Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Canon i950 - First Impression

1 view
Skip to first unread message

Jerry Schwartz

unread,
May 6, 2003, 11:07:24 AM5/6/03
to
Well, I bought the i950 to use as my only printer. Although right now I'm doing text more than photos, that hopefully won't be the case in the future. I can't comment on the image output, yet, but do have some initial observations and related thoughts:
  • After viewing the print sample in the store, I was initially aghast at the quality of the text: it was terrible, and I almost ruled the printer out altogether. Then I realized that the entire print sample was undoubtedly a bitmap, created who knows how; and that the text was probably mangled during the creation of the sample image. This turned out to be entirely true, and the text I've done at home is as good as you'd expect from a top-quality inkjet.
  • You cannot compare image quality using the print samples the stores generate. A friend of mine helped create some of those for HP, and a lot of work went into choosing and tweaking an image that looked just right. You might be willing to tweak a photo in your own work, but unless you have a bucket of money you can't have one printer for skin tones, one for sky and water, and another that excels in the primary colors prominent in advertising.
  • Whether this printer will be better, or worse, at printing photos than the others I considered is probably unanswerable for the reason I gave above. I remember almost buying an Alps thermal-transfer printer that did incredible skin tones, then noticing that it did a terrible job on dark grays.
  • Digital photo shops will take an image and do an Iris or Fuji print for you. If you give them "print-ready" copy, the price won't be that bad for occasional use. Sometimes it pays to rent.
  • The driver options are nice. Many are familiar to me from my Epson, such as print preview (a must when doing anything slow or expensive). Others, such as the built-in photo enhancements, I would probably never use since I have other ways of accomplishing the same things with more control.
    • The built-in monochrome options (sepia, etc.) may turn out to be useful because of their simplicity.
    • The automatic "quiet" schedule is an interesting feature, especially if your printer is in your bedroom. I haven't tried it, though, so I don't know how effective it is. It's not like this is a jackhammer to begin with.
  • Software installation was simple and straightforward. So far as I can tell, the drivers on the CD were the same as the ones I downloaded from the support site.
  • The included utilities can gobble up disk space, and may duplicate other software that you already have for cataloguing photos and the like. I haven't tried the Easy-WebPrint (Windows only), but if it works I will spend my next unemployment check to buy a Canon for my mother. Driving over there to help her print a web page without clipping the text is getting really tedious.
  • Hardware setup was simple, but while flipping the printer around to remove the packing tape I thought I had bent some of the plastic. It wasn't obvious until I unfolded everything that some parts are supposed to be curved. Getting the orange plastic seals off the ink cartridges without squeezing the cartridges takes a bit of practice.

--

Jerry Schwartz
 

Deathwalker

unread,
May 6, 2003, 1:30:21 PM5/6/03
to
Print 4 borderless on a4 and be prepared to be impressed.  This is with ez photoprint.  jpegs only and they can't be mega quality jobbies.  I have a negative scanner that produces a4 300dpi from a 35mm slide.  They are too much apparently.  30mb uncompressed.  Going to 48bit depth would be 60mb.  however for my fuji digital camera they are brilliant.  Produced 28 6x4 on 7 sheets of paper in under an hour.  Try that with an epson!
 
The web printer options are a god send.
 
The downloaded drivers give you the option of 5x7 borderless prints where as the cd ones don't.
 
It is damn quiet before quiet mode.  It just slows down the initial paper feed, actual printing is about the same (damn quiet). It doesn't do as many head cleans when sitting idle which it does in normal mode.  The auto switch on and off is good too.  You can set the times for quiet mode and the time idle before switching off.
 
 

--
Ian Lincoln Independent I.T Consultant
"Jerry Schwartz" <jerrys...@comfortable.com> wrote in message news:6-WdnZlhcfk...@comcast.com...

Jerry Schwartz

unread,
May 6, 2003, 7:11:53 PM5/6/03
to
I might have missed a file version difference somewhere, but the driver's overall version is the same on both drivers.
 
I thought about the quiet mode and the paper feed motion when the sound of a page being loaded shot me out of my chair.
 
I'm not sure what you meant about the file sizes; I've printed 60mb files on my Epson, although it is totally unnecessary most of the time. I'm not sure what the driver was doing to it by way of compression, but Epson uses a lot of disk space for rendering and I'm sure the Canon does as well. (The Epson 870 driver let you move the spool area to a different disk, that's how I knew where to look.) I don't know what the data stream to the printer looks like, so it might not vary as much as the input file sizes do.
 
