22 Nov
2006
22 Nov
'06
10:48 a.m.
My recent post re BBM manuals has left me with a bit of a
dilemma; I've got several more boxes of various manuals etc.
to dispose of and I'm not sure what's the best way.
On the one hand, it benefits everyone to have them in Al's
hands, who has the means, time and willingness (thanks, Al)
to make them available to us all.
On the other hand, anyone who actually owns the relevant
equipment would probably like to have original documentation.
When I put them up on the list, I'm inevitably going to disappoint
someone; should I just offer it privately to Al and rely on him, or?
Comments?
mike
22 Nov
22 Nov
11:03 a.m.
In article <01C70E3D.1427E020 at mse-d03>,
M H Stein <dm561 at torfree.net> writes:
When I put them up on the list, I'm inevitably
going to disappoint
someone; should I just offer it privately to Al and rely on him, or?
Al isn't the only one who scans in old docs and makes them available.
I've done that for a bunch of stuff I recently acquired and I have
some more stuff in the pipeline.
So just because something doesn't go directly to Al doesn't mean it
won't end up scanned and archived on bitsavers.
You could always add the condition that if someone else besides Al
takes ownership of the manual that they have to scan it and get it on
the net for bitsavers.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
12:01 p.m.
My recent post re BBM manuals has left me with a bit
of a
dilemma; I've got several more boxes of various manuals etc.
to dispose of and I'm not sure what's the best way.
On the one hand, it benefits everyone to have them in Al's
hands, who has the means, time and willingness (thanks, Al)
to make them available to us all.
On the other hand, anyone who actually owns the relevant
equipment would probably like to have original documentation.
I have Burroughs equipment, but still I think it should go to
bitsavers. Hey, if I want to, I can then try to get them from Al after
scanning, if I want. While others have also scanned (and we thank
you), bitsavers is still the best archive, and probably has the best
chance of long term survival.
--
Will
12:11 p.m.
In article <e1d20d630611221201u6d562f0bm1616052b60ea5292 at mail.gmail.com>,
"William Donzelli" <wdonzelli at gmail.com> writes:
[...] While others have also scanned (and we thank
you), bitsavers is still the best archive, and probably has the best
chance of long term survival.
I always email Al with a URL of the scanned PDFs for uploading to
bitsavers after I scan stuff. I also email the guy who maintains manx
(I forget his name right now) so that they can be found online through
that database.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
12:26 p.m.
William wrote....
> [...] While others have also scanned (and we
thank
> you), bitsavers is still the best archive, and probably has the best
> chance of long term survival.
To which Richard replied...
I always email Al with a URL of the scanned PDFs for
uploading to
bitsavers after I scan stuff. I also email the guy who maintains manx
(I forget his name right now) so that they can be found online through
that database.
I don't want to detract from what others besides Bitsavers have done. Any
documentation made available online is a huge plus, a great benefit to all,
and is sincerely appreciated. Many thanks to all those who have done it in
addition to bitsavers.
However, that being said... there is much more to the historical archiving
of a document for posterity than slapping it on a scanner and putting it on
the web. Al does not simply do that. He has specific expertise in the
correct and best way to scan the documents (with regards to settings and
such), uses other utilities to process the images before converting to PDF,
in many cases hand adjusting them, etc. It is an art, not a "press a scanner
button" thing, and Al has it down cold.
So long story short, emailing Al with a URL of the scanned PDF is not the
best situation (although it's certainly appreciated). The best situation is
to get the original to Al so he can scan it himself and thus we all get
museum quality results.
Just my own two millidollars worth.
Jay
12:55 p.m.
In article <00f601c70e74$8bfbaaf0$6700a8c0 at BILLING>,
"Jay West" <jwest at classiccmp.org> writes:
However, that being said... there is much more to the
historical archiving
of a document for posterity than slapping it on a scanner and putting it on
the web. Al does not simply do that. He has specific expertise in the
correct and best way to scan the documents (with regards to settings and
such), uses other utilities to process the images before converting to PDF,
in many cases hand adjusting them, etc. It is an art, not a "press a scanner
button" thing, and Al has it down cold.
