Details, DP2M and Pixel-Peeping ...

Started 2 months ago | Discussions thread
OP xpatUSA Forum Pro • Posts: 15,495
Re: Details, DP2M and Pixel-Peeping ...

D Cox wrote:

xpatUSA wrote:

We talk about "detail" a lot in this Forum, perhaps because the Foveon is quite good at recording said detail.

I often wonder what "detail" is. In my own mind, detail is objects no smaller than one pixel at the sensor. My DP2M has a 5um pixel spacing and a 30mm focal length lens but how does that translate to objects in the scene? By which I mean that a leaf shot at minimum focusing distance is hardly a "detail" - but that same leaf shot from a mile away is beyond even the resolution of the Mighty Merrill.

Easy enough really, if we now talk about angles. The DP2M pixel at 30mm subtends an angle of about 0.6 minutes of arc, i.e. 0.1666 milli-radians (mrad). On the other side of the lens, i.e. in the scene, such a detail is easily calculated. An object 1 meter away will be 0.167mm large, 100 meters away the same 1-pixel detail will be 16.67mm large. Piece of doddle.

This kind of calculation is useful when you have to photograph a large painting.

Although I have a DP2M, shooting anything with such detail is quite a challenge for me such that I find it necessary to go beyond 100% zoom in post to make sure my shots are at least half-decent. Of course, others here are so good at capturing "detail" that they need no such assurance and they cannot understand why I pixel-peep even at 100%.

So, in defense of pixel-peeping, I took a virtual shot of small piece of virtual mesh with exactly one-pixel dimensions with a virtual DP2M. Here is that virtual "shot":

So, how "good" is that mesh detail? Anybody?

No peeping now, y'hear? - but feel free to use a real magnifying glass on your monitor ...

More to come later ...

Good stuff. More please.

Thank you, Don.

With this post, I hope to show that image "detail" can not be assessed without zooming to more than 100% - a practice frowned upon by some people here (you know who you are ).

In the above image and in the post view, the virtual mesh looks like a gray square on my monitor. Application of a 10X jeweler's loupe reveals ... a gray square made up of RGB sub-pixels. No prob . . . go to "original size" ... Nope, even viewing with a powerful pocket microscope reveals only a hint of a mesh with very low contrast.

At this point, the anti-peeper would have no idea what the quality of that detail is, because he "never views at more than 100%".

On the other hand, the peeper can zoom in by integer factors, X1,X2,X3 etc. and, assuming that the stupid viewer doesn't apply smoothing, thereby see the detail pixels enlarged upon the monitor screen. Note that the image itself remains unchanged - so what you see is what you got. Amid howls of horror, here is my original virtual mesh shown at 50X, i.e. 5,000% zoom:

Extreme pixel-peeping shows clearly that the mesh detail was perfect in fact. Each original pixel is shown at 50 screen pixels on my monitor, thus screen resolution and sub-pixels become irrelevant. Equally, any lack of detail quality would have become blindingly obvious.

In a recent post, a very nice landscape image had been shot accidentally using Sigma's medium resolution mode. Pixel-peeping at 8X (800%) showed unusual detail artifacts but which the poster said he could not see on his monitor (at 100%, IIRC). Now we know why ...

-- hide signature --

FastStone Viewer - don't leave home without it ...


 xpatUSA's gear list:xpatUSA's gear list
Sigma DP2 Sigma DP2 Merrill Panasonic Lumix DMC-LX7 Panasonic Lumix DMC-G1 Sigma SD15 +17 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow