Image Sharpness - an investigation

[Sid Garige]

Well.. Here is how the whole story started. I posted an image on avian forum in which I thought both birds were sharp. Link here
http://www.birdphotographers.net/forums/showthread.php?t=42612

I have a Sony Vaio 13.3" laptop (1280x800 resolution and a pixel size of 0.2238 mm). I use a 19" Dell LCD monitor (1440 x 900 resolution and a pixel size 0.2842 mm) as external display primary display and extend desktop to 13" screen. So technically my 19" LCD monitor is the primary display unit.

Today I received quite a few critics saying both subject heads are too soft and Arite posted both heads are petty sharp. I did some research trying to understand why same image looked sharp to me and Arite not to others.

I studied the image of 4 different monitors with different resolution and pixel size. Later on I followed instruction from an online tutorial with calibration images and calibrated my monitor and I was stunned with the results. I will post link to the tutorial at the end of the message.

Monitor 1.
Dell laptop Screen -17" Resolution - 1440 x 900 Pixel size - 0.2543 mm
Image looked really soft. Did not see any details in the head.

Monitor 2
Dell LCD Screen -21" Resolution - 1680 x 1050 Pixel Size - 0.2692 mm
Better than 17" but not as good as 19" Monitor.

Monitor 3
Dell LCD Screen -19" Resolution - 1440 x 900 Pixel Size - 0.2842 mm
More or less same as what I saw on my screen.

I called couple of my co-workers and asked their opinion. Both of them express the same opinion.

Monitor 4
Sony Vaio LCD Screen - 13.3" Resolution - 1280x800 Pixel Size - 0.2238 mm
I was shocked to see image looked much sharper than the one on 19" monitor.

I came to a concussion that pixel size will effect the image sharpness. One can not judge image sharpness based on jpg image unless they have same resolution and pixel size.

While doing some more research on the same topic I ran into the following tutorial
http://www.lagom.nl/lcd-test/sharpness.php

I followed the instructions and calibrated my 13" laptop screen to show complete gray screen shown in second image. Another surprise, my image looked much sharper than before. Not only mine but I could see difference in every image.

Artie saw the same image at 21" screen 1680 X 1050 resolution with pixel size 0.2692 mm sharper than others.

In conclusion, looks like resolution, pixel size and sharpness settings governs how good you see original image. Unless we are using the same resolutions, pixels and sharpness settings we can not accurately judge image sharpness.

Any thoughts on this?

PS: I completely agree with anyone who says my image is not sharp. Objective behind this process is to figure out why there are so many different responses to one single image with respect to sharpness.

-Sid

[Axel Hildebrandt]

I think it is possible to judge if an image is soft, sharp or oversharpened. It's not an exact science but the differences should not be too significant. When I compare images on my 15'' laptop (1440x900) and on my 24'' monitor (1920x1200) I don't see a difference in sharpness that is worth mentioning.

One thing I noticed when all images looked soft was that I had to go to view > zoom > reset in Firefox. I had changed the size without noticing and this had a big impact.

[John Chardine]

Sid- If the original ibis image was sharp to you but many others saw it as soft, that suggests to me that your monitor was set too sharp rather not not sharp enough. The adjustments you have made have made your monitor sharper, which means you may now routinely undersharpen images for BPN posting and the web in general. Does that make sense?

I viewed the original image on my Apple Cinema display (1280 x 1024) and parts of the ibises including the heads do look soft to me. Unfortunately I have no way to adjust sharpness on my monitor as shown in the useful web link you provide.

[Sid Garige]

I think it is possible to judge if an image is soft, sharp or oversharpened. It's not an exact science but the differences should not be too significant. When I compare images on my 15'' laptop (1440x900) and on my 24'' monitor (1920x1200) I don't see a difference in sharpness that is worth mentioning.

One thing I noticed when all images looked soft was that I had to go to view > zoom > reset in Firefox. I had changed the size without noticing and this had a big impact.

Axel,
I agree with you to an extent but not entirely. As a test study the checked pattern of same image from this tutorial on two both your monitors at a distance of 15 feet. Tell me if you see the same image.
http://www.lagom.nl/lcd-test/sharpness.php

[Sid Garige]

[QUOTE=John Chardine;315436]Sid- If the original ibis image was sharp to you but many others saw it as soft,
Agree with you on this.

[John]that suggests to me that your monitor was set too sharp rather not not sharp enough.

[Sid] Sharpness is not only governed by sharpness setting on your display. I assume it is also governed by pixel size and resolution. I took same image and compared on two Identical dell boxes and they looked same. As soon as I changed the resolution, I did see difference. Of course, changing the sharpness did make difference too.

[John] The adjustments you have made have made your monitor sharper, which means you may now routinely undersharpen images for BPN posting and the web in general. Does that make sense?

[Sid] So what is web and BPN sharpness in general? We dont use any calibration sheet so everyone is working on their own standards.

[Desmond Chan]

I looked at your posted photo. On my monitor, I would say it looks a bit soft, too. I use a CRT monitor.

Artie says "pretty sharp." What does "pretty sharp" mean? It is sharp as it is not out-of-focus but could use some increase in contrast (which is what sharpening does actually) ?

[Axel Hildebrandt]

I think a common sense approach works best. Even if there are slight differences in sharpness, on my monitors the differences are negligeable. I describe images as soft that lack fine details.

[Sid Garige]

I looked at your posted photo. On my monitor, I would say it looks a bit soft, too. I use a CRT monitor.

[Sid] Desmond, I dont know much about CRTs and I would not comment on how it shows on CRT.

[Desmond] Artie says "pretty sharp." What does "pretty sharp" mean? It is sharp as it is not out-of-focus but could use some increase in contrast (which is what sharpening does actually) ?

[Sid] Contrast is the ratio of lightest part to darkest part of image. It is not related to sharpness. Sharpness is defined by boundaries between zones of different colors. For example on adult ibis head how good brown feathers are differentiated from black shadows caused by light.

[Sid Garige]

I think a common sense approach works best. Even if there are slight differences in sharpness, on my monitors the differences are negligeable. I describe images as soft that lack fine details.

Axel,
Agree with you on common sense. But what is the guarantee that your sharpness settings on your monitor is same as rest of the community? May I ask how did the sharpness chart showed up on your screen?

-Sid

[Axel Hildebrandt]

Axel,
Agree with you on common sense. But what is the guarantee that your sharpness settings on your monitor is same as rest of the community? May I ask how did the sharpness chart showed up on your screen?

-Sid

Most of the time people on the forum agree whether an image is sharp or not and sometimes there are borderline cases, which is good enough for me.

I will check the link when I have both monitors next to each other.

[Sid Garige]

More interesting stuff..

I went back and did some more research. I connected my laptop in dual monitor mode and put the same picture on both screens. Then I did a screen print and opened the image in photoshop. Cropped both images very carefully with grid and compared histogram to histogram. They is some difference just based on resolution and pixel size. I will repeat it at work tomorrow to see what happens with a 21" screen.

[Desmond Chan]

[Sid] Contrast is the ratio of lightest part to darkest part of image. It is not related to sharpness. Sharpness is defined by boundaries between zones of different colors.

As far as I know, sharpness has a lot to do with contrast. Sharpness is what your eyes see. Sharpness does not exist in time and space. A boundary itself can be seen as sharp and it can also be seen as not sharp. A boundary does not appear to be sharp simply because it is a boundary. Just because it is between two different colors does not mean it will appear sharp. A boundary between black and white always look sharper than a boundary between 18% gray and 25% gray. A boundary between yellow and blue will also look sharper than the one between green and blue for the reason that the contrast between yellow and blue is higher than that between green and blue. What sharpening does in photoshop and the like is to increase the contrast along both sides of a boundary for the reason that increasing contrast will enhance the perception of sharpness. So, bottomline is, sharpness has a lot to do with the contrast between both sides of the boundary. Sharpness is a perception and it is subjective.

Even if you can show a lot of scientific data to prove that the sharpness of your photo has met the "sharpness requirements" based on those data/calculations/experiments, etc., I don't know how it could change what is seen by other people. If that person does not see a certain image as sharp, all the sharpness data about that image means nothing, IMO.

[Simon Bennett]

Sid, Could this issue relate to sharpening/downsizing formula you use to prepare you images for web presentation? I use Manyks (http://fc07.deviantart.com/fs9/f/2006/044/e/8/Manyk_Web_Sharpening_Action.pdf) methods which seems to invariably produces crisp web sized presetations. Simon

[Sid Garige]

Even if you can show a lot of scientific data to prove that the sharpness of your photo has met the "sharpness requirements" based on those data/calculations/experiments, etc., I don't know how it could change what is seen by other people. If that person does not see a certain image as sharp, all the sharpness data about that image means nothing, IMO.

Desmond,

Did you see PS at 10.35PM

"PS: I completely agree with anyone who says my image is not sharp. Objective behind this process is to figure out why there are so many different responses to one single image with respect to sharpness"

First of all, I did not show any scientific data to prove that sharpness requirements were met. Also, I dont want to change any viewers opinion. All the scientific data I showed in the thread is related to JPG quality with respect to resolution and pixel size and how it effects sharpness. It is important for me to understand the scientific reason behind it to produce better images.

[Sid Garige]

Sid, Could this issue relate to sharpening/downsizing formula you use to prepare you images for web presentation? I use Manyks (http://fc07.deviantart.com/fs9/f/2006/044/e/8/Manyk_Web_Sharpening_Action.pdf) methods which seems to invariably produces crisp web sized presetations. Simon

Simon,

Thanks for the link. Let me read it.

-Sid

[Desmond Chan]

It is important for me to understand the scientific reason behind it to produce better images.

All right, thanks for the clarification.

[arash_hazeghi]

Sid,
What you say is absolutely right, sharpness depends on the resolution as well as the contrast of the screen, I have a Lenovo 15" screen with array 1600X1080 resolution and an HP as well as a DELL 21" screens (S-IPS panels) with 1600X1200 resolution. In the past I used to edit some pics with my laptop and then post it to the internet and find them very soft when looking at my 21" LCDs. I now use only desktop screens, but even with this I posted an image in the "Eager to Learn" forums which I am attaching here, and many thought it was soft, it looks razor sharp on my screen.
http://www.stanford.edu/~ahazeghi/Photos/birds/tern%20feeding%20chick.jpg

Now the question is even looking through the same screen sharpness is still subjective. It is hard to find a define a common standard, like audio quality that is another subjective factor.

[John Chardine]

snip

[Sid] So what is web and BPN sharpness in general? We dont use any calibration sheet so everyone is working on their own standards.

I agree Sid. This is subjective. However, what I have learned from critiques from excellent BPN photographers is that they like very sharp images, starting with the eye and head, but not sharpened to the point that artifacts such as halos and "that sharpened look" are present. If you try for this result, there is a surprisingly narrow band of tolerance between a soft and an oversharpened image.

[John Chardine]

Sid,
What you say is absolutely right, sharpness depends on the resolution as well as the contrast of the screen, I have a Lenovo 15" screen with array 1600X1080 resolution and an HP as well as a DELL 21" screens (S-IPS panels) with 1600X1200 resolution. In the past I used to edit some pics with my laptop and then post it to the internet and find them very soft when looking at my 21" LCDs. I now use only desktop screens, but even with this I posted an image in the "Eager to Learn" forums which I am attaching here, and many thought it was soft, it looks razor sharp on my screen.

Now the question is even looking through the same screen sharpness is still subjective. It is hard to find a define a common standard, like audio quality that is another subjective factor.

Here's a thought that dawned on me after posting an image a while back. It was of a teal at sunset and the "catchlight" in the eye was the lit horizon and the setting sun. The horizon was reflected in the water and so there were two. The image was very sharp IMO but the catchlight gave the impression of a soft image and a few people made general comments to this effect, without mentioning the eye specifically. What I learned from this is that we use eye sharpness as a general indicator of the sharpness of the whole image and if the catchlight is "funky" in some way, it can give the impression of a soft image when it is really sharp. I think this is what's happening in the tern image with the catchlight in the eye of the adult not being circular.

[Sid Garige]

Sid, Could this issue relate to sharpening/downsizing formula you use to prepare you images for web presentation? I use Manyks (http://fc07.deviantart.com/fs9/f/2006/044/e/8/Manyk_Web_Sharpening_Action.pdf) methods which seems to invariably produces crisp web sized presetations. Simon

Simon,
Tutorial talks about sharpening images for web. This tutorial is new to me and I tired it on couple of images as a quick test. Will spend more time later and let you know how it goes.

In this thread my interest is mainly in studying image sharpness with respect to resolution and pixel size. I will put little more time on this and let you know if Manky action makes any difference in this regard.

Thank again

-Sid

[Sid Garige]

I agree Sid. This is subjective. However, what I have learned from critiques from excellent BPN photographers is that they like very sharp images, starting with the eye and head, but not sharpened to the point that artifacts such as halos and "that sharpened look" are present. If you try for this result, there is a surprisingly narrow band of tolerance between a soft and an oversharpened image.

John,

First of all thanks for the reply.

I do understand what I am covering a topic which does not have enough supporting material to say "right" or "wrong". Every one likes a sharp image and dislikes a over sharpened image. My attempt is to find out how to make a sharp image which is accepted by majority. I learned exposure, flash techniques and post processing as a part of this process and I believe I handle it alright. But, this is the first time I discovered post processing on a 13" Laptop screen is not a right choice (my personal opinion) when majority are on 19" and 21" because of difference in pixel size. This is just another attempt to understand my equipment and use it in right way to produce better images.

-Sid

[Sid Garige]

Arash,

Here are your techs
15.4" 1680 x 1050 128.6 ppi 0.1974 mm (pixel size)
21.3" 1600 x 1200 93.9 ppi 0.2705 mm (Pixel size)

Do you see the difference in pixel size? Not a surprise to me you see soft image on other monitor.

