Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Pixels vs Printing  (Read 4118 times)
Zenny
Newbie
*
Offline Offline

Posts: 22


« on: November 07, 2010, 03:47:09 PM »
ReplyReply

How does a pixel value gets translated to print size? Just wondering what is the maximum printing size recommended from for example, a 12MP digital image without loosing details? Is there any way to convert pixel to ideal print size?

Thanks!

zenny

***Support http://www.thehumanape.org ***
***See http://www.onelifephotos.com/drbista ***
Logged
Graham Mitchell
Sr. Member
****
Offline Offline

Posts: 2282



WWW
« Reply #1 on: November 07, 2010, 05:06:36 PM »
ReplyReply

There are no hard rules about this, but today I finished printing my folio and I must admit there was a very noticeable difference in quality between the Canon 5D files and the Sinar e54 files, at A3 print size (using Epson 3880 on Hahnemühle Photo Rag Pearl). There was actually more difference than I expected - the medium format digital files look amazing at that size, with no visible artefacts.
Logged

Graham Mitchell - www.graham-mitchell.com
Dick Roadnight
Sr. Member
****
Offline Offline

Posts: 1730


« Reply #2 on: November 07, 2010, 05:12:23 PM »
ReplyReply

How does a pixel value gets translated to print size? Just wondering what is the maximum printing size recommended from for example, a 12MP digital image without loosing details? Is there any way to convert pixel to ideal print size?
The pixel value is the R,G,B colour of an individual pixel.

What you are asking is what is the maximum print size for the pixel dimensions of a digital photographic file.

Printers have default or optimum printing resolution... Epson printers work best on multiple or sub-multiples of 360 pixels per inch, and for HP printers it is 300.

Choose a res from 72 to 1440 and work out the print size... or tell the print menu the size of your paper,  and it will tell you what res it would print at to achieve that size.

Generally it is thought that...
720ppi is so good that the printer is producing more detail than the human eye can perceive,
360 is Excellent
240 is good
180 is adequate...
72 is awful

It depends on the quality of the file or pixels, and good files, particularly those from Anti-Aliasing filter free and/or multi-shot cameras can be enlarged or up-sampled to lower original camera pixels per print inch resolutions.

Doubtless someone will tell you that you can print any file to any size if you do not look at the picture too closely, but people use bigger cameras so that they can print bigger pictures (without the pictures looking awful close up) or so that they can do more to the picture in post-production, or so that they get a good file out of the camera so that they do not have to do so much work on the file in post production.

Logged

Hasselblad H4, Sinar P3 monorail view camera, Schneider Apo-digitar lenses
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7252


WWW
« Reply #3 on: November 08, 2010, 01:27:28 AM »
ReplyReply

Hi,

I have a small write up on the issue:

http://echophoto.dnsalias.net/ekr/index.php/photoarticles/24-how-many-megapixels-do-we-need

Proper output sharpening matters a lot and so does viewing distance. I have a 70x100 cm print from an 10 MPixel APS-C on my wall. It is not sharp at neither 25 cm nor 50 cm viewing distance but turns to be OK at around arms length distance.

In A2 size I have mixed experience. I would say that 12 MPixels at A2 are OK. Sometimes I can tell A2-size prints apart between 12.5 MP APS-C and 24MP full frame, but the difference in print is much smaller than what I see pixel peeping on the files.

Best regards
Erik


How does a pixel value gets translated to print size? Just wondering what is the maximum printing size recommended from for example, a 12MP digital image without loosing details? Is there any way to convert pixel to ideal print size?

Thanks!

zenny

***Support http://www.thehumanape.org ***
***See http://www.onelifephotos.com/drbista ***
Logged

Dick Roadnight
Sr. Member
****
Offline Offline

Posts: 1730


« Reply #4 on: November 08, 2010, 03:55:55 AM »
ReplyReply

Hi,

I have a small write up on the issue:

http://echophoto.dnsalias.net/ekr/index.php/photoarticles/24-how-many-megapixels-do-we-need

Proper output sharpening matters a lot and so does viewing distance.
Pixel quality makes a difference... and if the detail is not there, no amount of sharpening will create it, and the better the file, the better it responds to sharpening and other post-production.

In the real world, bigger files take more space and more CPU time, but, it you have a fast computer, does it take less time to process big files (for 360 dpi), as they look good without much post-processing, or do you check every pixel?

So how big looks good through reading glasses when you use Apo-Digitars, 48Mpx multi-shot and Hex stitch?
Logged

Hasselblad H4, Sinar P3 monorail view camera, Schneider Apo-digitar lenses
Dick Roadnight
Sr. Member
****
Offline Offline

Posts: 1730


« Reply #5 on: November 08, 2010, 06:59:28 AM »
ReplyReply

actually a MF is more of a "problem" with skins,
I try to focus beyond the face, on the ears, neck or hair, leaving the face slightly OOF.
Logged

Hasselblad H4, Sinar P3 monorail view camera, Schneider Apo-digitar lenses
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7252


WWW
« Reply #6 on: November 08, 2010, 08:23:57 AM »
ReplyReply

Hi,

The original posting was more about "How large can I print from 12 MPixels" than about MF...

