jagged angled lines

[Dewdman42]

I have a question that I hope someone here can answer. I have an older
postscript printer. It will be replaced soon, but I want to try to get
to the bottom of why it prints jagged angled lines, but when I create a
bitmap of the same angled line and print that, the printer displays no
jagged lines at all.

- The printer is HP 4ML (Postscript level 2), only 2MB of ram(whoo hoo)

- I have been having a lot of problems printing with the postscript
driver on this printer, presumably because its so old. Any PDF will
almost certainly spew errors to the page. Consequently, I am using the
non-postscript driver...I know this is postscript group...but you still
may have a handle on why this is happening, seems like it must be a bug
in the RIP firmware inside the printer.

- The main program I'm using is Finale (music typesetting)

When I print directly from Finale, or from a PDF that was created with
Finale, the angled lines have severe jagginess to them. Way worse than
what 300dpi should be able to do.

If I create a high resolution bitmap of the Finale document, the
jagginess is not there, and when I print this bitmp, the printer prints
the angled line with no jagginess visible to my aging eyes.

My theory is that there must be a bug or lame aspect of my old
printer's RIP capabilities such that this is how it renders angled
lines. Am I wrong to say this?

If I am right about this, is there any way that anyone knows that I can
get a RIP driver that will do this on the computer instead of in the
printer, but automatically whenever I print something? Perhaps
something that can convert postscript to a bitmap at the desire
resolution.before sending to the printer?

I am probably going to get a new printer anyway, but I'm just trying to
understand the issue completely to make sure its not a Finale problem
or a problem with Windows XP or some other issue like this.

Thanks in advance!

 

[Aandi Inston]

- I have been having a lot of problems printing with the postscript
driver on this printer, presumably because its so old. Any PDF will
almost certainly spew errors to the page. Consequently, I am using the
non-postscript driver...I know this is postscript group...but you still
may have a handle on why this is happening, seems like it must be a bug
in the RIP firmware inside the printer.

Sounds like lack of RAM. Some PostScript printers are quite smart
about this, dynamically reducing resolution until things fit in RAM.

If I create a high resolution bitmap of the Finale document, the
jagginess is not there, and when I print this bitmp, the printer prints
the angled line with no jagginess visible to my aging eyes.

Not really a fair comparison. Printing a PDF will put a much higher
demand on RAM, such as for storing fonts AS WELL AS the page bitmaps.

 

[Aandi Inston]

I have a question that I hope someone here can answer. I have an older
postscript printer. It will be replaced soon, but I want to try to get
to the bottom of why it prints jagged angled lines, but when I create a
bitmap of the same angled line and print that, the printer displays no
jagged lines at all.

Oh, and don't underestimate how jaggy 300 dpi is. A bitmap is likely
to contain some smoothing/antialiasing so the comparison may not be
equal.

 

[Dewdman42]

I hear what you're saying, but this is really way worse than what
300dpi is capable of. Its nearly as bad as what I see on screen at
72dpi. When I print what must be vector based material, the angled
lines are severely stair stepped, but if I render a bitmap first..it
looks much more correct(does not look antialiased).

At any rate, does anyone know of software RIP solutions besides
manually rendering a bitmap file each time I want to print?

I hear ya though, I'm probably going to get a 1200x1200 dpi printer
pretty soon and maybe that will end the whole problem. I just want to
make sure I understand what is gonig on here.

 

[Dewdman42]

Digging around a little bit. It looks like ghostscript may be the
thing I want to use, together with RedMon to provide an emulated
postscript printer where the RIP happens on my computer instead of in
the printer. Seems reasonable even though the HP 4ML does not appear
to be one of the directly supported printers. Hopefully Ghostscript
rips completely to a bitmap raster rather than convert to PCL commands
for HP laser printers it already knows about.