While I agree with the sentiment that scanning old documents isn't
something that can be done nonchalantly, I disagree with the
implication that Al is the only one who has the Special Knowledge to
scan old documents into a form worthy of bitsavers.
Being someone who has worked in graphics, I'm probably more aware than
most of how to get the best results out of a particular original and a
particular scanner.
So long story short, emailing Al with a URL of the
scanned PDF is not the
best situation (although it's certainly appreciated). The best situation is
to get the original to Al so he can scan it himself and thus we all get
museum quality results.
The assumption here, as above, is that only Al has the Special
Knowledge of how to scan documents. Al has many things on his "to do"
list. Certainly its better for me to scan using the same settings and
techniques that Al would use and make the end result available to him
than to make him the bottleneck for all document scanning.
ISTR that we discussed document scanning on this list and Al told us
the kinds of things that he does when scanning. I don't recall
anything magical, either in equipment or techniques, that made Al's
scanning methods beyond what I could do myself. I do recall that some
of the things he mentioned might not be obvious to someone who hasn't
scanned documents that are ultimately intended for OCR at some point,
but then again not all of us are complete dunderheads when it comes to
document scanning either.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
1:43 p.m.
Richard wrote....
While I agree with the sentiment that scanning old
documents isn't
something that can be done nonchalantly, I disagree with the
implication that Al is the only one who has the Special Knowledge to
scan old documents into a form worthy of bitsavers.
I did not say Al has unique
Special Knowledge (although he may, it's
certainly not MY area of expertise to know one way or the other). I did
imply that Al has Special Experience, which some others certainly may have
as well. I did not say his knowledge or experience were necessarily unique.
However, what I did imply (and in regard to your last phrase above) - is
that bitsavers is his, and HE decides what is acceptable for documents to be
included in the archive. And if he feels that the best result is scanning it
himself, that's his call. If he wishes to accept scans from others, that's
his call too. I'm all behind whatever he wishes to do. The quality (and
ubiquity) of his work speak loudly enough that I certainly don't dare
question his preferences :) Since Al is the only one who can actually put a
document into bitsavers, I suspect it is his call ;)
I definitely don't think bitsavers should be the only archive people
contribute to! But it is by far (IMO) the best archive according to most any
metric chosen. And with it mirrored on approximately 9 servers in
approximately 3 countries it's got a great chance of longevity. As I did in
my previous email, I praise other archives and the work they have done, and
don't feel it necessary to repeat here. My own personal decision is when I
get a spare manual no one has, send it to Al. Everyone should make their own
decisions of course.
but then again not all of us are complete dunderheads
when it comes to
document scanning either.
Not all of us are, nor did I say that. I certainly am
however; graphics is
blackmagic voodoo to me.
Jay West
2:08 p.m.
In article <002001c70e7f$3cf43020$6700a8c0 at BILLING>,
"Jay West" <jwest at classiccmp.org> writes:
[...] I did
imply that Al has Special Experience, which some others certainly may have
as well.
Well, we discussed it on the list a while back (I think I was even the
one bringing it up) and to the best of my knowledge the scans I've
been providing are up to Al's standards, since they have been
incorporated into bitsavers.
I even wrote up a web page with my advice on scanning so that others
can create scans that match the same quality of bitsavers.
<http://www.xmission.com/~legalize/vintage/docs.html>
In general, I think its a bad idea to have everything in any community
funnel through a single person. What happens if Al (knock on wood)
should get injured and can't do scans anymore? If we adopted the
"well, just let Al scan everything because he knows how to do it" and
it never gets documented on what is the best way to scan documents,
then we lose everything connected to document scanning when we lose
Al.
Its the same reason I don't really like the KB thing -- everything is
going through one person and no matter how devoted that one person is,
they will always end up being the bottleneck for someone else.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
4:48 p.m.
Richard wrote.....
Its the same reason I don't really like the KB
thing -- everything is
going through one person and no matter how devoted that one person is,
they will always end up being the bottleneck for someone else.
But yet you still contribute to bitsavers even though everything is going
through one person there. Awesome! I'm hoping you'll have the same
participation in the KnowledgeBase ;)
Jay West
5:04 p.m.
Richard wrote:
Well, we discussed it on the list a while back (I think
I was even the
one bringing it up) and to the best of my knowledge the scans I've
been providing are up to Al's standards, since they have been
incorporated into bitsavers.
I think that the only reason Al does as much scanning as
he does, is because he is willing to take the time to do it.
I'll bet that he would be thrilled to get documents in
a format ready to drop on his site.
I even wrote up a web page with my advice on scanning
so that others
can create scans that match the same quality of bitsavers.
<http://www.xmission.com/~legalize/vintage/docs.html>
In general, I think its a bad idea to have everything in any community
funnel through a single person. What happens if Al (knock on wood)
should get injured and can't do scans anymore? If we adopted the
Isn't the official prediction of injury "run over by a bus"?
It is from the same standard units as a "Library of Congress",
a "Station Wagon full of tapes", a "metric buttload", and
a "Human Hair".
"well, just let Al scan everything because he
knows how to do it" and
it never gets documented on what is the best way to scan documents,
then we lose everything connected to document scanning when we lose
Al.
Its the same reason I don't really like the KB
thing -- everything is
going through one person and no matter how devoted that one person is,
they will always end up being the bottleneck for someone else.
2:13 p.m.
Jay wrote:
...
I definitely don't think bitsavers should be the only archive people
contribute to! But it is by far (IMO) the best archive according to most any
metric chosen. And with it mirrored on approximately 9 servers in
approximately 3 countries it's got a great chance of longevity. As I did in
my previous email, I praise other archives and the work they have done, and
don't feel it necessary to repeat here. My own personal decision is when I
get a spare manual no one has, send it to Al. Everyone should make their own
decisions of course.
Scanning the document is most of the work in wall-time clock seconds.
My scans are at 600 dpi, 8-bit grey scale, and thus the "PDF" files are
huge (actually they are TIFF with a PDF "jacket" to make them easy to read).
I transfer them to Al, and Al does post-processing, squeezing them down
to approx 30-40% of the original size. See for example the XXDP listings.
I scanned all of those. You don't want to know the original file size!
A reasonable scanner will give results that are always better for somebody
else who does not have the document. Any documentation, regardless the
quality is better than no documentation at all!
my 2 ct, Henk.
PS. Al, if you read this, it would be great if you get an ftp server up
and running again...
This message and attachment(s) are intended solely for the use of the addressee and may
contain information that is privileged, confidential or otherwise exempt from disclosure
under applicable law.
If you are not the intended recipient or agent thereof responsible for delivering this
message to the intended recipient, you are hereby notified that any dissemination,
distribution, or copying of this communication is strictly prohibited.
If you have received this communication in error, please notify the sender immediately by
telephone and with a "reply" message.
Thank you for your cooperation.
but then again not all of us are complete
dunderheads when it comes to
document scanning either.
Not all of us are, nor did I say that. I certainly am
however; graphics is
blackmagic voodoo to me.
Jay West 
12:11 p.m.
Gooijen, Henk wrote:
Scanning the document is most of the work in wall-time
clock seconds.
My scans are at 600 dpi, 8-bit grey scale, and thus the "PDF" files are
huge (actually they are TIFF with a PDF "jacket" to make them easy to read).
I transfer them to Al, and Al does post-processing, squeezing them down
to approx 30-40% of the original size. See for example the XXDP listings.
I scanned all of those. You don't want to know the original file size!
The question does keep on sticking in my mind though of whether anyone's
worrying about scanning rarer documents at a sufficient quality / resolution
to be as close as possible to the original.