Translating the printer's resolution into an optimal bits per inch is tricky stuff. With different horizontal and vertical resolutions, and six colors of ink, I would guess that something like 600 - 700 bits per inch in the source is in the ballpark. My trigonometry is not up to this any longer, nor is my understanding of the way the ink interacts with the papers, so rough cut will have to do.
 
As for the difference between 24-bit, 32-bit, and 64-bit color depth I don't know if the printer has the gamut to care. Just because you can put 16 million colors into it doesn't mean you're going to get 16 million distinct colors out of it, and I haven't found any assertions about gamut from Canon yet.

--

Jerry Schwartz
 

Gregg E.

unread,
May 7, 2003, 4:05:28 AM5/7/03
to
What did it cost?

How much is replacement ink?

What operating systems are supported?

Does it support Macintosh?

Is it likely that there will be full functionality for the NEXT
version of Windows after XP? (Same for Macintosh if it supports
OSX.)

Since I have both Macintosh and PC boxes, I prefer to buy stuff
that will work with both systems when I can.

Ron Cohen

unread,
May 7, 2003, 5:44:26 AM5/7/03
to
Jerry,
I still have my Canon s820 so I can't comment from experience on the i950, but from everything I've seen on the net, it's an outstanding printer. The drivers should be similar to the s820, so with a little tweaking of the printer profiles you should have a printer capable of all the things you mentioned in the second bullet.  For your Mom, take a look at the i850. I'm sure you've already seen the massive amount of postings in praise of that printer.  Also, check out Red River Paper.  http://www.redrivercatalog.com/  They have great paper at really good prices.  BTW, I checked out your web site. You did some interesting (quite good) artwork and the recipe section had a couple that my wife and I had been looking for.
--
Ron Cohen
drc...@ivwnet.com
"Jerry Schwartz" <jerrys...@comfortable.com> wrote in message news:6-WdnZlhcfk...@comcast.com...

Jerry Schwartz

unread,
May 7, 2003, 9:24:06 AM5/7/03
to
I can't answer most of this, since I don't have a Mac handy. You'd have to ask Canon about support for future versions, in any case; but they won't give you a straight answer for legal reasons. The printer definitely comes with Mac drivers that look like the Windows drivers, but the manual only describes OS 9. The architecture of OS/X (or any UNIX variant) is so different that I can't speculate on what they would look like. Here's what Canon says:
 
"Borderless printing is supported for Mac OS X v10.2 (Jaguar) or later. This printer driver is compatible with the iMac, G3 & G4 series computers that can run Macintosh OS X and are equipped built-in USB ports."
 
Some of the supplied utilities are for Windows, some are for Mac, and their features might or might not be identical. You can get more information on the Canon web site, including a PDF version of what passes for a manual.
 
Base price for the ink cartridges is $12 each; I'm sure you can order them more cheaply on the Internet but haven't investigated that yet. I haven't seen multipacks in stores, but I have only been looking for a couple of days.

--

Jerry Schwartz
 
"Gregg E." <gre...@valint.net> wrote in message news:3EB8BE48...@valint.net...

Deathwalker

unread,
May 7, 2003, 7:01:58 PM5/7/03
to
I shall try to clarify.
 
whilst looking into all this digital photography stuff i came across the accepted belief that scanner res and printer res are not the same.  Even with a 4800x1200 printer the highest you set a scanner for a 1:1 reproduction is 300dpi.  if you want to double the size of the copy you scan at 600dpi.  300dpi is more than enough even for a £250,000 fuji frontier digital laser printer which i beleive is a digital/wet chemistry highbrid.  I have had great results from www.photobox.co.uk. 
 
They say that inkjet printers need not go above 240dpi scanning.  Anyway to scan a 35mm transparency so it produces an a4 print at 300dpi requires a true optical res of 2800dpi or thereabouts.  a 900% magnification.  This requires a dedicated film scanner.  A flat bed using a lamp in the hood and interpolation to get the same res is not even close. Such a scan holds 30mb of uncompressed info if you are using 8bits per colour channel.  Using 16bits takes the size up to 60mb.  I am currently using 8 bit because there is only limited functionality within photoshop if you have 16bit files.
 
As for canon easy print, that little utility is limited to jpegs.  jpegs of 1.5mb of larger don't seem to register.  However photoshop doesn't mind printing from the canon one little bit.
 
With my digial camera i can reduce the resolution to 250dpi and thus increase the print size without resampling.  As it is only a 2megapixel camera (producing 4.5mb uncompressed files or 720kb minimum compression jpegs) i can get an a4 print without resampling at around 220dpi.  At this magnification the focusing seems a little soft sometimes but you still don't see the individual dots on the printed page.
 