 

[Edward Reid]

Based on your description, this does *not* sound to me like a printer
problem. A PostScript level 1 printer should have no problems with Finale
output, and 2MB of RAM is plenty for a 300dpi B&W printer unless you're
doing something pretty unusual (like placing a large backgournd image --
and I don't even know if Finale can do that). A 15-year-old LaserWriter
should work fine.

I thus strongly suspect some configuration problem on your computer. Since
you say it looks like 72dpi, then it probably is -- something is resulting
in printing a bitmap image of your screen rather than the PostScript
output. I'm sure this could be due to lots of things, but I don't know
what. But if I'm right, then you might have the same problem with a new
printer.

Try printing to a file and see if you can then print the resulting
PostScript file. Or look at the file with a text editor -- if the selected
printer was a PostScript printer, then the file should obviously contain
PostScript code. If you don't know what PS code looks like, look for a line
near the start that contains "%!PS-Adobe-3.0".

Try installing Finale on a different computer and printing to the same
printer. Do you get the same problem? If not, then try reinstalling your
printer drivers. Look for the latest 4ML driver on HP's web site -- the
printer isn't THAT old, and basic PostScript really hasn't changed at all,
and the XP driver should be readily available. You might even have to
reinstall Windows, but that's a last resort.

Check Finale to make sure there are no options which could be causing this
result. (I've never used the full version of Finale, so I'm just guessing.)

Or just continue doing the bitmap generation thing, since you say it works.

Yes, the GhostScript method should work. But I don't think you've
identified the problem. In fact, since you haven't identified the problem,
you might have the same problem even with that method -- if something is
intercepting the output and substituing a screen bitmap before getting to
the printer driver, then you'd probably have the same problem.

Edward

 

[Dewdman42]

Thanks for the feedback Edward. I think we're on the same wavelength.
I have actually tried getting the latest postscript driver from HP and
I've also tried the one from Adobe using the HP PDD. Here is a bit
more info I thought of

1 - I actually can't print much of ANYTHING with the postscript driver.
I have tried to print various PDF's and they all just produce error
codes on the page. I know I used to print postscript stuff on this
printer in years past. I'm also using the printer as a network
printer, connected to a wireless print server which is connected to a
machine on the network and my finale machine is even a different
machine than that. So there could very well be a problem somewhere in
that chain that is munging up the PS data between Finale and the final
destination.

2 - I have also tried to just print the Finale score using the PCL
driver from HP. Theoretically, this should work fine too. It
certainly prints PDF documents without any problem at all. So it is
actually the PCL driver that is printing the angled lines with jaggies.
I can't seem to get PS driver to work right now. Its not clear to me
whether Finale embeds PS commands directly into the printer output or
sends something generic to the driver. It must send something generic
or else PCL wouldn't be able to print much of anything at all. But PCL
does print the right stuff...its just that angled lines look like crap.

3 - I have been able to produce a PS file of the score and inspect it
in GSView. When I used that neat application and set the zoom
resolution to 300dpi, and then zoom way in, I can get a pretty good
idea I think of what kind of jaggies I should expect to see from a
300dpi printer. What I see in GSView is definitely less jagged then
what my printer is outputing via PCL...and also i see that GSView in
fact does some anti-aliasing in addition which also smooths things out
even more.

4 - I don't think the printer is printing at lower than 300dpi in
general because all the music fonts and symbols and horizantal and
vertical lines and even the curved slurs all look excellent. The
curved slurs look good, not excellent, but not jagged either. Its only
the angled straight lines that are jagged, almost like the printer's
PCL RIP is just flat out not very good at RIPing angled lines from the
PCL vector data.

-steve

 

[SaGS]

...
3 - I have been able to produce a PS file of the score and inspect it
in GSView. When I used that neat application and set the zoom
resolution to 300dpi, and then zoom way in, I can get a pretty good
idea I think of what kind of jaggies I should expect to see from a
300dpi printer. What I see in GSView is definitely less jagged then
what my printer is outputing via PCL...and also i see that GSView in
fact does some anti-aliasing in addition which also smooths things out
even more.
...