For some documents there are very few surviving copies, and it would seem
sensible to preserve those ones now in a form that was as close as reasonably
possible to the original - colour scans where colour exists in the document
(or the paper is non-white or the text non-black I suppose), greyscale for
text rather than bi-level, sufficient resolution for photos and diagrams etc.
A reasonable scanner will give results that are always
better for somebody
else who does not have the document. Any documentation, regardless the
quality is better than no documentation at all!
Oh, absolutely. Getting the raw information saved is definitely a priority -
but with storage costs getting cheaper all the time, maybe we're at a point
where a coordinated effort could be made to preserve the really rare items in
a form a little closer to the original?
cheers
Jules
27 Nov
27 Nov
4:30 p.m.
In article <4564AEE4.1040709 at yahoo.co.uk>,
Jules Richardson <julesrichardsonuk at yahoo.co.uk> writes:
For some documents there are very few surviving
copies, and it would seem
sensible to preserve those ones now in a form that was as close as reasonably
possible to the original - colour scans where colour exists in the document
(or the paper is non-white or the text non-black I suppose), greyscale for
text rather than bi-level, sufficient resolution for photos and diagrams etc.
I scan this way -- I use color or grayscale where necessary and use
bi-level on text only B&W pages. bi-level also works for diagrams
that consist only of line art without fine detail.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
29 Nov
29 Nov
1:11 a.m.
Richard wrote:
In article <4564AEE4.1040709 at yahoo.co.uk>,
Jules Richardson <julesrichardsonuk at yahoo.co.uk> writes:
It's OK if you have top-quality documentation. But lots of computer docs out
there are old, faded, dirty, creased, well-thumbed etc. and unless someone's
prepared to visually check every scanned page, there's a chance that the
bi-level algorithm in use will corrupt the data and it'll go unnoticed.
If using greyscale then such artifacts can be dealt with at a later date (i.e.
at OCR time) rather than at scan time. Of course, OCR is by its nature slow as
the data has to be messed with at input anyway and checked upon output, so
some extra tweaking there isn't so bad - but it's useful if the initial scan
process can be as quick as possible (it's time-consuming enough as it is!)
I suppose it's one of those situations where you end up throwing away
information no matter what (after all, any scan is essentially a digital
representation of analogue data), but there's a danger of throwing away too
much data - and for rare docs you might only get the chance to scan them once.
For rare items I'd rather have maximum quality "just in case", even if it
does
mean more storage space.
cheers
Jules
For some documents there are very few surviving
copies, and it would seem
sensible to preserve those ones now in a form that was as close as reasonably
possible to the original - colour scans where colour exists in the document
(or the paper is non-white or the text non-black I suppose), greyscale for
text rather than bi-level, sufficient resolution for photos and diagrams etc.
I scan this way -- I use color or grayscale where necessary and use
bi-level on text only B&W pages. bi-level also works for diagrams
that consist only of line art without fine detail. 
6:26 a.m.
Jules Richardson wrote:
I suppose it's one of those situations where you
end up throwing away
information no matter what (after all, any scan is essentially a digital
representation of analogue data), but there's a danger of throwing away
too much data - and for rare docs you might only get the chance to scan
them once. For rare items I'd rather have maximum quality "just in
case", even if it does mean more storage space.
Exactly. Storage is incredibly cheap these days. It's easy to throw
disk space at a problem.
If you had to fit all the scans onto RL02 packs, I'd see your problem...
Gordon
30 Nov
30 Nov
6:40 a.m.
Gordon JC Pearce wrote:
Jules Richardson wrote:
Oh, I don't think it's *that* easy quite yet - think of something like
bitsavers if it suddenly needed a 10x increase in storage and bandwidth :-)
But I do think it's possible for 'special cases' where there may only be a
handful of surviving printed copies left in the world.