So in summary.  inkjet technology is inferior to industrial and dye sublimation printers.  Optical resolution of an image need be no more than 264dpi at the final output size.  although the printer will still need to be set at maximum res 4800x1200.
 
The canon easy print utility is limited to jpeg files and these jpeg files must be smaller than 1.5mb.

--
Ian Lincoln Independent I.T Consultant
"Jerry Schwartz" <jerrys...@comfortable.com> wrote in message news:yuudnRRe7bH...@comcast.com...

Jerry Schwartz

unread,
May 8, 2003, 11:57:31 AM5/8/03
to
[Comments interspersed...]

--

Jerry Schwartz
 
I shall try to clarify.
 
whilst looking into all this digital photography stuff i came across the accepted belief that scanner res and printer res are not the same.  Even with a 4800x1200 printer the highest you set a scanner for a 1:1 reproduction is 300dpi.  if you want to double the size of the copy you scan at 600dpi.  300dpi is more than enough even for a £250,000 fuji frontier digital laser printer which i beleive is a digital/wet chemistry highbrid.  I have had great results from www.photobox.co.uk. 
 
I think what you said above is completely right, but I'm not sure I understand your point. I'm going to belabor this for the sake of others, since I'm sure you understand what's really going on.
 
You are correct that if you scan at 300dpi and print at 300dpi, you'll get 1:1 reproduction: a 1" x 1" photo will generate a 1" x 1" printout. You are also correct that even the fanciest commercial printers have relatively low resolution, compared to what you'd expect; but they produce good results because of the technology they use. Those high-end boxes use a photographic process, ultimately. Essentially, they generate an image digitally (on what, I'm not quite sure) and then transfer that image using an optical/wet chemistry process; it's like taking a standard photograph of an image displayed on your computer monitor (but with carefully-designed specialty equipment). This technique actually goes back to the dark ages, when you could buy special monitors with Polaroid cameras bolted on. The "resolution" of the photographic portion of the process vastly exceeds that of the digital portion, so if you can get a good digitally-displayed image you're home free (home expensive, anyways).
 
This whole resolution / dpi business confuses a lot of people who ought to know better, mostly because programs in the "artist" market emphasize it. It all comes down to pixels: how many you've got, how many you need. The dpi (or ppi) is a "hint" that describes the intended physical size of the image, but it really means about the actual image. It is commonly said that screen images are 72 ppi or  96 ppi, but the same 1000 pixel x 1000 pixel image is going to look a lot different on a 14" monitor vs. a 21" monitor (let alone a handheld) so the dpi means absolutely nothing in this context. DPI is only another way of specifying the size of the printed output that you want. 
 
They say that inkjet printers need not go above 240dpi scanning.  Anyway to scan a 35mm transparency so it produces an a4 print at 300dpi requires a true optical res of 2800dpi or thereabouts.  a 900% magnification.  This requires a dedicated film scanner.  A flat bed using a lamp in the hood and interpolation to get the same res is not even close. Such a scan holds 30mb of uncompressed info if you are using 8bits per colour channel.  Using 16bits takes the size up to 60mb.  I am currently using 8 bit because there is only limited functionality within photoshop if you have 16bit files.
 
This is where things get really dicey. No existing hardware technology that I know of really, truly works with pixels at the physical level: monitors work with red, green, and blue phosphor dots or red, green, and blue liquid crystal cells; printers work with ink droplets, toner particles, puffs of sublimated dye, or (in the case of those high-end jobs) crystals of various chemicals diffused and suspended in layers of transparent gelatin; scanners do whatever voodoo that they do using lights, filters, and photoreceptors; digital cameras substitute external light sources, but otherwise work like scanners; and your eyeballs, for that matter, work with arrays of photoreceptors called cones and rods that are sensitive to a few specific colors. So from the get-go, a "pixel" is an abstraction; all the rest is done by interpolation in your software or brain.
 
Your example of scanning a 35mm transparency at 2800dpi is, I assume, arithmetically correct for A4 paper (we unilateralists here use "letter" paper). The term dpi is used to mean "pixels per inch" (it must drive you folks crazy to mix things up like this). And you're right, unless you have a dedicated film scanner you're not going to get the best results; it's like using a camera with a mediocre lens. However, the scanner hardware and software collaborate to produce a number of pixels that corresponds to the area you scanned and the dpi you requested.
 