I would recomend to (1) "print to file" one of the files that are
creating you problems, using the PostScript driver, (2) put the
resulting file on the Web somewhere publicly available, (3) post the
URL of the file here. Then one could inspect the file and try to locate
the problem, or at least confirm the file (and thus the application and
the driver) are OK. I had a LaserJet 4ML some years ago, and I remember
it as working very well.

 

[Dewdman42]

I will do that. Maybe one of you postscript gurus can figure out what
the heck is going on.

I also was just inspecting a postscript file with gsview. Finale has a
compile postscript option to generate postscript directly. Anyway, I
can't help but notice how razor sharp the edges of fonts look, even
when at angles. Curves, etc...but the angled lines just look like
absolute garbage..even when I set the gsview resolution to 1200. its
late now, but tomorrow I will post several things. i will post some
actual postscript. And I will post a PDF generated from the
postscript. and I will post some screenshots of what I see in gsview
at various resolutions.

part of me really wonders if Finale is doing something stupid..perhaps
only on windows since a high percentage of their user base is on the
mac...especially the serious ones, its possible...dunno..that the
windows version is doing something dumb.

 

[Dewdman42]

I will do that. Maybe one of you postscript gurus can figure out what
the heck is going on.

I also was just inspecting a postscript file with gsview. Finale has a
compile postscript option to generate postscript directly. Anyway, I
can't help but notice how razor sharp the edges of fonts look, even
when at angles. Curves, etc...but the angled lines just look like
absolute garbage..even when I set the gsview resolution to 1200. its
late now, but tomorrow I will post several things. i will post some
actual postscript. And I will post a PDF generated from the
postscript. and I will post some screenshots of what I see in gsview
at various resolutions.

part of me really wonders if Finale is doing something stupid..perhaps
only on windows since a high percentage of their user base is on the
mac...especially the serious ones, its possible...dunno..that the
windows version is doing something dumb.

 

[me]

I hear what you're saying, but this is really way worse than what
300dpi is capable of. Its nearly as bad as what I see on screen at
72dpi. When I print what must be vector based material, the angled
lines are severely stair stepped, but if I render a bitmap first..it
looks much more correct(does not look antialiased).

At any rate, does anyone know of software RIP solutions besides
manually rendering a bitmap file each time I want to print?

I hear ya though, I'm probably going to get a 1200x1200 dpi printer
pretty soon and maybe that will end the whole problem. I just want to
make sure I understand what is gonig on here.

What are the sizes of the bitmap and the pdf you are sending to the
machine? Does your driver allow for a variation in postcript version?

What resolution is the pdf being produced at? Have you tried printing
the same pdf on a different printer?

 

[Edward Reid]

Its only
the angled straight lines that are jagged, almost like the printer's
PCL RIP is just flat out not very good at RIPing angled lines from the
PCL vector data.

Either in PCL or PS, a line should be drawn at the maximum resolution of
the printer as long as the actual imaging is done in the printer -- which
should be the case if everything is working right.

Slurs, being curved lines that are nearly horizontal, are a worst case for
a printer. At 300dpi the slight jaggies -- where it adds or drops a pixel
from the curve -- just will be noticeable. It sounds like the printer is
doing the best a 300dpi printer can with them.

What exactly are these straight diagonal lines? Crescendo and diminuendo?
Finale would have to draw these; characters from a font would not do. Yet
it has to draw slurs too, and you say those are OK; I would think that if
straight lines were OK, curves would be too. Unless it is an interaction
between Finale and the printer driver ...

Anything that's drawn in PS should be completely smooth in GSView (to the
limits of your monitor) no matter how much you magnify it. You'll notice
that every time you magnify, GSView redraws the page -- it is
re-interpreting the PS code at the new resolution. PS code is resolution
independent (except when there's an embedded bitmap image), so the result
looks excellent at any resolution. For some reason that diagonal line isn't
being generated as a PS command.

