Why a resized photo is no use in showing camera quality

[Mark1616]

Originally Posted by onlooker

During the discussion in this thread Chato made the comment about a "good guide to which P&S doesn't have heavy NR..." I now have an Olympus SP590 and have recently noticed that if I move to the P, A, S, M modes (and I think some others) I get 'NR' in a rectangle at the top of the monitor or view finder along toward the right hand corner. I have searched the camera manual but can find no reference to this information ....

s o o o o..... what is 'NR'?

Thanks for un-mystifying me.

Hi Bernice,

NR = noise reduction, which takes out some of the grain from a digital photo at higher ISO settings. It can also have the effect of softening a photo making it look fake/plastic depending on how strong NR is.

[onlooker]

Originally Posted by Mark1616

Hi Bernice,

NR = noise reduction, which takes out some of the grain from a digital photo at higher ISO settings. It can also have the effect of softening a photo making it look fake/plastic depending on how strong NR is.

Thanks Mark. Much appreciated.

[daved163]

Interesting thread - I wish you could persuade all posters to include EXIF information. The new HS10 thread, which I'm following owng to an imminent purchase, is a classic example and it's driving me mad. There are very few HS10 samples around and the handful posted on this site don't give even basic details.
What prompted me to join in is the discussion about monitors. The best way to view any picture on a screen is to use an old fashioned CRT screen. This is because, unlike flat screens the dots on a CRT bleed into each other which is more akin to a printed photograph. Graphics designers were the last stalwarts using CRTs for this very reason but even they seem to have ditched them in favour of LCD and plasma screens (probably the accountants telling them to save energy!)
The problem is that any and all parts of the finalising process lead to differing, and sometimes widely differing, results. Unfortunately there is just no perfect solution that fits all circumstances and all users. And, at the end of the day, beauty is in the eye of the beholder and producing a picture for technical disection is not the same as producing one as a piece of artwork. Sadly, I bet that not 5% of users calibrate their monitors so it's debatable whether or not any comments about colour rendition, at least, have any bearing on reality.
Dave

[TCav]

Originally Posted by daved163

... The best way to view any picture on a screen is to use an old fashioned CRT screen. This is because, unlike flat screens the dots on a CRT bleed into each other which is more akin to a printed photograph. Graphics designers were the last stalwarts using CRTs for this very reason ...

The phosphors in a CRT are laid out on the inside of the glass tube, the front of which is quite thick, which would cause some blending of the colors, but very little, and probably not enough to be noticeable. The reason graphics designers were the last hangers-on (some still are) is because the CRT has a lot more bandwidth for colors, so the display adapter is the limiting factor. With LCDs, the number of descrete colors is dependant on the monitor circuitry, not the the display adapter, and is a lot less than CRTs are capable of.

Originally Posted by daved163

... but even they seem to have ditched them in favour of LCD and plasma screens (probably the accountants telling them to save energy!)...

Actually, most companies don't pay for their electricity. They're tenants, and the building's owner pays for the electricity.

Originally Posted by daved163

... The problem is that any and all parts of the finalising process lead to differing, and sometimes widely differing, results. Unfortunately there is just no perfect solution that fits all circumstances and all users. And, at the end of the day, beauty is in the eye of the beholder and producing a picture for technical disection is not the same as producing one as a piece of artwork. Sadly, I bet that not 5% of users calibrate their monitors so it's debatable whether or not any comments about colour rendition, at least, have any bearing on reality.

True, but there's 'calibrate' and then there's 'CALIBRATE'. Often, just applying the ICC color profile for a device is enough to produce good results. And most manufacturers apply the ICC color profiles in their drivers without the user knowing about it. So it's not as big an issue as it once was.

[daved163]

Originally Posted by TCav

The phosphors in a CRT are laid out on the inside of the glass tube, the front of which is quite thick, which would cause some blending of the colors, but very little, and probably not enough to be noticeable. The reason graphics designers were the last hangers-on (some still are) is because the CRT has a lot more bandwidth for colors, so the display adapter is the limiting factor. With LCDs, the number of descrete colors is dependant on the monitor circuitry, not the the display adapter, and is a lot less than CRTs are capable of.
Actually, most companies don't pay for their electricity. They're tenants, and the building's owner pays for the electricity.
True, but there's 'calibrate' and then there's 'CALIBRATE'. Often, just applying the ICC color profile for a device is enough to produce good results. And most manufacturers apply the ICC color profiles in their drivers without the user knowing about it. So it's not as big an issue as it once was.

Forgive the cynicism about accountants - it comes naturally (possibly because I seem to be continually chased by mine!) It doesn't matter who owns the building, the electricity still has to be paid for one way or another.
I do agree about the video adaptor - it's the weak link. CRTs themselves, though, are analogue devices, hence the truer colour rendition, assuming an appropriate colour profile is used. And how many web uses know the colour profile of the input device, to which they should be matching their screen? Digital audio sounds OK to most of us because we can't discern the discrete changes but we are somewhat more sensitive to digitised visual input. The reason the colours in a CRT bleed is because when a phosphor dot glows it spreads its light across to neigbouring dots. The shadow mask will reduce the effect so that it is almost imperceptable, but almost is still not quite the same as not-at-all. Also, despite the refractive index if the front glass being as good as it can be there are still internal reflections which add to the effect. LCDs, particularly cheaper ones and those that are set up with interpolated instead their natural resolution also suffer from aliasing effects. How any web users have top-of-the-range monitors anyway?
The point is that I question the validity of some technical discussions about the quality of an image posted on the web, particular in regard to colour. In photgraphic terms it's a relatively crappy medium. But popular . The bottom line is that a photo either appeals or it doesn't, or it lies somewhere in between. A technically poor picture can still be very appealing and a technically brilliant picture can be thoroughly boring. Which is the more important?

