Resolution concern?

[Bifurcator]

At the bottom of that chart it says:

"At 150 BPI printed images will have visible pixels and details will look "fuzzy"".

So why is it that an image displayed on my monitor at 80 BPI can look pixel free and perfectly sharp?

Different media. One is a lit (luminous) pixel the other is a reflected light off of paper. It's different and the eye's blur it differently. But you're both right here 150 is more of a pro grade standard if I can call it that. 122 and maybe lower looks awesome to me!

 

[Moglex]

I actually did a couple of prints at 75 DPI and there was no visible pixilation except when examined under a magnifying glass.

As to 'acceptable' sharpness, this always seems like a moveable feast.

A 10 x 8 from a half frame camera could look perfectly sharp and detailed.

Compare it to the same scene from a full frame and it looks much less impressive.

Compare the full frame to a Hasslebal shot ...

Until you end up as Ansel Adams there's always somewhere to go.

It really does depend entirely on the purpose of the finished print - so much so that stating an arbitrary figure of 300 DPI is a nonsense.

 

[TamiyaGuy]

I agree with the likes of Reg, Moglex, Sw1tch, etc. The number of pixels in a camera nowadays has very little meaning, yet camera maufacturers constantly bring out cameras with more and more pixels. Why, you might ask? Because it's almost the only thing the makes up camera's quality that can be expressed in cold, hard numbers. Back in the days when the absolute top-of-the-range models had 2.7mp, this had SOME meaning (although as Sw1tch explained, not all that much), but now it's just hype.

Many, many other things in a camera make much more of a difference than megapixels ever can, like image noise, sensor quality, lens quality, image processing, and many many more things. The problem is, they can't be expressed in numbers, so manufacturers keep with the "more pixels = better photos" motto (I guess you could express those things in numbers, but you'd need a PHD in computing to understand what they meant )

And just remember that at the end of the day, you're the one taking the photo, not the camera. If you're a terrible shot, a 200MP camera can't make up for it. Likewise, if you're a really good photographer, you can take photos with a 2mp camera and sell them for hundreds.

 

[Bifurcator]

I actually did a couple of prints at 75 DPI and there was no visible pixilation even when examined under a magnifying glass.

Don't most printer drivers apply diffusion dithering? Mine even let's me choose diffusion or ordered. This is going to smooth out what would otherwise be noticeable pixelation.

@TamiyaGuy
I guess I know what you mean but pixel resolution does indeed matter more than you seem to be implying.

 

[Moglex]

Don't most printer drivers apply diffusion dithering? Mine even let's me choose diffusion or ordered. This is going to smooth out what would otherwise be noticeable pixelation.

The printer dithers within a source pixel.

If it started dithering between them it would end up damaging the resolution of the material it was supposed to be printing.

If you send it something at a resolution of 100 DPI and it is capable of printing 1000 DPI that gives it a 10x10 area with which to play in as it combines the ink colours it has available to make up the desired colour for the source pixel.

I did wonder if the people who made the chart were getting confused between something that printed at 150 DPI, which would certainly show pixilation (due to not having enough pixels to adequately dither the colour palatte) and somthing with a source resolution of 150 DPI which should not show pixelation provided the printer has the printing resolution to cope.

 

[Bifurcator]

The printer dithers within a source pixel.

OK, that'll still smooth out the apparent resolution tho right? And what would look blocky on a non-dithering device will appear smooth... or at least smoother! True?

 

[Moglex]

OK, that'll still smooth out the apparent resolution tho right? And what would look blocky on a non-dithering device will appear smooth... or at least smoother! True?

Technically, if it did not dither within a pixel it would look posterised, which would certainly make the pixels stand out so you could legitimately say it would look pixilated, yes.

However, this (dithering) will have no benefit in making the source look smoother because all it will do is make the pixels in the source the correct colour.

The printer cannot take it upon itself to smooth the data it is given otherwise it might smooth out information that was actually wanted!

So the bottom line is: if the data is at such a low resolution that it should appear pixilated then the printer cannot assist in making it look non-pixilated.

 

[Bifurcator]

Well, I dunno. I kinda think that's one-a-dem laws which don't translate into reality like the physics equations that say it's impossible for bumblebees to fly.

Let's look and use our own powers of observation:

So I printed a photograph to an "L" sized print on my brand new MX850 inkjet uber-hires printer at it's finest setting with it's highest grade "special" photopaper. I sent the data over at 122 PPI - which is supposedly what some higher resolution magazines generally use as source image data.

Here's a crop of the image with the area we're looking at marked:

I also cut off the bottom and side a little bit in a crop to make it generally smaller.