I can't help with the PCL driver. But if you're having trouble printing
anything at all via PS, that's where I'd look for the problem. So ... the
print file has to go over Ethernet to another computer, then via wireless
to a print server box, then to the printer? Lots of places for something to
go wrong. Would it be possible to try a more direct connection?

Edward

 

[Dewdman42]

Again..thanks for your feedback. Well today I installed the newest
drivers available from HP, instead of using the ones that came with XP
(which is apparently what I was using). The HP website makes a comment
that the XP drivers are more generic and missing a few things. In any
case, installed the driver from HP, which calls itsself a Postscript
driver...though it probably sends some combination of postscript and
PCL to the printer I bet. In any case, whala..problem
solved..everything prints fine using that driver....no more postscript
errors (assuming its actually sending PS to the printer and I don't see
why it wouldn't be).

Now regarding the quality of the angled lines when I use this new
driver. its looks exactly the same as the other driver as far as I can
tell. Fonts, horizantal/vertical lines, and even the slur curves all
look fine. I can't see any jagged edges at all. But beams and
crescendo/diminuendos (as you pointed out) as the main culprits.

It seems that the the closer the lines are to horizantal, the more
obvious the jagg is. When it angles enough up towards 45 degrees or
even 30 degrees..it seems to do a much better job of representing the
line edge. But at low angles, the lines look like a series of
connected horizantal lines..quite visually. This could be bad
programming on the part of finale I suppose. Or it could simply be
that I'm hitting the limit of my 300dpi and I need to face the music.


I do find it interesting, that on the very same printer I can print a
PDF which contains a scanned image of some Finale music. (someone
printed out a finale score on a high res printer and scanned it to
create a page in a PDF file. When I print that page with this same
printer, the low angled beams look absolutely perfect..no jagged edges
at all, even when they are only slight angles.

This leads me to believe that either Finale should be using a different
method to represent beams, hairpins and other angled lines; or else
this particular printer just has a really lame algorithm for rendoring
vector data that Finale is sending it. I'll try to put some PS code up
on a web server later so maybe you gurus will be able to look at it and
see what they are doing that possible could have been done better. My
gut feeling is that with a newer printer this problem will go away
anyway. But it would be nice to understand why a bitmap representation
of an angled line looks great while a vector representing prints with
blatant jaggedness...

I do see the same jaggedness in GSView, particularly if I am viewing it
at 300dpi..it basically has about the same jagged edges that my printer
does.

Accent articulations which basically look like > symbol...look totally
perfect. They are font based, whatever difference that makes. For
some reason, fonts seem to scale MUCH better than the angled lines seem
to be rendered by PS...and obviously...pixel for pixel..the printer is
capable of representing a much nicer angled line than what PS appears
to be able to do with the current RIP.

(shrug)

 

[Edward Reid]

installed the driver from HP

Good move, obviously. At least that variable is removed.

which calls itsself a Postscript
driver...though it probably sends some combination of postscript and
PCL to the printer I bet.

One or the other, not both.

Now regarding the quality of the angled lines when I use this new
driver. its looks exactly the same as the other driver as far as I can
tell. Fonts, horizantal/vertical lines, and even the slur curves all
look fine. I can't see any jagged edges at all. But beams and
crescendo/diminuendos (as you pointed out) as the main culprits.

It really does sound like Finale, for some reason, is generating bitmaps
rather than PostScript for these. I don't know why. One possibility is that
the Windows API calls for printing don't allow Finale to shape these
elements exactly as they should be -- I can particularly see that for
beams, which should have the ends vertical rather than perpendicular to the
edges, though I'm not sure where a crescendo would have a problem. And even
a beam could easily be drawn as a long narrow parallelogram.

I think I suggested before: go through the Finale options in detail,
looking for a related option. Possibly there is some option for the
resolution at which to draw elements which Finale make bitmaps. If there's
still a user manual, search for the words resolution and bitmap.