Best regards
Erik


Totally agree.
With good post prod, enhancing areas that give the impression of "details" we can fool the eyes to some extend. But there are limits. So more the file is big and quality more it will allow logically more extensive uses.
The Dslr files in fashion for example handle very well because actually a MF is more of a "problem" with skins, BUT...and this is an important but, for fabrics, clothes etc...MF capacity is way in front and it's clearly visible. It seems that the best compromise between Canon look and MF look is that D3x that sits right in the middle. Never worked with one so far so I can't comment any further on that, just it has this reputation.
So again, back to what James Russell was saying about the more tool we have, the more expression mediums we gain, there are situations in fashion where the use of MF would be really much more indicated, for example when a very strong focus has to be given in the textures. You will see the difference in magazine print, absolutly!

Logged

Phil Indeblanc
Sr. Member
****
Offline Offline

Posts: 1104


« Reply #7 on: November 08, 2010, 12:20:16 PM »
ReplyReply

How does a pixel value gets translated to print size? Just wondering what is the maximum printing size recommended from for example, a 12MP digital image without loosing details? Is there any way to convert pixel to ideal print size?

Thanks!

zenny


___
Viewing distance becomes a key factor. 

I have regularly made billboard ads that are 40feet tall and wider with a 300 dpi file that is no larger than a magazine size page.  The distance and interpolation used to enlarge files is rather forgiving.  For upclose or in the home, or gallery, it is strikingly different. Even for some distant Duratran work, the file needs to measure up.

Application is the question, hence pricing to do the work changes due to these factors in commercial work.
Logged

If you buy a camera, you're a photographer...
Christoph C. Feldhaim
Sr. Member
****
Offline Offline

Posts: 2508


There is no rule! No - wait ...


« Reply #8 on: November 08, 2010, 12:49:01 PM »
ReplyReply

Quote
Viewing distance becomes a key factor.

I must second that. The resolution of a perfect human eye is at something around 0.3 arc minutes.
If you'd view an image from 1.5 times the image diagonal you'd have a diagonal viewing angle of

2* arcsin (0.5/1.5) = 38,9 Degree = 2336 arc minutes

2336/0.3=7788 dots per diagonal. At 12 MP (3000*4000) you have a diagonal of 5000 pixels which already is not too far from that. People with slight eyesight problems might not notice at all.

Changing the viewing distance changes that equation. So - if you want to pixel peep a 2*3 meter print from 25 cm distance thats a different story.
Logged

Schewe
Sr. Member
****
Online Online

Posts: 5419


WWW
« Reply #9 on: November 08, 2010, 02:19:59 PM »
ReplyReply

The resolution of a perfect human eye is at something around 0.3 arc minutes.

You sure? I thought it was 1 arc minute for 20/20 vision (at least according to this Wikipedia article: Visual acuity).
Logged
Christoph C. Feldhaim
Sr. Member
****
Offline Offline

Posts: 2508


There is no rule! No - wait ...


« Reply #10 on: November 08, 2010, 02:42:22 PM »
ReplyReply

You sure? I thought it was 1 arc minute for 20/20 vision (at least according to this Wikipedia article: Visual acuity).

There sources are contradicting and not constant. I was reading 0.59 arc minutes from clarcvision for a line pair, but I know - the Wikipedia article tells a different story.
Source: http://www.clarkvision.com/articles/human-eye/index.html
I was referring to the stricter value, to be on the "safe side". But actually that raises an interesting point: The optimum human vision also depends on lighting conditions. E.g. recently I was watching old prints from Andreas Feininger and others in the Hamburg exhibition hall near the station and the lighting was far less than optimal - probably for not stressing the prints. I believe my vision under this conditions was closer to the 1' than 0.3' .....

So what shall we conclude from that ?  
Print out a sector star, carry it to the location of your friendly gallery or museum nearby and see what you can actually resolve yourself under the given conditions ... Sad

This could be the guide to what resolution is really needed ...
« Last Edit: November 08, 2010, 02:52:48 PM by Christoph C. Feldhaim » Logged

archivue
Sr. Member
****
Offline Offline

Posts: 415


« Reply #11 on: November 08, 2010, 02:53:34 PM »
ReplyReply

for big prints, there a huge difference between a 22 MP file from an Aptus 22 and digital lenses and a canon 5DII...

with the aptus 22 a 60x80 inkjet is really really nice !
Logged
Christoph C. Feldhaim
Sr. Member
****
Offline Offline

Posts: 2508


There is no rule! No - wait ...


« Reply #12 on: November 08, 2010, 02:57:07 PM »
ReplyReply

for big prints, there a huge difference between a 22 MP file from an Aptus 22 and digital lenses and a canon 5DII...

with the aptus 22 a 60x80 inkjet is really really nice !

This again tells us, that resolution is far from being the only predominant parameter defining IQ.
Logged

BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3454


« Reply #13 on: November 08, 2010, 02:59:12 PM »
ReplyReply

There sources are contradicting and not constant.