Here's where things get really nasty. Let's assume that we can get the number of pixels we want; but how many pixels do we need for best output on a particular printer at a particular size? As I said, printers don't really have pixels. What they do is create images of pixels using whatever technology they're designed around. A printer with 6 inks will do it slightly differently from a printer with 4 inks or a printer with 27 inks. The number of inks, per se, does not equate to quality. Wet photography works quite well, thank you, with a limited number of "real" colors available: look at what Ansel Adams was able to do with the choice of "black" and "not-black." Printers do their best by putting clumps of whatever coloring agents they use onto paper. Some put the different agents side by side, some can layer them on top of one another; some can control the size of the individual bits of coloring agents, some just make them all the same but put more or less into a given area. There's no one best way to do it, which is part of why this newsgroup has such high traffic.
 
Those clumps are related to what professional printers (the people, I mean) call "half-toning." At its simplest, this means using clumps of really small dots to create the appearance of different shades by varying the density and shape of the clumps. That's not very easy to picture, but take a magnifying lens to a comic book and you'll see that they don't use evenly-spaced dots of ink; if they did, they'd get a very harsh and grainy appearance. They take advantage of the fact that our eyes (and associated wetware) do a lot of smoothing and interpolating. Half-toning isn't done by trial and error, it's done by math that involves the number of fundamental colors you have to play with.
 
Note that the difference between an Ansel Adams print and a newspaper photograph is that wet photography operates on a scale which isn't obvious to the human eye, whereas newspapers (for economic reasons) operate on a scale that can be seen. Wet photography doesn't need half-toning because we can't see that well.
 
I said before that we should assume we can get the number of pixels we want; let's further assume that we can get the number of colors (gamut) we want. (That's where the number of bits per pixel comes into play.) How do we get something that looks right onto the page? Here we are completely at the mercy of the printer, because unless the manufacturer tells us we just don't know.
 
If the printer just took the raw pixels and recreated them on the page, it would be easy to figure out. That's how the early laser printers and their associated drivers did it: give me a pixel, and I'll put a black dot on the paper or I won't. The results were pretty gruesome: it was like looking at a silhouette through a screen door. Dithering, which is sort-of like half-toning (more precisely, half-toning is one specific way of dithering), helps: one black dot and one non-black dot should come out medium gray, except that there might be space between the dots and you might get better results using six dots out of twelve, and you can use a cluster in the middle of the space or you can do it odd-even-odd-even, and do you use a 4x3 pattern or a 3x4 pattern and....
 
If you think about it, even a 300dpi B&W laser printer ought to be able to give a pretty good B&W picture (at least when seen from a distance); the problem is, how do you distribute the toner dots? Modern printers, whether monochromatic or color-capable, do a much better job because either the printer or the driver has sophisticated ways of dithering. Some, especially the laser printers, can control the actual size of the toner dots somewhat; others achieve a similar thing by overlapping dots (the dots are bigger than the smallest space between them); and so forth. They can use all kinds of different patterns, to avoid that unsightly checkerboard look.
 
But just because a printer has the ability to lay down a single 2 picoliter droplet of ink in one particular cell of a 4800 x 1200 matrix doesn't tell you how good the skin tones will be. These higher resolutions are more a statement about the precision with which a printer does its magic than anything else; and the same goes for the number of ink colors. Higher precision implies greater control which implies the potential for better results.
 
The coloring agents themselves, of course, will largely dictate such things as color saturation and purity. If the pure yellow isn't really yellow, or looks washed out, you'll never get that lemon to look right. And does the coloring agent completely obscure the underlying paper, the way a good housepaint would? or does it rely on the the basic color of the paper to give the blues more bite, or to tone down a too-harsh magenta needed to get saturated reds? Does the ink sink in and spread, giving a smoother (or blurry) look to everything? Does it sit on top of the paper like spattered paint, giving purer colors but a harsh, grainy look? How big a dot does a 2 picoliter drop of ink make, anyways? If we put a magenta droplet next to yellow droplet, will they merge into a single (but bigger) splotch of orange? or will they just sit next to each other like ham and eggs on a plate?
 
That gets us to the whole question of paper. Printer manufacturers will tell you to use their own paper for best results, and then price it to make money. That might seem harsh, but they aren't lying. How the coloring agent interacts with the color, glossiness, and absorptive qualities of the paper will greatly affect the final results. Professional photographers know this, and are used to selecting papers for different effects. Kodak doesn't just make "photorealistic" papers that match what the naked lens sees; they produce premium papers that have lower or higher contrast, that deliberately introduce warmth or texture, and so forth. The paper most suitable for a microbiologist documenting little icky things won't be flattering to a bride's complexion, and they know it. Wet photographers have one big advantage, though: they start out with light, and the paper then does what it is supposed to do. They don't have to worry about what using "after-market" light (well, technically photo processing does rely upon a standard light source behind the negative, but those are indeed standardized) might do to a particular paper unless they are deliberately trying to create special effects.
 
