what's wrong with pixel peeping?

[Fluorite]

I'll tell what wrong with it, it's that most photographers produce prints, not 100% views on their computer.

I say this in reference to the editorial blog "Downsampling to reduce noise, but by how much". I think that the conclusions reached in this blog are correct, but only if you use your photos in this way (100% on screen).

This argument isn't anything new, it's been discussed many times, but I still wonder why DPreview doesn't take printed results into account in any of their testing, other than speculating that "you probably won't see the difference in print". What about testing the theory?

This blog might have been an interesting and relevant, but ignoring print quality in this kind of "argument" seems pointless!
--
fluorite

 

[MPCat]

The only people who care about what a 100% view of a photograph looks like are idiot photographers.

 

[Fluorite]

I wouldn't go as far as saying "idiot" because they might just be "beginners" or they have just forgotten that for most situations, the print is a very good test of image quality, because that's how the majority of photos are viewed.

Anyway, just as well there are web sites we can go for that kind of info. Should also mention, I think DPreview still has a valuable place in the world of camera reviews. I think it's good to understand the strengths and weaknesses of each of the review websites.
--
fluorite

 

[Mike_PEAT]

I'll tell what wrong with it, it's that most photographers produce
prints, not 100% views on their computer.

Except for the fact that this isn't a photography site, it's a camera review site...and that's all about pixel peeping!

It's like going to a site that talks about Ford Mustangs and complaining there's no discussions about road trips and all they talk about is 0-60 times and horsepower.

Personally I come to DPR for the tech talk about my camera, but I go to other sites and even a "brick & mortar" camera club for my photographic pursuits.

That being said there IS a "Samples & Gallery" forum here for pictures.

 

[kwa_photo]

Everyone one of us has pixel peeped at sometime. Then I think back to the film days. Did I look at my prints with a magnifying glass? I did use a loupe on my slide film, but even then, it was no where close to doing what we do with digital peeping.

I've had some noisy 100% digital files print buttery smooth up to 16x20", after that, it was pointless to peep.
--
Ken

 

[Russell Ervin]

If people didn't pixel peep, all the photos would have the same quality. Then people would have actually look at the photo itself to judge if it was any good. BUT...then they would only have themselves to blame if it wasn't a good photo..heheh.

Anyone get my point here? People should have more pride in their photos, instead of their tools. Unfortunately on DPR this seldom true.

I have photos mounted on my walls. When people ask me about them, I can honestly say, I remember the place, I remember what I was thinking about when I took the photo and why, but I usually can never remember which camera I was using at the time.
Russ

--



http://www.flickr.com/photos/quietrvr/
Gear=A camera with a lens.

 

[Soenda]

I'll tell what wrong with it, it's that most photographers produce
prints, not 100% views on their computer.

Except for the fact that this isn't a photography site, it's a camera
review site...and that's all about pixel peeping!

Here's text from the home page:

"Welcome to Digital Photography Review

Here you will find all the latest digital photography and imaging news , reviews of the latest digital cameras and accessories, the most active discussion forums, a large selection of sample galleries, a digital camera database and buyers guide and the most comprehensive database of digital camera features and specifications. "

While gear is a main feature here, it's not named Digital Camera Review or Digital Photography Gear Review.

Until Phil says different, it's a photography site. ;^D

--
http://www.pbase.com/soenda

 

[KEITH-C]

The whole point about pixel-peeping is that like for like comparisons are still valid as a means of seeing which camera renders the best image at high magnifications.

Small differences in resolution are less apparent in prints where other factors such as contrast & dynamic range & colour accuracy are much easier to judge. The correct balance of these factors combines to create 'dimensionality' which isn't so easy to define with numbers.

Keith-C

 

[Brunobl]

I'll tell what wrong with it, it's that most photographers produce
prints, not 100% views on their computer.

Except for the fact that this isn't a photography site, it's a camera
review site...and that's all about pixel peeping!

It's like going to a site that talks about Ford Mustangs and
complaining there's no discussions about road trips and all they talk
about is 0-60 times and horsepower.

Personally I come to DPR for the tech talk about my camera, but I go
to other sites and even a "brick & mortar" camera club for my
photographic pursuits.

That being said there IS a "Samples & Gallery" forum here for pictures.

Except that the OP wasn't talking about samples, picture-sharing, "see my image", "what should I photograph" or other photographic pursuits. Instead, He made a point about printed photographs being a more relevant final camera IQ indicator than the 100% image in the computer monitor.

