Full frame sensor vs small sensor

Everyone keeps saying that the price of full frame sensor?s will go

down. I say who cares! I like my 20D sensor, I don?t want a large

sensor. I like the extra mm?s I get with my zoom! Also, they will

surely come out with a small sensor that has 10 MP maybe even

15MP... And I bet that will happen fasters then the larger sensors

coming down in price! I guess only time will tell?

I too don't care greatly about the 24 x 36mm sensor. Afterall, I can still shot chromes if I need my wide angles (and I often do). Plus I still enjoy film for what it is. However what I miss, and miss bightime, is the large, bright and vivid viewfinder I took for granted on my EOS 3. The viewfinder of the 10D/20D is like a peephole in comparison.

In keeping With Bobs' nudie theme in general archive; We're boys....we're supposed to like peepholes! Kidding aside, I'd sure like to actually see what it is I'm composing in the viewfinder.

I care. Now go play w/your 1.6.

I think I'd be very happy with the current line if they used Foveons with the noise characteristics of the 20D.

 

DI

I think Nader has an interesting point, "...extra mm's I get with my zoom."

 

A very funny thing happened tonight. I had my camera packed back in my Tamrac bags and placed them in my car. The next thing I knew, about four police cars pulled over three cars and all of the officers had the guns drawn.

 

I didn't want to open two Tamrac bags. 1st solely hold my EF 70-200 2.8L and the 2nd is the huge Tamrac bag that holds evening else.

 

Here is where those extra "mm's" come in handy. 20D to 1600EI, Partial Metering with an EF 135mm 2.8 SF. It was very nice that my 135mm became a 216mm

 

"The moral of the story is: If you ever throw a party, go over your guest list. Some of your invitees might be from rival gangs."

In my humble opinion, the size of the sensor is pretty crucial. You can digitally zoom to your heart's content, and if canon ever bring out a 25MP+ full frame I'll probably get it despite the undoubtably astronomical price tag. Why? because I have enough pixel density to still zoom to more than 1.6 in photoshop and have my "extra mms" just as if the camera was a 1.6x crop APS-C sized sensor, but I ALSO get the full 17-40mm on my ultra wide zoom. If you never record the information, you can't add it later...

 

All the best

Damien, digital zoom is quite worthless. The value in a larger sensor is the field of view.

The smaller sensors do give you more value with your long lenses, though.

I think the sensor size it is not the main issue on digital cameras. I belive that the <i>dynamic range</i> of sensors is the real one. For the moment I think that overall digital sucks.

Michele, my 10D has 10 stops of useful dynamic range in color, and about 12 for B&W

work (from RAW files). This beats color and slide film.

Sheesh my 10D only has about 5 stops of dynamic range. I can squeeze out another stop

of shadow detail by processing RAW files with differing amounts of digital exposure comp

and blending in PS layers. Two stops and the noise bothers me and the shadows look too

flat. However, that still falls really short of what I can see on a Portra 160 NC neg. Of

course no paper can reproduce 10 stops of dynamic range anywho.

Bigger sensor = Bigger viewfinder. For me, this is the only appeal in the bigger sensor. I consider the old 1D instead of a new 20D mainly for that reason.

 

Happy shooting,

Yakim.

Yeah. . I only get five stops. . .assuming my exposure was dead nuts on to begin with. I am not sure the answer isn't less. . .. this is my biggest disappointment with digital. No one even TALKS about improving the woefully inadequate dynamic range.

I think even Canon only claim 8 stops of range on a 10D. I certainly don't find my 10D has

as good dynamic range as my old "analog" SLR, raw or otherwise.

 

Andrew, if you do the sums you'll find that a 1.6x digital zoom on something like a 1DS

mkII with a 16mp sensor gives a comparable resolution image than a "normal" image on a

10D with a 6mp sensor. So, digital zoom does have its place, providing you have enough

resolution to play with. Given the hypothetical 25mp full-frame sensor above, if you did

1.6x crops you'd still be left with 9.76mp images.

 

To look at it another way, a 1.6x linear enlargement enlarges the area by 2.56x. That

means that a 1.6x zoom/crop on a 16mp image leaves you with 6.25mp. Or, in other

words, the pixel density on the 1DS mkII sensor is about the same as the 10D sensor, you

just have more pixels to play with because the sensor is bigger; if you start cropping the

sensor down by 1.6x you get a 10D sensor, more or less.

Thanks David - I'm glad I read your post before writing essentially the same thing myself. My reasoning was that the 20D has 8.2MP and is therefore equivalent to a 21MP full frame sensor. Anything over 21 and I can crop and magnify back to exactly where my current 20D is in terms of pixel resolution.

 

BTW I agree about the dynamic range being far more of an issue, but that wasn't the post topic!

<i>my 10D has 10 stops of useful dynamic range in color, and about 12 for B&W work (from RAW files)</i>

 

<p>ROFL :-). ...and 20D has 20 stops of useful dynamic range, hence the name...