We, on the other hand, have to worry about the physical interaction of the ink we buy and the paper we buy. Using after-market papers and inks introduces far more variations into an already confusing situation. Unless you work in quantity and are careful to document your experiences, you should avoid switching around; and for many of us, that means sticking to the printer manufacturers' recommendations.
 
So, back to where we started: what resolution should you use? As I said, unless the printer manufacturer tells use we can only guess. The stated resolution of the print mechanism is used to create dither patterns which are in turn used to simulate pixels which never really existed in the first place until your camera or scanner fed its version of reality into your software. Giving the printer 5,000 pixels per inch of paper used might be overkill if the printer mechanism only does 1,000; the software has to interpolate down to little shots of color in certain places. Depending upon how it interpolates, certain resolutions might mislead it under certain circumstances and produce moire patterns, color shifts, chiaroscuro problems, and such. Giving the printer too little information will generally lead to "soft" images, which you might even like! (No wedding portraits should feature acne, after all.) Again, though, you are dependent upon the techniques used inside the printer hardware and software.
 
The best way to look at this, actually, is from the standpoint of information: how much information (how many pixels) does the printer need to generate its best possible results? How much information do you have to begin with. (If you think in terms of information, that whole "dpi" business falls into perspective.) Too much information is wasted, and too little relies upon the printer to fill in the gaps with guesses. Regardless of what you start with, the print mechanism moves and shoots the same way. Either trial and error, or information from the manufacturer, is required to find the magic number. And even then, different papers and different pictures will give you different amounts of leeway.
 
Although 240 pixels per inch is the number often thrown around for inkjet printers, I believe that is an old estimate that has become enshrined in popular wisdom. It just doesn't make sense that the same number I was given with my first photo inkjet is still universally applicable.
 
As for canon easy print, that little utility is limited to jpegs.  jpegs of 1.5mb of larger don't seem to register.  However photoshop doesn't mind printing from the canon one little bit.
 
I haven't used the Canon software. JPEGs are lossy compression, and I rarely use them except for web sites. Since most cameras store images as JPEGs, you've already lost some of the original resolution by the time the image gets to the computer. A 2 megapixel camera is a 2 megapixel camera; the dpi setting is just a wish about the size of some printout that doesn't yet (and might never) exist. If you were doing no compression at any step, then a 2 megapixel image at 24-bit color resolution would create a 6291456 byte file. Does your camera offer the ability to capture images as TIFFs or some such? If not, you've already lost some information by the first time you touch the file. When you say (below) that you have a 2 megapixel camera that produces a 4.5mb uncompressed file, what you really are saying is that you've already lost around 1/3 of the original information. You've actually lost more, because an uncompressed JPEG isn't the same thing as a never-compressed JPEG.

Deathwalker

unread,
May 9, 2003, 6:17:16 AM5/9/03
to
I do like that a flame war hasn't been stated.  I did put that this was "my understanding after sifting through the info" On the scanner groups this quickly dissolves into verbal abuse and general name calling and no actual clarifications.
 
feel free to skip through this to Summarise 7 or 8 paragraphs down. (bloody hell 8.  this is getting out of hand).
 
At the end of my long passage i had upped the maximum necessary optical dpi (what did you call that?) to 264.  For important stuff i work with transparencies so i can more easily determine what the original colour was.  I also get higher resolution.  I also work with uncompressed tifs. I will also up the res to 300dpi just to be sure.  I doubt that even my new £160 printer can top fuji frontiers best efforts, so no need to go higher.
 
With my epson 600 the photodeluxe readme suggested 150 dpi optical was all that was necessary. Eyeball tests without magnifying glass caused me to agree.
 
As for jpegs well my eyeballs cannot see the difference providing i don't keep resaving the image over and over.  My digital camera only does jpegs and i only select minimum compression (finest quality).  I do have the option to convert to tiffs during the transfer to hard disc.  I will then work on that in tiff format.  When i'm satisfied that no further changes are made it goes back to jpeg 10 on the jpeg quality scale (12 being no compresson 1 being maximum)  As the original jpeg file was 720kb resaving it as a 4mb jpeg seems pointless.  There are possible flaws in this line of thought but having filled a 40gb drive with tiffs i've gone back to jpegs for practical reasons (until more money is availabe sigh!)
 