I have an old copy of PrintMusic somewhere, but I don't have it installed
at the moment, and if I understand correctly it's actually a different code
base so even if I could examine its options, that wouldn't be likely to
help (if you follow all that).

It seems that the the closer the lines are to horizantal, the more
obvious the jagg is.

This is to be expected with any kind of "aliasing" problem (which is the
name for this kind of problem).

Or it could simply be
that I'm hitting the limit of my 300dpi and I need to face the music.

Your story about the scanned files argues against this.

This leads me to believe that either Finale should be using a different
method to represent beams, hairpins and other angled lines; or else
this particular printer just has a really lame algorithm for rendoring
vector data that Finale is sending it.

The latter is highly unlikely, but here's an easy way to test it, since you
have GSView already installed. Make a text file and paste in the following
PostScript program:

100 200 moveto
3 setlinewidth
500 205 lineto
stroke
showpage

Open that in GSView. Then print it, making sure to select the "PostScript
Printer" option. That will send the actual PS code to the printer. It will
draw a fairly thick, barely angled beam most of the way across the page.
You should have to look really closely -- perhaps using a magnifying glass
-- to see the 21 steps it makes. You can increase the angle by increasing
the "205" in the code.

If it looks better than the Finale output, then the problem is that Finale
is rendering the lines as either multiple lines or as a bitmap. And will
probably do the same with a better printer.

I'll try to put some PS code up
on a web server later so maybe you gurus will be able to look at it and
see what they are doing that possible could have been done better.

I could take a quick look, but usually the PS generated by printer drivers
contains so much excess cruft that it's really hard to figure out what it's
doing.

My
gut feeling is that with a newer printer this problem will go away
anyway.

My gut feeling is that you will be terribly disappointed if you act on that
gut feeling. My first laser printer was an Apple Personal Laserwriter NT,
around 1992, which IIRC was several years older than your printer. I never
had this kind of problem with that printer.

But it would be nice to understand why a bitmap representation
of an angled line looks great while a vector representing prints with
blatant jaggedness...

That's the argument that says Finale isn't using a vector graphic to draw
those lines, but rather (probably) generating a bitmap and doing it at the
wrong resolution -- maybe using screen resolution when printing.

I do see the same jaggedness in GSView, particularly if I am viewing it
at 300dpi..it basically has about the same jagged edges that my printer
does.

Another argument that Finale isn't putting out a vector.

Accent articulations which basically look like > symbol...look totally
perfect. They are font based, whatever difference that makes.

That guarantees that they are drawn with PS code.

For
some reason, fonts seem to scale MUCH better than the angled lines seem
to be rendered by PS

No, PS (even HP's PS emulation) will do exactly as well at drawing simple
lines as at drawing lines within glyphs.

I think you have a Finale problem and should be asking Finale support for
help. For the price of Finale, they ought to be able to answer a question
like this.

Edward

 

[Dewdman42]

Finale outputs vectors..not Bitmap. the beams that look good actually,
is a scanned image I have that was printed with finale, scanned and
then put into a PDF as a bitmap image. the bitmap looks good. Finale
definitely prides itself in postscript use...and the likelihood that it
is outputing a bitmap beam is barely slim to none. It is vector. I'm
actually thinking right now that postscript and PCL both basically suck
at printing low angle lines as vectors on 300dpi printers. The jagged
edges show up in GSview too, so I don't think its my HP printer or
driver. Its the actual postscript, perhaps what Finale is generating.


Don't ask me why the scanned bitmap looks so smooth, but the ones
produced the vector way suck. I need a lot more data to put up on a
website someday to validate this assertion and get better feedback
because at this point anything I or anyone says is a guessing game
without looking at the results and seeing it yourself... At some point
I would like to share the output PS file so that some of you can look
at it and see if you can figure out if there is something stupid in the
PS that could be done another way.