Hi Nader

 

I think the point about full frame sensors is that you get a better signal-to-noise ratio, as each site is larger therefore more light per site.

 

Nick

Would you folks like to experiment?

 

Use ACR, and shoot a scene so the histogram is bumped up all the way to the right. Then,

reduce exposure a stop. Keep repeating this until you get an image that looks black.

 

Then process those images in the ACR, and add exposure. I find that the color suffers

beyond 8 stops or so, but I can squeeze 12 stops with RAW processing.

 

Try it, and then get back to me. Bear in mind, this is worthless for JPGs.

<I>Would you folks like to experiment?<P>

 

Use ACR, and shoot a scene so the histogram is bumped up all the way to the right. Then,

reduce exposure a stop. Keep repeating this until you get an image that looks black.<P>

 

Then process those images in the ACR, and add exposure. I find that the color suffers

beyond 8 stops or so, but I can squeeze 12 stops with RAW processing.</i><P>

 

Aha! Now Andrew wants to use data instead of theory! (snide and unworthy reference to a

previous thread in which the two of us crossed swords over theoretically-derived MTF

curves). It will be interesting to see the

results of the proposed experiment. Supposedly the functional color depth of a RAW file is

about 12 bits, which is extrapolated to 16 bits in a typical TIFF or Photoshop conversion.

Since a bit

is essentially a factor of two, you have 2<SUP>12</SUP> possible steps (about 4,000) in

each of the RGB (or is it CMY?) channels from a RAW file (but only 256 per channel in an 8-

bit RGB file). An f-stop

scale is also calibrated in factors of 2 (one stop = a doubling or halving of light level), so

in theory there is some support for the suggestion that a DSLR image could have a

dynamic range of 12 stops. <P>

 

BIG caveat: I'm not sure how well theory translates to practice here, or how one can

interpret an approximate 12-bit color depth into a B&W dynamic range. My own

impression is that my 10D and 1D images have more dynamic range than slide film

(especially in shadow detail), but I'm dubious about a dynamic range of 8-12 stops. I'd

guess about 5.

I feel the need to point out two things:

<BR><BR>

1) There are no "extra mm's". The focal length of the lens <I>does not change</I> when you put it on a different body. What changes is your <I>field of view</I>. That may be semantics, but thre IS a difference; I view an increased focal length as an advantage (potentially) while a decreased fov is a definite disadvantage. I'd love to post an image or two to demonstrate but I can't afford a full-frame camera, and even if I could I don't know if there are two sensors (full and APS-C) out there that are the same density/composition.<BR><BR>

2) Digital zoom is not the same as cropping. Again, semantics, but IMHO digital zoom is more akin to resampling than cropping. If you crop, you are changing the <I>amount</I> of original data, but not the <I>quality</I> of it. If you zoom digitally, you are affecting the quality because you are creating data that does not exist. Hence I agree with the statement digital zoom is worthless, but also the rationale for why "it" isn't if not the term being used.

What I say can easily be reproduced by experimentation. I actually tested this a while

back, but have thrown away all the images. I'm not so excited to prove myself - believe

me or not. Trying this for yourself will convince you more effectively than all the typing I

can do.

 

A few notes. I find that grossly underexposed images (even ones that appear to be solid

black) can be brought to life with the exposure compensation feature in ACR. Color data

suffers, however. Little can be done about blown highlights, though. I usually convert to

ProPhoto RGB, and reduce contrast slightly when making a RAW conversion, which can give

back some lost highlights compared to Adobe RGB or SRGB.

<I> I view an increased focal length as an advantage (potentially) while a decreased fov is

a definite disadvantage. </i><P>

 

A decreased FOV is usually advantageous if you are using telephoto lenses. You don't

need as long a lens to get the desired FOV, or conversely, you have a bigger and more

satisfying fraction of the frame filled with the subject.

<I>No one even TALKS about improving the woefully inadequate dynamic range.</i><P>

 

It all depends on your point of view. If you shot slides back in your film days, a DSLR is a

wondrous improvement that

offers <I><B>better</b></i> dynamic range than slide films. Of

course,

it's a different story for color negs and BW films (or so I hear -- I've had little experience

with those).

<I> If you zoom digitally, you are affecting the quality because you are creating data that

does not exist. Hence I agree with the statement digital zoom is worthless</i><P>

 

So, it's worthless to interpolate a digital image upwards (increase the pixel count) to match

the optimal dpi on an inkjet printer to the final print size you want? I wouldn't go that far,

but that's essentially what you suggest by saying that digital zooming is worthless.

Worthless is probably too strong a word, particularly in the context of relating the term to resampling. It's just that every time I hear the words "digital zoom" it means an in-camera function to me, and I've never seen a good one. I can't really comment on the whole optimum ink-jet bit, as I've only recently heard it mentioned. Haven't done any reading or testing on my own, though it sounds reasonable and is something I'll have to give a shot.