The interesting thing about jpegs is that the losses involved aren't as big as the loss from digital image to inkjet print so in practice the loss isn't noticeable (at least to my eyes).  No doubt purists reading this will baulk at that.
 
I would like to know the exact maths involved in the resolution of my camera.  I read that only 3/4 of the sensor is actually usable.  2.1 actual sensor but 1.8 usable or some such.  They also require 3 sensors (rgb) to produce a specific colour so you take the usable res and divide by three. 
 
The final image from my digital goes 1600x1200.  Now adjusting the dpi in photoshop with resampling unchecked keeps the resolution at 1600x1200 but the image size changes.  so i imagine its simply a matter of 1600x1200 pixels divided over an area.  The smaller the area the greater the dpi. (pixels per inch to be precise).  The bigger the area the further apart each individual pixel becomes.
 
Finally my original passage also went into too much detail.  The point i was trying to make is that your typical chemist photo has less than 300 optical dpi.  Therefore even with a million pound flatbed scanner there is only 300dpi optical information you can get from it.  scanning at 600dpi will not produce a better enlargement than scanning at 300 and resampling in photoshop.  ( The method of resampling, fractal or bicubic, going up in 10% increments as opposed to a 1 jump enlargement is not something i want to get into.  I'm getting bogged down again.)
 
So to summarise.
I believe that 264 is currently the maximum optical output resolution but the printer must always be set to maximum quality.  The quoted resolutions of scanners and printers have nothing in common.
 
300dpi if you want to take no chances.
 
enlarging prints will always be useless regardless of flatbed not due to any part of flatbed technology but due to the fact that there is only 300dpi of information to be obtained.  If you are worried about quality then "film scanning" is the way to go.  Having said that for home use scans at 300, and without resampling, printing at 264 or even 220 will produce a reasonable print but may appear soft focused at 220 or less.  The imaging package will tell you how large the resultant image will be.
 
 
 
 

--
Ian Lincoln Independent I.T Consultant
"Jerry Schwartz" <jerrys...@comfortable.com> wrote in message news:2mqdnVtivdA...@comcast.com...

Jerry Schwartz

unread,
May 9, 2003, 8:14:08 PM5/9/03
to
[Comments interspersed]

--

Jerry Schwartz
 
I do like that a flame war hasn't been stated.  I did put that this was "my understanding after sifting through the info" On the scanner groups this quickly dissolves into verbal abuse and general name calling and no actual clarifications.
 
That's a shame. I've moderated quite a few forums over the years, and although I've been pretty successful I have had one spectacular failure. Fortunately, the particular medium allowed me to expel the offender - eventually.
 
feel free to skip through this to Summarise 7 or 8 paragraphs down. (bloody hell 8.  this is getting out of hand).
 
Yep. ;<)
 
At the end of my long passage i had upped the maximum necessary optical dpi (what did you call that?) to 264.  For important stuff i work with transparencies so i can more easily determine what the original colour was.  I also get higher resolution.  I also work with uncompressed tifs. I will also up the res to 300dpi just to be sure.  I doubt that even my new £160 printer can top fuji frontiers best efforts, so no need to go higher.
 
When working with a scanner, I tend to do much more than necessary and reduce later in software, where I have both greater control and an undo feature.
 
With my epson 600 the photodeluxe readme suggested 150 dpi optical was all that was necessary. Eyeball tests without magnifying glass caused me to agree.
 
As for jpegs well my eyeballs cannot see the difference providing i don't keep resaving the image over and over.  My digital camera only does jpegs and i only select minimum compression (finest quality).  I do have the option to convert to tiffs during the transfer to hard disc.  I will then work on that in tiff format.  When i'm satisfied that no further changes are made it goes back to jpeg 10 on the jpeg quality scale (12 being no compresson 1 being maximum)  As the original jpeg file was 720kb resaving it as a 4mb jpeg seems pointless.  There are possible flaws in this line of thought but having filled a 40gb drive with tiffs i've gone back to jpegs for practical reasons (until more money is availabe sigh!)
 
The stuff I print tends to scanned photographs, but I also do a lot of work with computer-generated graphics. The JPEG algorithms are pretty good, overall, but I've run into some particularly horrendous exceptions even with moderate compression. The combination of sharp edges and many colors can give rise to a halo effect that makes text hard to read, for example. That's why you shouldn't use JPEGs for web page buttons if you can avoid it. I've done it, but it took some careful tweaking of the image to make it look okay.
 