[TCav]

Originally Posted by daved163

Forgive the cynicism about accountants - it comes naturally (possibly because I seem to be continually chased by mine!) It doesn't matter who owns the building, the electricity still has to be paid for one way or another.

Yes, the landlord pays for the electricity but does not pass the cost on to the tenants directly. Office buildings have one meter, and the landlord pays for the electricity. If each office space had it's own meter, then the meter would have to be adjusted every time a tenant expanded into adjacent vacant space, or when a tenant gave up some space to reduce costs, and every time a new tenant moved in, another meter would have to be added. The cost of repeatedly adjusting the meters would be prohibitively expensive. So landlords just absorb utility costs and pass the cost onto the tenants in the form of rent increases. That's why office buildings use florescent lights and gas water heaters. They're cheaper to operate.

So the accountants for the landlord don't tell the tenants to save electricity; they tell the landlord to raise the rents.

[pboerger]

I take exact opposite view. I believe 100 percent crop is the worst way to judge picture quality for most shooting. A 100 percent crop represents a gigantic print size that will almost never be used. A 40 percent crop, at most, is more realistic in terms of family pictures and that equalizes out a lot of cameras. My point is that for the vast number of casual shooters intending to simply enjoy the shot, you do not need to spend the kind of money required to get a good 100 percent crop. Your example picture is very good at the size you posted. Are you really going to enjoy a picture of an eyeball? This is truely posted with all due respect. I just don't get the 100 percent crop theory of judging cameras. I own a DSLR, but it is used primarily for low light and fast FPS. Instead of relying on crop, I zoom or get closer. At 8x10 with post processing, you can put some point and shoot high ISO shots against some very expensive cameras and not tell the difference. I just don't like to see people spend a lot of money to get way more camera than they need based on a 100 percent crop judgment.

[Alan T]

Originally Posted by pboerger

...exact opposite view. I believe 100 percent crop is the worst way to judge picture quality for most shooting...

You are correct if we interpret "judge picture quality" as meaning "assess fitness for purpose of an image or images".

However, the bulk of this thread has been about comparing image quality between cameras under various conditions. You correctly point out that for practical purposes, it's often impossible to see the difference. But if you can't see the difference you can't do a comparison.

I have owned 5 digicams since 2002, and am still regularly involved with images from another 3, which I purchased for family members. Those three, plus three of mine are still in regular use. Each has its good and bad points, which is why they're all used. All of them produce good results that are generally 'fit for purpose', and the cameras should be judged against each other according to that purpose. In many instances it would be difficult to judge among them.

BUT, most of this thread is about the image quality aspect of "camera quality". If you can't see see the difference you can't make a comparison.

Occasionally I want to make an 18x12 print from an artistically slightly cropped frame, to hang on my or someone else's wall. If I wish to compare quality between cameras, clearly I then need to to compare things that are different. Therefore it's necessary to inspect the image at a magnification that shows the differences. Hence the "100% crop".

Once an image has been resized, using whatever algorithm, the original relation between adjacent pixels has gone for good, and can't be meaningfully compared any more. Hence the 100% crop of the original file straight from the camera.

As for the "get closer" or "zoom in more" comment, wouldn't it be lovely if we always had the luxury of being able to do that? I often find I can't.

[camvard]

Thanks for this thread. My Canon SD4000/300HS takes pictures that look great when zoomed out. But when you zoom to 100%, you really notice how it lacks detail one would find even in older cameras in the same series.

pboerger says that a 100% crop represents a gigantic size that will never be used. As someone who will crop images to get only the part/detail I want, I disagree. Being able to crop the image to focus on the detail I want to highlight is a very relevant use case, at least to me.

[TCav]

Originally Posted by pboerger

I take exact opposite view. I believe 100 percent crop is the worst way to judge picture quality for most shooting. A 100 percent crop represents a gigantic print size that will almost never be used. A 40 percent crop, at most, is more realistic in terms of family pictures and that equalizes out a lot of cameras. My point is that for the vast number of casual shooters intending to simply enjoy the shot, you do not need to spend the kind of money required to get a good 100 percent crop. Your example picture is very good at the size you posted. Are you really going to enjoy a picture of an eyeball? This is truely posted with all due respect. I just don't get the 100 percent crop theory of judging cameras. I own a DSLR, but it is used primarily for low light and fast FPS. Instead of relying on crop, I zoom or get closer. At 8x10 with post processing, you can put some point and shoot high ISO shots against some very expensive cameras and not tell the difference. I just don't like to see people spend a lot of money to get way more camera than they need based on a 100 percent crop judgment.

What we're talking about here is that if you reduce an image so it can be posted here, the resampling of the image will eliminate the flaws in an image that will help to identify the problem. Posting a 100% image preserves a portion of the original image without any resampling. To identify things like motion blur, focus, softness, or CA, a 100% crop is useful, while a reduced image is not.