TAKO. INTERNET SEIT 1996.
Olympus-OM

[OM] Re: E-1 ISO 800 Acceptability

Subject: [OM] Re: E-1 ISO 800 Acceptability
From: Moose <olymoose@xxxxxxxxxxxxxxx>
Date: Fri, 10 Mar 2006 17:17:25 -0800
Joel Wilcox wrote:

>On 3/10/06, Moose <olymoose@xxxxxxxxxxxxxxx> wrote:
>  
>
>>Joel's full pixel samples of his daughter are more like what I saw. Not
>>awful, and not a problem at all reduced for the web, but significantly
>>limiting the cropping and large print options.
>>    
>>
>
>I ran the printer this morning and made prints of the samples I posted
>yesterday, with variations.  Yesterday's samples are crops of the
>original ISO 200 and 800 RAW files.  I'm not sure what percentage was
>actually cropped, but I would estimate that the printed portion of the
>5x7s are the equivalents of letter-size prints had I not cropped.
>  
>
More like 6x8.

>I printed....................
>I discovered two things by doing this:
>
>1) Not only are the ISO 800 captures "not awful," as Moose says, they
>are potentially very good -- provided you do due diligence by getting
>exposures right and possiblity shooting RAW, etc.  
>
Yup.

>A little post-process NR is OK.
>  
>
Weeellll, that depends. See examples below.

>2)............
>
>We're in something of a similar circumstance with digital cameras,
>it's just that we use our monitors as magnifying glasses in addition
>to normal viewing.  My beef is when too much magnification starts to
>be considered "normal."
>  
>
Well, I don't know what normal is in this context. When viewing the 
whole image, I use whatever % fills the viewing area. For 4000 dpi scans 
of 35mm, that is 12.5%, 25% for the 300D, 50% for the S110. On the other 
hand, I know from experience that I need to view at higher levels to see 
what a print will show.

>I don't mind viewing at 100% pixels to sharpen, but at 100% nothing
>looks very good to me, no matter what the source.  
>
I don't know what's up with your equipment for viewing, but I have many 
images from my three digicams that look good to me at 100%. Here are 
three from the F10, a tiny sensor P&S. In all cases the full frame shot 
at about the equivalent of 25% precedes the 100% crop 
<http://galleries.moosemystic.net/Iron4/pages/IRON4021.htm>, and, 
<http://galleries.moosemystic.net/Iron4/pages/IRON4011.htm> and 
<http://galleries.moosemystic.net/Iron4/pages/IRON4025.htm> .

And I don't think this from the 300D is half bad at 100%. Notice iso 
800, too 
<http://www.moosemystic.net/Gallery/MPhotos/Iggy/pages/CRW_2224c3.htm>. 
Any camera that produces images that can't be viewed and look decent at 
100% is giving away a lot to those that can.

>Even after sharpening I like to shift everything down to 50% to look at 
>details. 
>That's where my monitors seem to do best.
>  
>
Again, I don't understand. There is a great deal of detail I can see at 
100% that I can't see at 50%. It's sort of like buying a 5 mp camera, 
then using it at 2.5 mp. To make comparisons easier, I've stacked 100% 
crops of your samples on top of each other 
<http://www.moosemystic.net/Gallery/Others/E1noise.htm>.

The iso 200 is really nice. 800 is pretty noisy, but retains just about 
all the detail in the eyebrows. The NR800 simply loses lots of detail 
compared to 200. Now none of that matters a bit if you like a bit of 
noise and/or if you only post on the web and print full images or modest 
crops. With the image at hand, the iso 200 image would probably look 
nice cropped to a head only shot at 8x10. The 800 would not. If that 
limitation doesn't bother you, everything is fine. What some others like 
me are saying is that there are other DSLRs in the same price range, or 
even cheaper, that will give comparable results at 800 to what you get 
at 200.

>Why is 100% correct for viewing?  Why not 200% or 400%?
>
It isn't always, perhaps even seldom. But why shouldn't a camera produce 
an image that is good at it's native resolution? Clearly, higher mags 
aren't desirable for most uses because they introduce artifacts from the 
interpolation process. But lower ones don't show all the detail in the 
image.

>Does anyone actually print at 100% pixels?
>
Yes, I'm pretty sure you do. The 2560x1920 image from an E-1 is 7.8x10.7 
in. at 240 dpi. At 10x14, it is 180 dpi. In any case, your printer 
driver is upsampling from the source image to print at it's actual drops 
on paper dpi.

If you mean do they print at the equivalent dpi on paper of the pixel 
pitch on a computer screen, the answer is still yes. A full E-1 frame at 
the 96 dpi of my screen is 19.5x26.7 in. So a slightly cropped E-1 frame 
printed at the 12x16 size many of our printers can do starts to get 
there pretty fast. Folks like Tom S. are already well past that and Bill 
B. is on his way to a bigger format printer.

>Is this a target for sensor designers?
>  
>
I think so. Otherwise all they'd have to do is put in a half rez sensor 
and upsample it to get the mps for marketing, knowing nobody would ever 
use more than 50% of the rated resolution.

>Just as we went through this about 10 years ago with inkjet printers,
>  
>
And yet, whatever the advertised dpis of current inkjets, they make MUCH 
better images that 10 years ago. They must be doing something right.

>I think we're experiencing a little bit of silliness about noise.
>  
>
Depends on what one is doing with the image. What's good for one may be 
insufficient for the uses to which another may put his/her images.

>I'm not putting this out there to discredit anybody or their decisions.
>
Nor am I trying to discredit yours. Just trying to point out what is 
true for some others, where pixel peeping may not be the equivalent of 
navel gazing.

In fact, I'm really pleased that you and many others like your E-1s and 
have induced others to buy more. That increases the likelihood that Oly 
will build the camera I want one of these days.

Other than a couple of what one hopes are fixable firmware shortcomings, 
the E-330 looks real close.

Moose


==============================================
List usage info:     http://www.zuikoholic.com
List nannies:        olympusadmin@xxxxxxxxxx
==============================================

<Prev in Thread] Current Thread [Next in Thread>