By the way, TIFF allows for loss-less compression. That might save you some space, but watch out - the TIFF standard is so extensible that it isn't hard to write a TIFF from one program that another program can't read.
 
The interesting thing about jpegs is that the losses involved aren't as big as the loss from digital image to inkjet print so in practice the loss isn't noticeable (at least to my eyes).  No doubt purists reading this will baulk at that.
 
I'm sure you're right, in most cases, as long as you don't get some of the really obnoxious artifacts such as a halo, streak, or banding.
 
I would like to know the exact maths involved in the resolution of my camera.  I read that only 3/4 of the sensor is actually usable.  2.1 actual sensor but 1.8 usable or some such.  They also require 3 sensors (rgb) to produce a specific colour so you take the usable res and divide by three. 
 
The final image from my digital goes 1600x1200.  Now adjusting the dpi in photoshop with resampling unchecked keeps the resolution at 1600x1200 but the image size changes.  so i imagine its simply a matter of 1600x1200 pixels divided over an area.  The smaller the area the greater the dpi. (pixels per inch to be precise).  The bigger the area the further apart each individual pixel becomes.
 
That's almost right. The pixels actually get larger when rendered; otherwise the picture would become overcast by whatever the color between the pixels works out to be (white or black). Pixels aren't always the same shape, either; I don't remember the specifics, but some devices use oblong pixels, some square pixels, and some round ones. Software treats them all the same, as points (neither size nor shape).
 
Finally my original passage also went into too much detail.  The point i was trying to make is that your typical chemist photo has less than 300 optical dpi.  Therefore even with a million pound flatbed scanner there is only 300dpi optical information you can get from it.  scanning at 600dpi will not produce a better enlargement than scanning at 300 and resampling in photoshop.  ( The method of resampling, fractal or bicubic, going up in 10% increments as opposed to a 1 jump enlargement is not something i want to get into.  I'm getting bogged down again.)
 
Are you sure that a typical (whirr, click, translate) drugstore photo processor does less than the equivalent of 300 dpi? I think the processing is actually better than that, if the equipment is well maintained, but when you take into account the quality of the camera you're probably right. When HP first made its push into photo printing, they claimed that their top-end printer was "better than photo quality." I found that statement downright baffling, until I figured out what they were using as a standard: the typical point-and-click camera with consumer-grade moderately fast film.
 
Film and photographic paper have grain, which is the equivalent conceptually of resolution, but I think it's a lot better than 300 dpi even for consumer-grade. The terminology is completely different, and I couldn't begin to establish a correlation even if I had the raw data to begin with.

Model Flyer

unread,
May 9, 2003, 11:19:48 AM5/9/03
to
Ian, this message of yours was just on 57KB, some of us are still
using dial in systems, so please use the snips.
--

.
--
Cheers,
Jonathan Lowe
modelflyer at antispam dot net

Antispam trap in place

************** old message ******************


"Deathwalker" <ian-l...@blueyonder.co.ukspam> wrote in message

news:MeLua.245$d56.2...@news-text.cableinet.net...

Deathwalker

unread,
May 10, 2003, 5:50:53 AM5/10/03
to
okay dokey. There is an attachment on all these that i didn't put in.

--
Ian Lincoln Independent I.T Consultant

"Model Flyer" <anti...@null.net> wrote in message
news:oxXua.12679$pK2....@news.indigo.ie...

Deathwalker

unread,
May 10, 2003, 5:57:52 AM5/10/03
to
 

--
Ian Lincoln Independent I.T Consultant
That's a shame. I've moderated quite a few forums over the years, and although I've been pretty successful I have had one spectacular failure. Fortunately, the particular medium allowed me to expel the offender - eventually.
When working with a scanner, I tend to do much more than necessary and reduce later in software, where I have both greater control and an undo feature.
 
The stuff I print tends to scanned photographs, but I also do a lot of work with computer-generated graphics. The JPEG algorithms are pretty good, overall, but I've run into some particularly horrendous exceptions even with moderate compression. The combination of sharp edges and many colors can give rise to a halo effect that makes text hard to read, for example. That's why you shouldn't use JPEGs for web page buttons if you can avoid it. I've done it, but it took some careful tweaking of the image to make it look okay.
 
Are they are usually gifs aren't they?
 
By the way, TIFF allows for loss-less compression. That might save you some space, but watch out - the TIFF standard is so extensible that it isn't hard to write a TIFF from one program that another program can't read.
 
I only use photoshop.
 
I'm sure you're right, in most cases, as long as you don't get some of the really obnoxious artifacts such as a halo, streak, or banding.
 
That's almost right. The pixels actually get larger when rendered; otherwise the picture would become overcast by whatever the color between the pixels works out to be (white or black). Pixels aren't always the same shape, either; I don't remember the specifics, but some devices use oblong pixels, some square pixels, and some round ones. Software treats them all the same, as points (neither size nor shape).
 
Well it to enlarge they are usually squares.  So it just makes them bigger?  suppose why it gets blocky when i zoom in.  They are actually squares.  Don't know if that is just for my screen.  Does the printer produce tiny squares?  If i enlarge will it simply make the splodges (dots) larger.
 
 
Are you sure that a typical (whirr, click, translate) drugstore photo processor does less than the equivalent of 300 dpi? I think the processing is actually better than that, if the equipment is well maintained, but when you take into account the quality of the camera you're probably right. When HP first made its push into photo printing, they claimed that their top-end printer was "better than photo quality." I found that statement downright baffling, until I figured out what they were using as a standard: the typical point-and-click camera with consumer-grade moderately fast film.
 
I picked up that assumption on scanner group.  There is a bloke representing dp now who recently agreed.  Only he refers to "resolving power".
 
Film and photographic paper have grain, which is the equivalent conceptually of resolution, but I think it's a lot better than 300 dpi even for consumer-grade. The terminology is completely different, and I couldn't begin to establish a correlation even if I had the raw data to begin with.
 
Most drug store stuff has colour casts are soft focused and the paper is extra cheap.

Jerry Schwartz

unread,
May 10, 2003, 8:34:36 PM5/10/03
to
This is easier to grasp if you stick to thinking about the information content of an image.
 
A bitmap consists of pixels, each of which is two pieces of information: location and color (not size). That's really all you have, regardless of how the bitmap got into your computer (scanner, digital camera, paint program).
 
If you change the dpi of an image, without doing anything else, you are telling the rendering engine (printer, for example) how far apart the pixels are. The amount of information you have (number of pixels) remains unchanged. In practice, the rendering engine will fill the space occupied by the image by making the pixels larger or smaller, rather than leaving gaps between them.
 
If you "enlarge" the image in your software, you are creating extra pixels - but you don't have any more information than you started with, so the software has to do clever things to guess what the new pixels should be like. If it guesses wrong, the results will be coarse, show patterns, or make the image darker or lighter.
 
If you make the image smaller, you are reducing the number of pixels - and actually discarding information. Again, the software has to be careful or you'll get unpleasant results beyond simply losing detail.
 
When you assert (with justification) that 300 ppi is the upper useful limit for scanning a typical snapshot, you are saying in effect that 300 ppi is all the useful information a snapshot has to offer.

Larry

unread,
May 11, 2003, 2:48:59 AM5/11/03
to
I print 2-3MB jpegs all the time with Canon Easy-PhotoPrint.
 
In fact, my camera is a 4Mpixel and all of the pictures are well over 2MB.
 
There is an updated PhotoPrint application on Canon's site.  Maybe you need it ?
 
Larry
 

Deathwalker

unread,
May 11, 2003, 1:50:49 PM5/11/03
to
When an image from my camera is first imported it is insanely large but only 72dpi.  I therfore uncheck resampling and change dpi to 300.  That will produce an image 4x 5.88 inches.  I think 264 = 4.5x6 inches which is the paper photobox uses as default on their frontier machines.  I hope i am not throwing away any info.
 
 
As mentioned in an earlier post it is recommended if you resample to do it only 10% bigger at a time.  Not one jump to the final size.  Also there is a plug in called genuine fractals that supposedly does a better job than the bicubic resampling alogorhythm in photoshop.  Again it is suggested enlargements are incremental. 
 
I am most certainly saying that there is only 300dpi worth of information available even on a wet chemistry print.  Whether i am correct or not is another matter.  But as fuji haven't built thermal dye sub or frontier machines with higher res than that they obviously don't believe the wet chemistry has now been matched.

--
Ian Lincoln Independent I.T Consultant
"Jerry Schwartz" <jerrys...@comfortable.com> wrote in message news:zCadneZOOds...@comcast.com...

Jerry Schwartz

unread,
May 11, 2003, 3:23:26 PM5/11/03
to
What happens depends upon the software you are using. Changing the dpi without resampling sounds as though it would keep the same number of pixels.

--

Jerry Schwartz
 
0 new messages