So far as I'm aware it's not an issue that any of the big institutions are
thinking about though as (rightly so) the goal in recent years has just been
to preserve the text for human readability rather than addressing quality
issues. Better to do something than nothing at all, and all that. Things are
probably at a stage now where the recommendations for special cases can be
different, though [1].
[1] Docs from the 50s and early 60s are often good candidates - typically
they're bound in something other than paper/card, printed on very off-white
paper, and by now are getting rather dirty. A typical bi-level scan not only
loses the spirit of the original somehow, but introduces major headaches for
subsequent OCR (like OCR isn't enough of a headache already!)
I suppose it's one of those situations where
you end up throwing away
information no matter what (after all, any scan is essentially a
digital representation of analogue data), but there's a danger of
throwing away too much data - and for rare docs you might only get the
chance to scan them once. For rare items I'd rather have maximum
quality "just in case", even if it does mean more storage space.
Exactly. Storage is incredibly cheap these days. It's easy to throw
disk space at a problem. If you had to fit all the scans onto RL02 packs,
I'd see your problem...
:-) What's the transfer rate for a typical RL02 drive? I suspect that just
getting the data on and off would be a problem, even if you had enough of them
to meet capacity...
(there's a fun calculation - what would the power consumption of the bitsavers
archive be if it was held on RL02 packs?)
cheers
Jules
29 Nov
29 Nov
1:41 p.m.
Jules Richardson wrote:
It's OK if you have top-quality documentation. But
lots of computer
docs out there are old, faded, dirty, creased, well-thumbed etc. and
unless someone's prepared to visually check every scanned page,
there's a chance that the bi-level algorithm in use will corrupt the
data and it'll go unnoticed.
Well I do look at each page in the final scan - pretty briefly just
to make sure that all the pages are in the correct order. Even so
I've never noticed a problem. I've never tried scanning a line-printer
listing and most stuff I scan is a reasonable manual, but even on
the few scans I've made of photocopies, I've never seen a problem.
It would have to be a gross problem to affect legibility of text
by a human (almost anything affects OCR :-( ).
I suppose it's one of those situations where you
end up throwing away
information no matter what (after all, any scan is essentially a
digital representation of analogue data), but there's a danger of
throwing away too much data - and for rare docs you might only get
the chance to scan them once. For rare items I'd rather have maximum
quality "just in case", even if it does mean more storage space.
I scan at 600dpi bi-level G4-encoded and I end up with some manuals
in the hundreds of MB. I think at least one of the Digital Semi
manuals hits nearly 400MB. If I went to greyscale (8 bits) then
some of those manuals would take half a DVD[1] to store. Perhaps
a few years from now I'll come back and look for some way of
erasing this post ("2GB, that's nothing, I can downlaod that
in 10 mins!!") but right now 300MB is manageable (even as a download)
whereas 2GB is a stretch.
For my first few manuals I did redo pages in greyscale if they
had a photo but I stopped because I could not see a difference.
But by all means scan in 8-bits, just don't scan text as a JPEG
unless the alternative is not scanning at all :-)
Antonio
[1] Maybe more - I think G4 only works on bi-level, so by using
greyscale I think you lose some lossless compression too.
3:13 p.m.
In article <002c01c713ff$1105ced0$4804010a at uatempname>,
<arcarlini at iee.org> writes:
For my first few manuals I did redo pages in greyscale
if they
had a photo but I stopped because I could not see a difference.
I noticed a difference on the ones I scanned. I just scan /pages/ in
grayscale when that page contains a photo. I don't scan the whole
manual in grayscale just because there's the occasional photo.
How the source material was printed makes a big difference in how you
should scan it.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
3:10 p.m.
In article <456D4EB3.7000106 at yahoo.co.uk>,
Jules Richardson <julesrichardsonuk at yahoo.co.uk> writes:
It's OK if you have top-quality documentation. But
lots of computer docs out
there are old, faded, dirty, creased, well-thumbed etc. and unless someone's
prepared to visually check every scanned page, there's a chance that the
bi-level algorithm in use will corrupt the data and it'll go unnoticed.
I check every scanned page as I scan it. You have to anyway, because
if it doesn't scan right you have to rescan it to get it right.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
30 Nov
30 Nov
6:27 a.m.
Richard wrote:
In article <456D4EB3.7000106 at yahoo.co.uk>,
Jules Richardson <julesrichardsonuk at yahoo.co.uk> writes:
The main problem I find with that is that it's time-consuming to check every
page at the scanned resolution (i.e. 1:1 mapping between on-screen pixels and
scanned dots). However at a typical "fit page to window" zoom level it's
easy
to make sure that the page was scanned straight etc., but easy to miss things
which might hinder some future OCR process.
No process is going to be perfect, of course, but we are maybe at a point in
terms of storage availability and transmission speed that we can handle a
quality improvement for the really hard to find stuff.
cheers
Jules
It's OK if you have top-quality
documentation. But lots of computer docs out
there are old, faded, dirty, creased, well-thumbed etc. and unless someone's
prepared to visually check every scanned page, there's a chance that the
bi-level algorithm in use will corrupt the data and it'll go unnoticed.
I check every scanned page as I scan it. You have to anyway, because
if it doesn't scan right you have to rescan it to get it right.
7:16 a.m.
In article <456EEA43.7050601 at yahoo.co.uk>,
Jules Richardson <julesrichardsonuk at yahoo.co.uk> writes:
[...] it's easy
to make sure that the page was scanned straight etc., but easy to miss things
which might hinder some future OCR process.
To be honest, almost every time I have tried to OCR something (even a
pristine original), it was simply faster and more accurate to type it
in myself. I don't know why but I have been singularly unimpressed
with OCR software. Obviously lots of people do OCR, but the amount of
rework and editing necessary to get high accuracy is just as much work
as typing it in yourself for someone like me that is a fast touch
typer.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
8:12 a.m.
On Thu, 30 Nov 2006 08:16:14 -0700, Richard wrote:
.... Obviously lots of people do OCR, but the amount
of
rework and editing necessary to get high accuracy is just as much work
as typing it in yourself for someone like me that is a fast touch
typer.
For the followers of the Columbus School of Typeing which promots the Discover and Land
three finger approach to communicateing
using a keyboard, a few trackball or mouse strokes combined with well timed button
flicking is easier.
I like the idea of the Dragon voice editing, but have not taken the time to master it
...... yet
Just an alternate thought
Bob Bradlee
9:22 a.m.
In article <200611301612.kAUGCARD075057 at keith.ezwind.net>,
"Bob" <caveguy at sbcglobal.net> writes:
For the followers of the Columbus School of Typeing
which promots the
Discove r and Land three finger approach to communicateing > using a
keyboard, a few trackball or mouse strokes combined with well timed b
utton flicking is easier.
Whatever works for you is a good solution :).
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
11:35 a.m.
On Thu, 30 Nov 2006 08:16:14 -0700, Richard wrote:
Discover and
Land three finger approach to communicateing
.... Obviously lots of people do OCR, but the
amount of
rework and editing necessary to get high accuracy is just as much work
as typing it in yourself for someone like me that is a fast touch
typer.
For the followers of the Columbus School of Typeing which promots the using a keyboard, a few trackball or mouse strokes
combined with well
timed button flicking is easier.
I like the idea of the Dragon voice editing, but have not taken the time
to master
it ...... yet
Just an alternate thought
Bob Bradlee
I dictated / edited a 50 page User's Manual for a Military COMM Unit with
excellent results ( 90+ % ) using Dragon. Beats the hell out of Voice Pad
and some of the other crap out there. I do miss the days of paying my kids
$1 a page for typing though...
regards, SAC
8:12 a.m.
Richard wrote:
In article <456EEA43.7050601 at yahoo.co.uk>,
Jules Richardson <julesrichardsonuk at yahoo.co.uk> writes:
Oh, I agree. Twenty years down the line I expect it'll be a lot better though,
but by then the original paper copies of some of the material out there
might be long-gone - hence my concern about improving the quality of some scans.
I suppose a vague rule of thumb might be that if it's not readable by a human
then it's never going to be readable via OCR :-) Thing is, to maximise
chances, every single letter in every single scan would have to be proof-read
for legibility - which is obviously unrealistic.
Hence my feeling that bi-level just isn't good enough for some docs, because
it won't necessarily discriminate between real text and a hair / dirt / pen
mark where greyscale *might*. It's not infallible either of course - a blue
biro mark might be indistinguishable from the faded text below it after
scanning; give it five years and I'll probably be advocating full-colour scans :-)
cheers
Jules
[...] it's easy
to make sure that the page was scanned straight etc., but easy to miss things
which might hinder some future OCR process.
To be honest, almost every time I have tried to OCR something (even a
pristine original), it was simply faster and more accurate to type it
in myself. I don't know why but I have been singularly unimpressed
with OCR software. Obviously lots of people do OCR, but the amount of
rework and editing necessary to get high accuracy is just as much work
as typing it in yourself for someone like me that is a fast touch
typer.
9:25 a.m.
In article <456F030B.4080202 at yahoo.co.uk>,
Jules Richardson <julesrichardsonuk at yahoo.co.uk> writes:
Richard wrote:
Ahem. That's what they said 20 years ago! :-) Yeah, it is better now
than it was, but its still pretty frustrating to use, particularly
when compared against the hype of the products.
I think the worst is when you have an original that is several
generations of 1970s or 1980s copier away from the real document.
Then there's the aspect of printer output where the character
baselines aren't lined up on the printout, like old line printer
listings and stuff.
To be honest, almost every time I have tried to
OCR something (even a
pristine original), it was simply faster and more accurate to type it
in myself. [...]
Oh, I agree. Twenty years down the line I expect it'll be a lot better though Hence my feeling that bi-level just isn't good
enough for some docs, because
it won't necessarily discriminate between real text and a hair / dirt / pen
mark where greyscale *might*. It's not infallible either of course - a blue
biro mark might be indistinguishable from the faded text below it after
scanning; give it five years and I'll probably be advocating full-colour scan
s
:-)
If I was scanning a low quality ancient original, then yeah, I'd
probably do grayscale to capture as much detail as possible for my
"archive" original and then put out a bi-level PDF for download.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
4:11 p.m.
Jules Richardson wrote:
scanning; give it five years and I'll probably be
advocating
full-colour scans :-)
The RSX & RT-11 manuals I scanned had various pages with colour -
for RSX-11M & RSX-11MPLUS it matters because various sections
only apply under certain conditions and colour is used to
indicate that. So there's B&W, red text, blue text, pink
background, grey background and probably blue background too.
I scanned the entire manual in B&W on one scanner and then passed
the colour pages through a colour scanner: 600dpi @ 24-bit colour.
I know I should be able to get those pages down to a few 100KB
each but currently they're all still 24MB or so. I'm guessing
a decade before you'll be advocating full colour scans - which
is probably how long it will take me to find some way of filtering,
compressing and layering the scans to produce something
reasonable :-)
Antonio
4:40 p.m.
In article <004001c714dd$415adbf0$4804010a at uatempname>,
<arcarlini at iee.org> writes:
The RSX & RT-11 manuals I scanned had various
pages with colour -
for RSX-11M & RSX-11MPLUS it matters because various sections
only apply under certain conditions and colour is used to
indicate that. So there's B&W, red text, blue text, pink
background, grey background and probably blue background too.
With a little care, you can capture this accurately and efficiently by
quantizing the truecolor image to a carefully designed color palette.