One may or may not agree with that assessment, but nonetheless it IS a gear-related view that belongs in a gear-oriented forum just as much as shutter lag, high-ISO noise, or OVF versus EVF.

--
Best regards,

Bruno Lobo.



http://www.pbase.com/brunobl

 

[Barry Fitzgerald]

This blog might have been an interesting and relevant, but ignoring
print quality in this kind of "argument" seems pointless!

Problem is, I tried it in print, and I cannot disagree wtih the conclusions that dp came to. It really is NOT the magic bullet some folks have suggested.

 

[Jeppe76]

I'll tell what wrong with it, it's that most photographers produce
prints, not 100% views on their computer.

Except for the fact that this isn't a photography site, it's a camera
review site...and that's all about pixel peeping!

What? Nonono.

Pixel peeping has very little to do with reviewing photographic equipment. Evaluating sensor performance does matter - but simply looking at 100% crops and using that to divine which camera is better is useless.

The sensor resolution is getting so high that the requirements to the glass will increase to a point where the glass will have to be exteremly expensive and heavy in order to do the sensor justice. So, people will gradually have to admit that dynamic range and tonality matter more and are cheaper to improve than pure resolution.

Some of us are starting to see the light. Hopefully Phil will join us.

--
http://flickr.com/photos/jeppe-photos/

 

[ejmartin]

Nothing is "wrong" with it, so long as you keep in mind the context in which the image is being viewed (in particular the scale in the image that is being viewed), and what conclusions are appropriate to draw from what is seen.

However, when these caveats are ignored, and pixel-peeping is regarded as some sort of absolute measure of an image rather than its measure at a particular scale, problems arise. Particularly pernicious are comparisons made between cameras of different pixel count via pixel-peeping, as the test protocol at this site is wont to do. Comparisons of the properties of images of the same scene from different cameras made by comparing them at different scales are inherently flawed, and yet are used to support a misguided agenda that seeks to suppress technological advances in camera engineering.

The latest foray in this crusade is Phil's blog post, which purports to refute the notion "that we're being unfair on high megapixel cameras which show higher levels of noise than lower megapixel cameras". By which he means higher noise std dev, which already displays a lack of understanding of what the std dev measures. Phil proceeds to set up a straw man -- a set of sample images that have little to no noise power at high frequency (pixel-level scales) -- and then knock it down by showing that downsampling, whose effect is to remove high frequencies from the image, doesn't remove much noise. Well, duh! It wasn't there to begin with.

Never mind of course the fact that the whole blog post has little to do with the object of the crusade (the evils of high MP cameras), since it makes no comparison between cameras with the same levels of noise at comparable scales, but having different pixel counts. Anyone interested in the test that should have been done to address that issue may be interested in some of the posts in a recently filled thread, in particular

http://forums.dpreview.com/forums/read.asp?forum=1000&message=30165429
http://forums.dpreview.com/forums/read.asp?forum=1000&message=30169513

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[Fluorite]

I think dynamic range and tonality are commented on here in the reviews, but maybe not given the attention they deserve.

I think the methods used on this site are skewed to "scientific analysis" more than "impression", even though there is some subject comment given.

My intention in this thread is just to point out that analysis of prints (rather than speculative comment) would be a useful addition to what DPreview does here.
--
fluorite

 

[Fluorite]

I don't know if the lomo camera is still popular, but you might know of it's use by some very talented photographers for its unique visual look.

To me this is a very interesting idea, because the lomo camera, if reviewed by DPreview, would be totally panned for it's lack of resolution, severe corner shading, lack of features, etc etc. But this camera was very successfully used by some people to do great work.
--
fluorite

 

[Fluorite]

Thanks again, good comments for us to digest....
--
fluorite

 

[GordonBGood]

Comments as follows:

Nothing is "wrong" with it, so long as you keep in mind the context
in which the image is being viewed (in particular the scale in the
image that is being viewed), and what conclusions are appropriate to
draw from what is seen.

However, when these caveats are ignored, and pixel-peeping is
regarded as some sort of absolute measure of an image rather than its
measure at a particular scale, problems arise. Particularly
pernicious are comparisons made between cameras of different pixel
count via pixel-peeping, as the test protocol at this site is wont to
do. Comparisons of the properties of images of the same scene from
different cameras made by comparing them at different scales are
inherently flawed, and yet are used to support a misguided agenda
that seeks to suppress technological advances in camera engineering.