Hi Christoph,

Yes, although 1 arc minute is a more common rule of thumb. Of course the story is more complex, because contrast also plays a role. The human Contrast Sensitivity is a useful concept, and it shows an optimum around 8 cycles/degree (with a range of also resolvable spatial frequencies around it):
http://www.cg.tuwien.ac.at/research/theses/matkovic/node20.html#SECTION00535000000000000000

Cheers,
Bart
Logged
Christoph C. Feldhaim
Sr. Member
****
Offline Offline

Posts: 2508


There is no rule! No - wait ...


« Reply #14 on: November 08, 2010, 03:27:36 PM »
ReplyReply

Hi Christoph,

Yes, although 1 arc minute is a more common rule of thumb. Of course the story is more complex, because contrast also plays a role. The human Contrast Sensitivity is a useful concept, and it shows an optimum around 8 cycles/degree (with a range of also resolvable spatial frequencies around it):
http://www.cg.tuwien.ac.at/research/theses/matkovic/node20.html#SECTION00535000000000000000

Cheers,
Bart

Thats an interesting article. But the article, if I understand right does not suggest a resolution of 8 cycles per degree, but a maximum of sensitivity in that frequency range. This means we can distinguish 8 cyc/deg even at lowest contrast as opposed to the higher frequencies.
The function is said to be meaningless above 60cyc/deg because of an absolute limit. this would be 1 cycle / minute.

This article from the same website illuminates the problem of the resolution vs. viewing conditions:
http://www.cg.tuwien.ac.at/research/theses/matkovic/node16.html#SECTION00531000000000000000

I'm afraid in the end we land at the need of trying for ourselves under the conditions of image presentation, like I proposed (not 100% seriously) above.

What I also ask myself now is, if we can draw conclusions about optimum sharpening in post from that function.
Logged

Schewe
Sr. Member
****
Online Online

Posts: 5419


WWW
« Reply #15 on: November 08, 2010, 04:32:57 PM »
ReplyReply


I'm afraid in the end we land at the need of trying for ourselves under the conditions of image presentation, like I proposed (not 100% seriously) above.

What I also ask myself now is, if we can draw conclusions about optimum sharpening in post from that function.

Well, since all we have are our eyes, I guess that's what we need to do is look...Bruce Fraser in the 1st edition of Real World Image Sharpening determined that for an 8.5 x 11" print held at about 12 inches needed 355PPI to fool the eye into seeing continuous tone...at PPI less from that distance, he could see the lack of detail. Ironic that 355PPI is real darn close to Epson's 360 nozzles per inch. At distances further than 12 inches, the amount of resolution needed drops a lot real fast...so for real big prints you don't need a lot of resolution–unless the viewer is a photographer....Bruce used to say the normal viewing distance for a photographer is only limited by the length of his/her nose. (or as they get older, the strength of their reading glasses)
« Last Edit: November 08, 2010, 06:49:14 PM by Schewe » Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3454


« Reply #16 on: November 08, 2010, 04:35:35 PM »
ReplyReply

Thats an interesting article. But the article, if I understand right does not suggest a resolution of 8 cycles per degree, but a maximum of sensitivity in that frequency range. This means we can distinguish 8 cyc/deg even at lowest contrast as opposed to the higher frequencies.
The function is said to be meaningless above 60cyc/deg because of an absolute limit. this would be 1 cycle / minute.

Correct.

Quote
What I also ask myself now is, if we can draw conclusions about optimum sharpening in post from that function.

Well, in addition to boosting the limiting resolution (at the native printer resolution, e.g. 600 or 720 PPI), 8 cycles/degree seems a good angular resolution to give a boost, provided the viewing distance is known and constant. The full story is again more complex, because the system MTF ultimately plays a large role:
http://www.normankoren.com/Tutorials/MTF.html#Human_visual_acuity
http://www.imatest.com/docs/sqf.html#csf

Cheers,
Bart
Logged
Phil Indeblanc
Sr. Member
****
Offline Offline

Posts: 1104


« Reply #17 on: November 08, 2010, 05:41:06 PM »
ReplyReply

You see where this question ended up(some interesting stuff)?  How about some feedback as to what you are looking for..
:-)
Logged

If you buy a camera, you're a photographer...
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7252


WWW
« Reply #18 on: November 08, 2010, 10:20:22 PM »
ReplyReply

Hi,

If we have comparable optics a sensor of double height and having the same number of pixels would achieve about double MTF on a given feature size. MTF drops almost linearly with frequency. So micro contrast would be higher on the larger sensor would the lenses have equal MTF and used at the same aperture.

Excessive stopping down would reduce MTF on any camera.

The other issue is that it may be that resolution is fully utilized. To fully utilize sensor resolution we need dead on focus (which is notoriously hard to achieve without Live View), no vibration and optimum aperture.

Best regards
Erik

This again tells us, that resolution is far from being the only predominant parameter defining IQ.
Logged

Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad