5
\$\begingroup\$

If I put a DX lens on a Nikon D700, it will reduce megapixels to 5.1.

I understand that FX pixels are about 2.5 times larger than DX pixels. Would I be correct to assume that, therefore, my FX image using 5.1 megapixels will produce quality equivalent to at least a 10 megapixel image on a DX camera, all other things being equal?

Improve this question
\$\endgroup\$
5
  • \$\begingroup\$ Some of this is covered under earlier more general questions like Is crop-factor a bad thing? but I don't think it's bad to cover this specifically. \$\endgroup\$
    – mattdm
    Commented Sep 23, 2012 at 13:44
  • 1
    \$\begingroup\$ How do you measure "quality?" \$\endgroup\$
    – Blrfl
    Commented Sep 23, 2012 at 14:14
  • \$\begingroup\$ @Blrfl - there are various standardised mesure of quality. The standards to not always agree :-). see eg DxOMark's sensor evaluation site . \$\endgroup\$
    – Russell McMahon
    Commented Sep 24, 2012 at 1:14
  • \$\begingroup\$ @RussellMcMahon: Standards are wonderful because there are so many to choose from. I find it's better to make the best of whatever you're shooting with and not worry so much about what's on the spec sheets. \$\endgroup\$
    – Blrfl
    Commented Sep 24, 2012 at 2:38
  • \$\begingroup\$ @Blrfl - I was responding to a specific question and providing an example answer that addressed the question. I see now that you have asked a question and also criticised (critiqued if you must) a 'perfectly good' [tm] answer. that meet the formal definition of trolling. Go away :-). Looking at your answer I see it can be translated to "quality is what I like". That is indeed one valid answer, and one I employ often enough [tm again]. However, other people are liable to want a little more. Signal to noise ratio, and dynamic range of various parameters tends to be a little more universal. \$\endgroup\$
    – Russell McMahon
    Commented Sep 24, 2012 at 7:12

2 Answers 2

Reset to default
6
\$\begingroup\$

Assuming the same sensor generation, if you are comparing 5MP 8.45um pixels to 10MP 6um pixels, then the "quality" of the 5MPs will be better (better light sensitivity, less noise visible in high ISOs), so in a situation where that is important to getting the good image (regarding exposure, motion blur, etc. ) then that would be preferred. It you are making a huge print where the lower image resolution would be visible, and where you could get the right image with the lesser quality 10MP, that would be preferred. So the simple answer is: no, it is not the same, nor equal. But it is also not black and white, which one is best.

Improve this answer
\$\endgroup\$
7
  • \$\begingroup\$ I know the question says "all other things being equal", but just for clarity, how important is photosite size compared to technology improvements? What if we're comparing 8.45µm pixels from a sensor made in 2005 to 6µm pixels from a sensor made in 2012? \$\endgroup\$
    – mattdm
    Commented Sep 23, 2012 at 14:21
  • \$\begingroup\$ Yes, that is a very interesting comparison, which complicates an answer enough to make an enire PH.D thesis on that question alone. I read about how CMOS can add all kinds of technological advanced into the processing of each pixel (ie for higher dynamic range) but tehy do say it is at the cost of photosite size and adding transistors per pixel - they like to keep it at around 3T (transistors per site). In imaging the amount of light they can capture is always highest priority compared to the other things, and the area of each site is a huge factor for this purpose. \$\endgroup\$
    – Michael Nielsen
    Commented Sep 23, 2012 at 14:38
  • \$\begingroup\$ The other factors are lens specs and whether it is a back illuminated, front illuminated, or foveon chip(where the trouble is getting the light through the layers). So the question is if technological improvements help or add the need for larger sites to overcome the extra electronic noise? In machine vision there's a new CMOS sensor called CMOSIS with double quantum efficiency. I haven't read about how they achieve that, yet. \$\endgroup\$
    – Michael Nielsen
    Commented Sep 23, 2012 at 14:44
  • \$\begingroup\$ I believe your opening sentence is incorrect, unless you're comparing the same sensor generation. The pixel size on the original Canon 5D is larger than the pixel size of the Canon 1D4. The latter has better light sensitivity and lower noise at higher ISOs. \$\endgroup\$
    – Eric
    Commented Sep 23, 2012 at 14:50
  • \$\begingroup\$ yes, assuming you are comparing the same generation. \$\endgroup\$
    – Michael Nielsen
    Commented Sep 23, 2012 at 14:57
4
\$\begingroup\$

No, but let's start with why you might think the contrary:

They are the same pixels. Therefore each pixel has the same noise-characteristics, the same dynamic range and sits behind the same anti-alias filter. Assuming your sensor does not out-resolve the lens, meaning your DX lens is good enough.

What you get is a 5 MP crop from a 12 megapixels sensor. That's a very nice 5 MP image but its still 5 MP, you can print a smallish print but it will degrade quickly as you get past an 8" x 10".

With 10 MP DSLR (APS-C or not), you have enough resolution to make a sharp 10" x 15" print. This is of course until noise-levels start eating away image details and lowering dynamic-range.

That is where the full-frame sensor's bigger pixel show there advantage, as you compare at higher ISOs, say 1600+ on modern DSLRs, then, your full-frame camera will pull-off the print much better. At some very high-ISO, I guess that you will get a better image form 5 MP full-frame DX-crop versus 10 MP APS-C but it's only going to be in such limited circumstances.

Improve this answer
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.