OK, let's see what it looks like in photoshop zoomed up on that area with all the display diffusion dither settings turned off of course.

This is grabbed in PNG format so that there'll be no j-peg noise. These are the actual pixels that were sent to the printer.

Now let's look at what the printer and driver did to the data.

Wow that looks dithered inside and out and interpolated to me at first glance. But that's what my camera saw... Maybe it added something? OK, here then is a scan of the same area with all the color profiling, dithering, and other options turned off:

Hmm, yeah, I suppose it could be only internal dithering as you say. Whatever it is it sure made a smooth mess of the original pixels.

The data is here. What conclusions do you draw?

 

[Bifurcator]

BTW, the printed photograph looks just about perfect. I can notice jaggies if I look hard enough around any areas of contrast but only if I hold it about 10 inches from my eyes and look for them. Set on the desk 21 to 24 inches away it looks 100% perfect.

 

[Alex_B]

If resolution is my main concern when taking pics, should I be most focsused on getting a DSLR with a higher number of megapixels?

I could not be bothered to read rthe over 50 replies to this.

But anyway, resolution and megapixels are two different things!!

resolution depends on the whole optical system, including the lens. hence lots of MP require good lenses.

how big do you want to print? for the standard print on the wall, IMHO 8-10 MP will be enough, given there is not too much noise and the lens is ok.

 

[Garbz]

Bifurcator this is entirely dependant on software and printer drivers. There are many packages out there which will interpolate data if you zoom in to print independent of the driver. Some drivers may or may not do it too. My Canon IP4200 definitely produces blocky results if I print at 100ppi and then scrutinise the result.

 

[Bifurcator]

I didn't zoom in to print though. I printed the whole image and then pointed my camera at a tiny little section of it. The image was printed in Photoshop by loading the file, setting it to 122 DPI and pressing print. Also the driver is set to not "do anything" to the file and allow the printer to decide everything. If I set it to allow Photoshop to decide (as I did in the past once) it looks like total cr*p. Not sure why.

Or did you mean something else and it went over my head? (wouldn't be unusual for me )

 

[Garbz]

Not quite sure any more I am totally confused. So you set photoshop to print at 122dpi and nothing else changed? I'll have a play in a sec.

By the way the reason is looked like crap is if you do anything in photoshop that changes the default normal output you need to disable all colour management in the printer driver. This is a case of photoshop adjusting the colours to your printing space, the printer assuming that the incoming data is sRGB and converting it again. But it only really makes a difference if you have a decent wide gamut printer and a wide gamut source image too.

 

[Bifurcator]

OK, that's good to know. I don't remember if it was wide or not tho - that was 3 or 4 months ago.

1. Anyway, yes, all I did to the image was load and set it to 122 PPI and printed it.
2. Then I zoomed up in PS and did a screengrab of the massive square blocks you saw.
3. And then I found the same area on the printed output (deluxe "fine" Canon photo paper),
4. and both took a picture of it with a #14 close-up lens on a 200mm Macro,
5. and also put the printed photo on the flatbed and selected and scanned the same area (with all the scanner enhancement features turned off) at 2400 DPI.
6. I scaled everything in PS (making sure that no significant data was altered) so that everyone could see and compare the three samples:
   • 1 scan of the print,
   • 1 photo of the print,
   • and the screengrab of the actual pixels for that area.
   at a similar size and magnification.

You should be able to almost overlay them perfectly as is in layers for a good comparison.

Also with the scan and the photo of the print isolating the 3 (or 4 if you do CYMK) color channels of each and turning it into gray data is a little bit interesting. BTW this printer is a 5 cartridge printer but only 4 are used in printing photographs (C, Y, M, and K) the second (blacker) black I think is only used for text and vector graphics.

 

[Moglex]

Well, I dunno. I kinda think that's one-a-dem laws which don't translate into reality like the physics equations that say it's impossible for bumblebees to fly.

Let's look and use our own powers of observation:
The data is here. What conclusions do you draw?

I was hoping that someone with a longer track record here would answer this as they would presumably have more credibility but as no one has:

Firstly, I suppose it's always possible that Microsoft have introduced some entirely new way of talking to printer drivers that they have only made the people at Adobe and the like aware of. But that's extremely unlikely.

The reason that I state with almost 100% confidence that the dithering has been done in the photo program (whatever you may have told it to the contrary) is this:

When you send data to a printer (via its driver) from a Windows program you send it at the resolution of the printer.

Thus the printer driver and the printer are never aware of the resolution at which the source would be printed if you could magically cover the requisite amount of paper with the available pixels.

So the printer and its driver simply cannot perform interpixel interpolation/dithering as they never see the large pixels.

I hope that all makes sense.