Then you can store it as 8-bit per pixel (or eve 4-bit per pixel,
depending) and squish it down considerably.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
1 Dec
1 Dec
1:36 a.m.
arcarlini at iee.org wrote:
The RSX & RT-11 manuals I scanned had various
pages with colour -
for RSX-11M & RSX-11MPLUS it matters because various sections
only apply under certain conditions and colour is used to
indicate that.
I scanned the entire manual in B&W on one scanner
and then passed
the colour pages through a colour scanner: 600dpi @ 24-bit colour.
I know I should be able to get those pages down to a few 100KB
each but currently they're all still 24MB or so.
I have post-processed some of those manuals to split the colours into
layers, convert them to 1bpp and reassemble. This can make a spectacular
difference to file size.
For example, the original colour scan of AA-5286G-TC was 117 MiB. By
splitting the black and red into layers, the size has come down to 7.9 MiB.
I perform the image processing with a netpbm chain, and then assemble
the PDF with a Perl script, using PDF::API2.
--
Paul
9:19 a.m.
In article <456FF787.2060201 at frixxon.co.uk>,
Paul Williams <paul at frixxon.co.uk> writes:
I have post-processed some of those manuals to split
the colours into
layers, convert them to 1bpp and reassemble. This can make a spectacular
difference to file size.
I have some DEC manuals that have this black & red text thing going
on. What's the netpbm chain you used to split into 1bpp red and black
layers? That's exactly what I was planning on doing, but its handy to
leverage someone else's work in perfecting the command line :-).
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
30 Nov
30 Nov
8:55 a.m.
On Nov 30, 2006, at 10:16 AM, Richard wrote:
To be honest, almost every time I have tried to OCR
something (even a
pristine original), it was simply faster and more accurate to type it
in myself. I don't know why but I have been singularly unimpressed
with OCR software. Obviously lots of people do OCR, but the amount of
rework and editing necessary to get high accuracy is just as much work
as typing it in yourself for someone like me that is a fast touch
typer.
I just did exactly that myself a few weeks ago...I wound up typing
in an entire very fuzzy hex dump of a 4KB ROM. You wanna talk about
frustrating!
-Dave
--
Dave McGuire
Cape Coral, FL
9:26 a.m.
In article <3C4E741A-F12C-410C-9493-73B5FD7702D1 at neurotica.com>,
Dave McGuire <mcguire at neurotica.com> writes:
I just did exactly that myself a few weeks ago...I
wound up typing
in an entire very fuzzy hex dump of a 4KB ROM. You wanna talk about
frustrating!
Now that gives me an idea -- if you know that the text in question is
a HEX dump, then you ought to be able to optimize the OCR by
restricting its set of choices. However, I really doubt that OCR of
hex dumps is enough of a market to attract the attention of OCR
software developers to special case it for us :-).
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
11:29 a.m.
On Nov 30, 2006, at 12:26 PM, Richard wrote:
Yes, that was exactly my problem. *I* knew the only characters on
the page were 0-9 A-F, in pairs, but I had no way to tell the
software that. Eventually I said "screw it" and typed it in.
-Dave
--
Dave McGuire
Cape Coral, FL
I just did
exactly that myself a few weeks ago...I wound up typing
in an entire very fuzzy hex dump of a 4KB ROM. You wanna talk about
frustrating!
Now that gives me an idea -- if you know that the text in question is
a HEX dump, then you ought to be able to optimize the OCR by
restricting its set of choices. However, I really doubt that OCR of
hex dumps is enough of a market to attract the attention of OCR
software developers to special case it for us :-).
6364
days inactive
6373
days old
32 comments
13 participants
participants (13)
-
arcarlini＠iee.org -
cannings＠earthlink.net -
caveguy＠sbcglobal.net -
dm561＠torfree.net -
gordon＠gjcp.net -
henk.gooijen＠oce.com -
julesrichardsonuk＠yahoo.co.uk -
jwest＠classiccmp.org -
kth＠srv.net -
legalize＠xmission.com -
mcguire＠neurotica.com -
paul＠frixxon.co.uk -
wdonzelli＠gmail.com