The latest foray in this crusade is Phil's blog post, which purports
to refute the notion "that we're being unfair on high megapixel
cameras which show higher levels of noise than lower megapixel
cameras". By which he means higher noise std dev, which already
displays a lack of understanding of what the std dev measures. Phil
proceeds to set up a straw man -- a set of sample images that have
little to no noise power at high frequency (pixel-level scales) --
and then knock it down by showing that downsampling, whose effect is
to remove high frequencies from the image, doesn't remove much noise.
Well, duh! It wasn't there to begin with.

Never mind of course the fact that the whole blog post has little to
do with the object of the crusade (the evils of high MP cameras),
since it makes no comparison between cameras with the same levels of
noise at comparable scales, but having different pixel counts.
Anyone interested in the test that should have been done to address
that issue may be interested in some of the posts in a recently
filled thread, in particular

http://forums.dpreview.com/forums/read.asp?forum=1000&message=30165429
http://forums.dpreview.com/forums/read.asp?forum=1000&message=30169513

That was a very interesting thread and I'm sorry I missed it. In reading it, I found the answer to a question I posted to you on another thread about noise power and the effect of downsampling.

I see that the argument that different sized sensors need to be evaluated on an image basis went by most participant, and degraded into Phil's discussion on how to evaluate different sized and pixel density sensors by downsampling, which he saw as having no effect on noise. I think that you made most of the points I would have made, but there are a few things that perhaps should have been emphasized in your final posts to that thread, as follows:

1) That standard deviation is no longer a very good measurment of relative noise when compared between differing amounts of averaging/noise reduction/downsampling already applied and even though the noise power may be the same.

2) That downsampling is effective as a comparison tool when the right downsampling algorithm is used (including the right blur filter to the new downsampled real comparative resolution) even when the higher frequencies have already been stripped due to this again changing the nature of the frequencies or power envelope.

3) That downsampling could be done at the raw data level in which case it is exactly as effective as calculated for any raw data that has not already had noise filtering applied.

Points you did summarize are very good, and such points you made about pre-filtering before down-sampling and choise of down-sampling algorithms are exactly what I would have said.

You basically answered the questions of Phil's blog as to why downsampling didn't remove the noise in that the noise had already been stripped as if these images were from a camera that had already applied noise filtering and that proper algorithms and techniques need to be used. What I would like to see is something provided to Phil that he could post as an extension to his blog page that would show him and his public how proper comparisons of cameras of different pixel densities could be compared effectively. You made lots of posts showing all the pieces, but perhaps we should have a post showing how the Canon G10 isn't really worse than the G9 or or G7 (although I regret it doesn't have native raw, although I think the CHDK patch now works with it). Any takers? Or maybe I'll take a stab at it myself. In that way, we are addressing the real issue as you say.

It is in studies such as this that Pixel Peeping does have a role when used with proper perspective as you say.

Best regards, GordonBGood

 

[Patrick T. Kelly]

There have always been people who did things I don't do that occasionally I found useful. A man I worked with saved every piece of paper he ever received from the organization. So, if I needed a memo from five years before, he would have it.

I've have neither the skill nor the interest to do some of the technical investigations that some here find fascinating. Occasionally, though, they hit on something I find useful.

The world is divided into those who measure benchmarks and those who don't.
--
Patrick T. Kelly
Oaxaca, Mexico

 

[MusicDoctorDJ]

Most of the pixel peepers are people who really don't know what good image quality is, and no matter how big image sensors become or how good image quality gets, they will find fault with it and will never be happy.

Unfortunately, with digital imaging many people only use their images on a computer screen and never really see what their digital cameras are capable of image quality-wise.

No matter what anyone here says . . . photography is still intended to be printed as the final outcome.

And if you don't print any digital images, you are not really seeing what your digital camera can do.

--
J. D.
Colorful Colorado



Remember . . . always keep your receipt, the box, and everything that came in it!

 

[bm bradley]

Except for the fact that this isn't a photography site, it's a camera
review site...and that's all about pixel peeping!

BINGO!!!!!

 

[Fluorite]

The whole point about pixel-peeping is that like for like comparisons
are still valid as a means of seeing which camera renders the best
image at high magnifications.

Do you mean on-screen or in print?

Small differences in resolution are less apparent in prints where
other factors such as contrast & dynamic range & colour accuracy are
much easier to judge. The correct balance of these factors combines
to create 'dimensionality' which isn't so easy to define with numbers.

I completely agree. I think one of the reasons DPreview hasn't gone down the path of evaluating prints is that it probably isn't possible to measure the results, like you can with digital image data. I also agree with you that these minor differences are likely less visible in prints, which makes other factors, such as camera features, AF performance, ergonomics etc. more important when deciding on a camera to purchase.

Keith-C

--
fluorite