[Jay Gould]

http://www.lagom.nl/lcd-test/sharpness.php#sharpness-d.png


LCD monitors often have a "sharpness" control, which can emphasize or deemphasize boundaries between light and dark areas. Ideally, it does neither, unless you like a bit of fuzziness or like small letters to have more contrast. If the sharpness setting is correct (i.e., neutral), all blocks in the image below will blend in when you look at the screen from a distance or while squinting your eyes. If the gamma of your display is correct as well, then the central disc will blend in, but we will have a better look at this in the next test. If the sharpness setting is far off, you may be able to see fringes around the white and black bars in the center.

On a laptop, all I can do is adjust brightness.

At what "distance" do you believe you should be to do perform the "blend" test?

Wearing contacts and magnifying glasses for reading, at 1 1/2 feet, I can see all of the lines etc. sharply; at four feet with my reading glasses everything starts to blend. If I remove my reading glasses at four feet, everything is sharp and I can see all of the boxes and lines.

Laptop monitors, unlike stand-alone LCDs, do not have many controls.

[Sid Garige]

On a laptop, all I can do is adjust brightness.

At what "distance" do you believe you should be to do perform the "blend" test?
Wearing contacts and magnifying glasses for reading, at 1 1/2 feet, I can see all of the lines etc. sharply; at four feet with my reading glasses everything starts to blend. If I remove my reading glasses at four feet, everything is sharp and I can see all of the boxes and lines.

Laptop monitors, unlike stand-alone LCDs, do not have many controls.

Honestly, I dont know Jay. I experienced the same. At 1&1/2 feet I see all lines. At around 8 feet I see a blend. Like I said before, I am covering a topic which does not have enough supporting material to say "right way" or "wrong way". All I am trying to do is figure out a method that works for me.

[Roger Clark]

Arash,
It appears to me that your image has the focus in front of the birds. If you look at the details in the ground, they are very sharp close to the bottom of the frame, but the top edge of the (is that dirt covered cement?) dirt, just before the water, which is slightly behind the birds is slightly out of focus. The birds are closer to that position than to the dirt in front of them. That means the focus is slightly in front of the birds.

Roger

[Roger Clark]

A lot of what is being discussed in this thread is accutance, not sharpness. Some of these issues were covered in a thread from last year:

Important Sharpening Information!
http://www.birdphotographers.net/forums/showthread.php?p=133264#poststop

Unsharp mask increases edge contrast, accutance, and does not actually sharpen. It gives the impression of improving sharpness. But other methods are needed to actually improve sharpness and the above thread discusses some.

Regarding monitors, there was a thread discussing that too but I can't find it (too many hits on monitors). It's not just dot pitch of the monitor, but its modulation transfer function. CRTs generally have lower MTF than LCDs, and when I switched from CRT to LCD (I use 30-inch Dell S-IPS monitors) a lot of my web images looked over sharpened. I had to go back and redo them (probably have more to do). So there will be natural variations regarding what people see in your images and there is nothing anyone can do about it (same with color and calibrated versus non-calibrated monitors; same with contrast with varying gamma).

True sharpness (as opposed to accutance) would probably best be defined by the modulation transfer function of an image. For any real image you can't derive the MTF because you need to know the true answer as a reference in order to measure the drop in MTF for your image, and unless you are imaging test charts, you don't know that. So then it gets to be subjective. But at least you can know what your equipment is able to deliver.

For example, see:

DigitUnderstanding image sharpness part 1:
Introduction to resolution and MTF curves
http://www.normankoren.com/Tutorials/MTF.html

Roger

[arash_hazeghi]

Arash,
It appears to me that your image has the focus in front of the birds. If you look at the details in the ground, they are very sharp close to the bottom of the frame, but the top edge of the (is that dirt covered cement?) dirt, just before the water, which is slightly behind the birds is slightly out of focus. The birds are closer to that position than to the dirt in front of them. That means the focus is slightly in front of the birds.

Roger

Hi Roger,
both of the birds and the mud on the platform the Terns are standing on are in focus, however due to strong light and lack of texture on Tern feathers people get the impression that the birds are soft.

Here is a closer look at the RAW

http://www.stanford.edu/~ahazeghi/Photos/birds/crop.jpg

[Nagesh Mula]

Sid, Could this issue relate to sharpening/downsizing formula you use to prepare you images for web presentation? I use Manyks (http://fc07.deviantart.com/fs9/f/2006/044/e/8/Manyk_Web_Sharpening_Action.pdf) methods which seems to invariably produces crisp web sized presetations. Simon
Simon, it seems the Manyk's link that you provided is no more active. I would love to go through this link as I too am facing similiar problem.

Thanks.

[Simon Bennett]

Nagesh, The Manyk link seems to be working now: http://fc07.deviantart.com/fs9/f/2006/044/e/8/Manyk_Web_Sharpening_Action.pdf

[annmpacheco]

Thank you Sid for sending me this post, it is very helpful