea what the spacing between the characters is, which with right-justification is variable. I think I'll write Migraph. (I've had laryngitis for almost a week, which has limited my phoning. I haven't had anyone hang up on me, but people _do_ wonder what's happening when I start the conversation in a whisper.) Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 156 Mon Jan 25, 1993 K.VANDELLEN [Ken Van] at 06:21 EST I've found that sometimes OCR will show two characters in the "viewing box" and its interpretation of them will be correct, so I train it to recognize such combinations for what they are. I think I may have been the one who introduced the term "right justified" here. I've heard it used, but don't know where. Since the left margin is usually straight, it was my impression that it was the right margin that we need to discuss. I note that PageStream talks about block left (straight on left, ragged on right), block right (ragged on left, straight on right), and justified (straight on both sides). The important thing is that we understand each other. :^) Ken Van ------------ Category 5, Topic 10 Message 157 Tue Feb 02, 1993 J.MOSES2 [moses] at 20:40 EST Does anyone know if the hand scanner from Migraph is IBM compatible and/or if an interface card is available for it? Thanks, Jim ------------ Category 5, Topic 10 Message 158 Tue Feb 02, 1993 K.SPRINGER1 [FROZEN NORTH] at 22:52 EST I suspect it is. A friend of mine has a virtually identical scanner physically that he got with PC Paintbrush. The was a card for insides of his compatible. We've never had the guts to try and experiment by trying one scanner on the other's machine. I have a yen for a full page scanner now, so if that ever comes to pass I probably try my scanner on his machine. If the are identical, then both are DFIK hand scanners. Ken Bettles, AK ------------ Category 5, Topic 10 Message 159 Thu Feb 04, 1993 M.SHANNON at 01:32 EST gnox, yeah, I got that one straightened out too. How do we convince These guy\girls at the OCR factory to let us "teach" it just what we want recognized (like joining l+n=h)? Got any ideas? MS ps: Do you remember the teacher of the year in '92 or '93 and his acceptance speech having to do with forced education in America and how it was last enforced in 1913 in Maine? ----------- ------------------------------------------------------- Joe M. Thanks Joe, I think we got it this time :^) MS d8^, Please do write them and let us know whats up! Thanks ahead of time MS ------------ Category 5, Topic 10 Message 160 Sat Feb 06, 1993 K.VANDELLEN [Ken Van] at 06:41 EST I spoke with Paul at Migraph today about split letters. I told me to send him a disk with an IMG file on it, and I will put it in the mail tomorrow. He first said my v. 1.24 was old, and would send me 1.4, but then said Amiga was up to 1.4, but not Atari. I'll keep you posted on developments. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 161 Sat Feb 06, 1993 D.FARRINGTO1 [David <>< .] at 15:57 EST Ken: I'm watching this version experence with interest also. My OCR version is 1.10 for the ATARI. David <>< ------------ Category 5, Topic 10 Message 162 Sun Feb 07, 1993 G.FUHRMAN [gnox] at 07:43 EST Last weekend I did the biggest scanning job I've ever done with the OCR, and 1.24 performed beautifully; the input was clean, so the output was very accurate. No split letters at all. gnox ------------ Category 5, Topic 10 Message 163 Sun Feb 07, 1993 J.HOPKINS16 [John] at 15:53 EST Moses, Yes, according to my 12/92 Migraph News Letter, the PC card is available from Migraph for the Migraph and Golden Image handscanners. Cost: $99. It works with PC Touch-up. "Same as Atari Touch-up V1.8". The blub says that you can now use your Atari hand scanner on your PC. -- John ------------ Category 5, Topic 10 Message 164 Sun Feb 07, 1993 L.BUDNICK [Lorne] at 16:14 EST Gnox, I have experienced the same results with similar input, also using version 1.24. Lorne ------------ Category 5, Topic 10 Message 165 Sun Feb 07, 1993 K.VANDELLEN [Ken Van] at 16:28 EST David, Nice to hear from you again. How are things on Maui? The Migraph man, Paul, said they charge for major upgrades, but will send minor upgrades out to people who have problems or request them. I'm not sure what the improvements were between 1.10 and 1.24 (that was it?), but you might like to ask for the newer version. I asked about the possiblity of adding a feature to train the program to join split letters, but he said they didn't write that part of the program, so can't do it. I sent him an IMG file, so he can see what's happening. gnox, I don't always get split letters. Some text works fine. Paul asked if the breaks happen where there are no pixels or only one, but the weak places seem thicker than that. I get it with some text with my Wand turned to the deepest contrast, too. Maybe the Wand needs help. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 166 Mon Feb 08, 1993 J.GNIEWKOWSK [GE-Lamp ST] at 22:03 EST David & Ken, I'll be calling Migraph tommorow about the upgrade from 1.1 to 1.24. Let you all know what's up. Sincerly, John E. Gniewkowski Editor, GEnieLamp ST ------------ Category 5, Topic 10 Message 167 Tue Feb 09, 1993 K.VANDELLEN [Ken Van] at 23:26 EST I found a solution to split letters. I scanned my pages, loaded them into TouchUp (one at a time), did a clip to page, and ran BOLD under process. The OCR processing occurred with few errors, although a few letters were still split. It would be easier to make a slightly darker copy on photocopier, and I'm going to try that next. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 168 Sat Feb 13, 1993 J.MOSES2 [moses] at 18:32 EST Thanks, John! I get the newsletter but guess I missed that. Jim Moses ------------ Category 5, Topic 10 Message 169 Wed Mar 17, 1993 C.WORTON at 06:50 EST Hello, all! I'm new to OCR in general, and have a number of questions. First off, what I'm trying to do is scan stock market information out of financial newspapers. (There are services that will download this information directly, but the cost is exorbitant; hand scanning the info, if it can be made to work, is much more cost effective.) Can anyone tell me if this has been tried, and with what degree of success? I had the opportunity to play with the Migraph OCR product, but did not have access to a manual (or knowledgable salesperson). On the test file I brought in, the OCR would repeatedly 'link' characters from a higher line with characters from a lower line. There must be a way to tell the program not to do this, but none of us knew how. I'd like to get some confirmation that this can be done. I have seen some hand scanners in the IBM world that support 800 DPI. Are these available in the ST world, and are they supported by Migraph OCR? Again in the IBM world, it seems to be possible to do multiple scans using a hand scanner, and have all the individual scans pasted together and justified into one seamless whole. Is this feature also available in the ST world? The basic goal is to turn the scanned newspaper data into an ASCII text file; I can then write a program which will format the data to make it more usable for me. Can this be done at all, and if so, is a hand scanner going to be a workable solution - or would a flat bed scanner (more $$) be necessary to get a recognisable image? I appreciate your feedback - thanks! ....Charlie PS: The other use I might have for scanning would be to scan color photographs for use with a digital retouching program. This is not a priority for me, but would fit in the 'gee, it would be nice' category. I imagine that a color scanner, particularly a high res one, is lotsa dollars. Can anyone make recommendations as to current front runners, and cost of such beasts? ------------ Category 5, Topic 10 Message 170 Wed Mar 17, 1993 R.SNYDER6 [Roger] at 20:28 EST Charlie, My first thought on scanning stock info is the problem with errors. I don't think newspapers are the best source for OCR (the paper and printing are not the highest quality) and you always get some errors. With text you can use a spell checker, but with lots of numbers it seems you would have to check them all manually. And the data you are going to use would be important to get right. With all the stock market data offered by on-line services (like here on Genie), it seems to me it would be easer to download it and then import it to your program. (Woops, I see you have considered and rejected this option because of cost. It still might be cheaper in the long run, unless you have lots of free time.) --Roger ------------ Category 5, Topic 10 Message 171 Wed Mar 17, 1993 C.WORTON at 21:05 EST Roger, it's the cost factor; you would not believe the expense. To do daily updates via Dow Jones, it would cost me several hundreds of dollars PER DAY, if I were tracking a few thousand stocks. And to find bargains on a regular basis, you need to track that many. Yes, I can get daily updates on the NYSE through GEnie basic services, but ONLY NYSE quotes; For example, anything on the Chicago trade board or anything outside of NYSE or AMEX is inaccessable. And if I go on holidays, I have to obtain historical quotes through Dow Jones, and fork over that Kings ransom. There's another reason. Sometimes the best bargains are found on the small boards, the ones that don't get traded by the savvy wall street panthers. The Alberta stock exchange comes to mind, and there are many others. While these stock quotes are listed in newspapers, they are frequently unavailable via electronic information vendors. Dow Jones just started making the Toronto Stock Exchange available around two years ago; yet it is the largest, and best traded, exchange in Canada. I'm not at all sure that Dow Jones even has the Vancouver exchange listed, and its a dead sure bet that the Alberta exchange is unavailable. EVERYWHERE. Trust me on this. The ONLY way to show a serious, reliable profit in the market is to obtain a large amount of data CHEAPLY and process it quickly. The bottleneck is getting it into the computer. If OCR won't handle it, I might try voice recognition; but OCR seems the best bet. I don't mind sitting there, correcting every 20th symbol; thats lots better then typing pages of stock quotes in, every day, by hand! I need to determine if OCR will solve this problem. ....Charlie ------------ Category 5, Topic 10 Message 172 Thu Mar 18, 1993 R.SNYDER6 [Roger] at 01:50 EST Charlie, I guess if you are tracking that many, it does get expensive. Well OCR is probably worth a try. I'll try scanning in a stock list, and let you know how it comes out after running it through OCR. I did think of a way to pick up errors. If you are going to run it through you own program, a routine to flag price changes that seem to large would help prevent large errors. And I would guess that small errors would work out in the long run if you entered data frequently enough. --Roger ------------ Category 5, Topic 10 Message 173 Thu Mar 18, 1993 ABC.SOLN at 01:55 EST Charlie, with respect to OCRing stock quotations, forget it. I tried this using Migraph OCR and a hand scanner. I found the type too small for one thing. But this was compounded by the printing process, where ink smears and runs into the next character, and the various and sundry ink drops and smears. Newsprint quality also allows some print from the other side to show through at times. Take a magnifying glass to a stock page and have a good look. OCR needs a good scan of a good quality print in order to succeed. regards Peter ps. I see from a later message that you make frequent references to Canadian exchanges. If you are in the Toronto/Southern Ontario area, CRS has daily trading summaries of all TSE stocks (maybe other exchanges, I haven't checked recently) posted daily in one large zipped file. Cost is $100 annually, with no other download fees. ------------ Category 5, Topic 10 Message 174 Thu Mar 18, 1993 C.WORTON at 02:08 EST Thanks, Peter. I'm in Edmonton, Alberta - about 2,000 miles west of Toronto. Oh, well! The stock quotation source I'm using is the Financial Post. They usually do an better then average job of printing. To eliminate print through from the next page, place the page on something black (non reflective). That will remove the print through problem. When I did some scanned info, I found that the major problem was holding (and moving) the scanner consistently enough. Even a slight angle, tremor, or what-have-you caused skews and glitches in the characters. Roger, I would really appreciate your giving it a try for me. My best results happened when I used the 'line art' position, and the highest resolution - 300 or 400 DPI. The problem was not in having recognisable numbers on the screen, it was in program operation. Again, I'd be really thankful if you could play with it a little bit, when you can. Thanks, Charlie ------------ Category 5, Topic 10 Message 175 Thu Mar 18, 1993 J.MEEHAN3 [>> Joe M << ] at 05:13 EST You might try other general RoundTables, even the IBM RoundTable since your problems will be similar. From my experience I suggest you forgt it. Second since many programs will import data formated the way it is sent to you by GEnie's online service, that is the way to go. >> Joe M << ------------ Category 5, Topic 10 Message 176 Thu Mar 18, 1993 K.VANDELLEN [Ken Van] at 07:38 EST I agree with Roger. I have had fair success with newsprint, but if you have to check all of your data for accuracy, you might as well type them in in the first place. Either way is going to take a lot of time if you're going to input the whole NYSE plus a bunch of other stuff. As I recall, stock listings are smaller print than most of the paper, which would make them more difficult to do. The MiGraph OCR and the Wand go to 400 dpi. I think the hand scanner is still 300 dpi, making tiny print almost impossible to work with, I would think. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 177 Thu Mar 18, 1993 A.FASOLDT [Al Fasoldt] at 21:54 EST Charlie, I created the stock tables that are used at my newspaper (I'm an editor and pagination maven), and I can attest to the truth of Peter's remark about the quality of the type (6 points) in most stock tables. There's another issue, too, and that's whether your time is better spent analyzing these stocks rather than scanning them in. Even if a stock-quotaton service cost a lot of money, if it pays off, you are ahead. (But surely there are ways of getting this info cheaper.) Al ------------ Category 5, Topic 10 Message 178 Thu Mar 18, 1993 C.WORTON at 22:49 EST Al, if you have any suggestions for obtaining the information more inexpensively, I'd welcome them. I've searched for years, and found nothing. ...Charlie ------------ Category 5, Topic 10 Message 179 Fri Mar 19, 1993 J.RENNER1 [Jim] at 01:06 EST Charlie, You'll need a higher DPI scan for the stock market stuff. Although I don't know how you can easily organize it into usable information. That will be the hard part. There is a product called SCAN & MERGE from Migraph for combining scans into one file. The Epsom 600C is a good color scanner and i believe atleast one company has support for it in their software. I believe DMC Publishing (the Calamus folks) have or soon will software to do color scanning. Jim. ------------ Category 5, Topic 10 Message 180 Fri Mar 19, 1993 K.VANDELLEN [Ken Van] at 07:54 EST Whoa! I don't think Charlie needs anything beyond his OCR software to merge text files. He can do that in his text processor after running OCR, and the OCR program will append text files when directed to do so. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 181 Fri Mar 19, 1993 A.FASOLDT [Al Fasoldt] at 18:24 EST If I come across something, I'll let you know, Charlie. Al ------------ Category 5, Topic 10 Message 182 Sat Mar 20, 1993 K.SPRINGER1 [FROZEN NORTH] at 23:35 EST Charlie, Jim has inadvertantly combined the names of 2 programs from Migraph. One is SCAN & SAVE, the other is MERGE IT. Scan and Save is basically an accessory version of the scanning portion of Touch Up, and lets you save greyscales if you wish. Merge It lets you piece together two or more side by side scans into one large scan file. Coalesce from Dr. Bob does the same thing. I have the Migraph products, but have never been able successfully have two scans "seamlessy" join together. They always misalign somewhere. As for the jiggling and the difficulty of scanning straight, there is a company in So. Cal that sells a do-dad that fits over the Migraph hand scanner, and lets you run the scanner down the side of a straight edge provided with the kit so there are no side to side movement problems. I find it works very well. Ken S. ------------ Category 5, Topic 10 Message 183 Sun Mar 21, 1993 G.FUHRMAN [gnox] at 07:34 EST Charlie, What Ken said. I own both Merge-It and Scan & Save and never use either of them any more. Merge-It has been replaced by scanning in landscape mode (OCR) and Megapaint (for merging graphics over 4 in. wide). Scan & Save is replaced by Scanlite (WizWorks). gnox ------------ Category 5, Topic 10 Message 184 Sun Mar 21, 1993 J.MEEHAN3 [>> Joe M << ] at 14:12 EST Follow up to FROZEN NORTH message # 182 I have Dr. Bob's Coalesce and it works well as long as you have a good pair of scans. Using a "Tray" I can reliably get seamless results with the tools in Coalesce. >> Joe M << ------------ Category 5, Topic 10 Message 185 Sun Mar 21, 1993 K.VANDELLEN [Ken Van] at 16:58 EST I also saw a file in this library that has a plan for a scanner guide you can make. I can't tell you any more than that. Sorry! You might do a search. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 186 Sun Mar 21, 1993 J.RENNER1 [Jim] at 22:55 EST My hand is held out where all my slap it. Sorry, i knew that but was "typing challenged", Thanks for clearing that up. Jim. ------------ Category 5, Topic 10 Message 187 Mon Mar 22, 1993 AEO.3 [Lyre] at 22:26 EST BTW, anyone interested in Migraph's OCR Jr, scanners, Merge It and Scan & Save, etc. might want to take a look at the current issue of Atari Explorer Online. Their is information on a sale on these items in the Developing News column. Lyre News Editor, ATARI EXPLORER ONLINE 7:41 pm, March 22, 1993 ------------ Category 5, Topic 10 Message 188 Thu Apr 01, 1993 J.SCHRAM at 04:48 EST Ken, I did several searches for those plans to build a scanning guide and couldn't find it. I searched all libraries for scan, scanner, guide, tray, and maybe some others. Has anyone else found this file or have any more information that may help me locate it. The uploader's address or about how old the file is might help. If it was in a text mag I may never find it... Joe Schram ------------ Category 5, Topic 10 Message 189 Fri Apr 02, 1993 K.VANDELLEN [Ken Van] at 07:59 EST Joe, I thought I ran across that file just the other day. Hmm! If I get a chance I'll see if I can spot it and let you know. Maybe you should run an ad for a used Tray. I sold one a while back when I got The Wand. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 190 Fri Apr 02, 1993 J.SCHRAM at 23:54 EST Ken, I just talked to Toad about getting Coelesce and the tray. They tell me that WizWorks no longer sells it with the tray so they are shipping me Coelesce alone. They did say that they found out where the tray came from and would be selling them seperate and it would run about the same price. I saw a scanning guide in Computer Shopper for about $35 that I think would work. I may take your advice about running an ad but would really like to check out those plans first if possible. How much did you get for your old tray? Joe Schram ------------ Category 5, Topic 10 Message 191 Sat Apr 03, 1993 K.VANDELLEN [Ken Van] at 07:51 EST Joe, Can't remember. Maybe $30. I checked for that file in my archives back to 10/91, and haven't found it. Maybe before that. I forgot what you searched for. Try hand, scan, and guide. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 192 Sun Apr 04, 1993 J.SCHRAM at 00:44 EST Ken, I tried scan, scanner, guide, tray. I didn't try hand so I will try that now... Joe ------------ Category 5, Topic 10 Message 193 Sun Apr 04, 1993 M.VONSEGGER [Marky Mark] at 13:44 EDT I am getting a TT for This Purpose. I myself have had the same problem even with the latest versions out. I think with the TT extended memory may work. ------------ Category 5, Topic 10 Message 194 Sun Apr 04, 1993 K.VANDELLEN [Ken Van] at 15:35 EDT Joe, Good luck, as you continue your search. I'm sure I saw that file here, but I'm beginning to have my doubts, now. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 195 Mon Apr 05, 1993 A.FASOLDT [Al Fasoldt] at 03:47 EDT Marky, :) The TT does not have extended memory. That's a horror in the PC world. Al ------------ Category 5, Topic 10 Message 196 Tue Apr 06, 1993 K.SPRINGER1 [FROZEN NORTH] at 21:33 EDT Someone was talking about scanners and trays. I just got a sale flyer from Tiger Software, and they have the same tray that Migraph and others sold for $40. Ken S. ------------ Category 5, Topic 10 Message 197 Sat Apr 10, 1993 J.SCHRAM at 16:55 EDT I ordered a Scan Guide from O.S. Computer City today for $29 + $8 shipping. The guy checked to make sure it would work on a Migraph and he said it was listed. It comes with tray, scan pad, trackers, attachments, and accessories for many hand scanners. I should have it in about a week, about the same time I will get Coelesce from Toad... Joe ------------ Category 5, Topic 10 Message 198 Mon Apr 12, 1993 K.VANDELLEN [Ken Van] at 18:08 EDT I received a phone call from Paul at Migraph today in response to sample files and printouts I sent him. (I think he's the co-author of OCR.) I had several questions. My IMG had some vertical lines on it, which he said could be caused by dust on the scanner. He suggested blowing it off before using it. I was having trouble with certain letters showing up split, such as h, d, p, etc. These were rather thin where the two parts were connected, and were being interpreted as two letters. I've found the reworking the IMG by bolding it with TouchUp helped some. Perhaps a darker photostat would do the same thing. While we were talking, I mentioned that I had gone back to using EZ Draw, because it does some things no other drawing program I have will do, such as grids. He said they had been well along the way to an upgrade of it, but dropped it because of the poor Atari market. Let's hope the Falcon takes off and improves the user-base. Maybe they'll finish the upgrade. There doesn't seem to be an EZ Draw topic, so I thought I'd put the news here. The good news is that Migraph is continuing to support the Atari community. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 199 Sat Apr 17, 1993 J.SCHRAM at 00:50 EDT I didn't get my tray yet but I did get Coalesce ver 1.42 today. I thought it was an ACC that would work from within OCR. Is it strictly a stand alone program? Do I need to save the merged document and then load it into OCR? Joe ------------ Category 5, Topic 10 Message 200 Sat Apr 17, 1993 K.VANDELLEN [Ken Van] at 21:20 EDT Joe, As I recall, the answer to both questions is yes. Ken Van Dellen d8^) ------------ Category 5, Topic 10 Message 201 Sun Apr 18, 1993 J.SCHRAM at 02:54 EDT Thanks Ken, I can live with that. It would be nice to merge from within OCR but I guess even with 4 megs that would be stretching it. Some of those IMG files get pretty big... Joe Schram ------------