But...at this point I have purchased a 1200x1200 dpi printer and expect
the problem to go away.

It sure would be nice to know why, though, that I can print with a
different notation package I have...Overture..which is inferior to
Finale in hundreds of ways but for some odd reason it prints perfectly
smooth angled beams and lines. So... There is definitely something in
the mix somewhere that is not doing things the optimal way. When I get
a chance I will put up a website with pictures and files and try to get
you gurus to figure it out.

 

[Edward Reid]

It sure would be nice to know why, though, that I can print with a
different notation package I have...Overture..which is inferior to
Finale in hundreds of ways but for some odd reason it prints perfectly
smooth angled beams and lines. So...

So ... it works fine in Overture and fails in Finale, but you refuse to
consider that even Finale might have a flaw?

I have news for you: all large programs have flaws. I've been programming
for forty years, and I've seen plenty of them.

And it's unlikely that anyone here will take the time to look at your PS
output if you're not willing to take the time to run the five-line PS test
that I gave you.

Edward

 

[Dewdman42]

Edward, first of all, calm down please.

I didn't say finale was without failure. In fact I pointed out my
suspicion that it is doing something stupid. But I do not think it is
using BITMAPS to do it. Its either using some braindead vector
thing...which results in jagged lines in both PCL and
PS......or......PCL and PS fundamentally do not look as good on a
300dpi printer as do a scanned bitmap image of a similarly slanted line
which would point to a downside of vector base printing..

And don't worry..when I get some time I will put together a webpage
with PS files and PDF's and bitmaps and everything I can so that I can
understand the issue.

ps - relax.

 

[Dewdman42]

and by the way, I must have missed the 5 line PS test you sent. I Will
try that...as soon as I can find it...

 

[Dewdman42]

I tried your simple PS test. Thanks for that. Its jagged on my 300dpi
printer. The HP printer does attempt to smooth out the edginess of
each jag, but its still at the end a jagged edge...and the attmpted
smoothing does not look like nice aliasing.

Now this is an older printer and a newer one with higher dpi and
perhaps greyscale support, etc..will undoubtedly do a much nicer job.

but what I still don't understand is what a scanned bitmap image looks
excellent with no jagged edges and why the other software also seems to
be able to print smooth beams. Its entirely possible, actually, that
the other program is in fact using a bitmap get a smooth angled
line...i dunno. But anyway, I have a new printer ordered, so I suspect
it will be non-issue after that.

 

[Edward Reid]

I tried your simple PS test. Thanks for that. Its jagged on my 300dpi
printer. The HP printer does attempt to smooth out the edginess of
each jag, but its still at the end a jagged edge...and the attmpted
smoothing does not look like nice aliasing.

And you're saying that it has many fewer than 21 steps, which is the number
I would expect that line to have on a 300dpi printer? I have to ask, since
eyes vary so greatly; some people see smooth where others see jagged. (Some
people are happy watching movies on VHS.)

If that's the case -- and I'm assuming that you did pick the "PostScript
Printer" option in GSView -- then indeed something is wrong with your
printer. Hard to say what, but it sounds like the PS emulator might be
corrupted. Any PS emulator should render that line to the limits of the
printer's resolution.

In GSView on the screen, that line remains as smooth as possible on the
screen, no matter how much you magnify it, right? (If the line is jagged in
GSView, then you have some serious configuration problem in your computer.
I very much hope this is not the case. It's almost inconceivable for that
simple test to come out wrong on both the printer and in GSView ... but
I've seen stranger things.)

If you are seeing 21 steps on the printer, then your eyes are just very
sensitive to the unavoidable jaggies at 300dpi. I don't recall what
resolution enhancement (if any) this model has; it may be that some methods
utilize the smoothing methods better. But from your past descriptions, I
think you are saying that the line has fewer than 21 steps.

In either case, I now agree that you are correct and the new printer should
resolve the problem. If you ordered the 1320, I think you will like it.

Edward