Will camera performance increase by selecting smaller resolution?

I have a 5MP camera, I don't crop heavily nor print poster sized

prints. I was wondering if by setting the camera to shoot at 3MP will

it shorten the time between shots.

Also, will my images be less noisy since the sensor is bigger

relative to pixel count?

"I was wondering if by setting the camera to shoot at 3MP will it shorten the time between shots."

 

Probably.

 

"Also, will my images be less noisy since the sensor is bigger relative to pixel count?"

 

No.

"Also, will my images be less noisy since the sensor is bigger relative to pixel count?"

 

No.

 

Is it because a smaller image just uses a smaller section of the sensor?

>Is it because a smaller image just uses a smaller section of the sensor?

 

Hi Robert, No, noise is a function of the signal-to-noise ratio of the chip inside your camera, so the only thing that reduces noise is to shoot at the lowest possible ISO. A smaller image size won't buy you a less noisy picture, unfortunately. The only thing I can think of that might is to deep-freeze your camera; digital camera sensors produce less noise when they're very cold. Best wishes . . .

"Is it because a smaller image just uses a smaller section of the sensor?"

 

No, the camera will be using the whole chip area regardless of the resolution you choose. Noise will be a function of the size of the chip and the ISO setting- the smaller the chip, the more noise; the higher the ISO, the more noise.

<em>Noise will be a function of the size of the chip and the ISO setting</em>

<p>

For the data directly from the senor, yes, but 3MP isn't going to be data directly from the sensor. If the camera captures a 5MP image and then downsizes it to 3MP, it will be averaging pixels. When you average pixels you may tend to lower random noise.

<p>

However the obvious answer to this question is "You have the camera, you tell us". Take shots at ISO 400 at 5MP and 3MP and compare the noise levels. Do they look different?

"If the camera captures a 5MP image and then downsizes it to 3MP, it will be averaging pixels. When you average pixels you may tend to lower random noise."

 

Hmm, then reducing the image size may reduce some noise but not for the reason I thought (bigger sensor relative to image size).

 

Now the other issue is that perhaps the shot to shot time does not improve if the camera captures at 5MP regardless and then resize to 3MP.

"The only thing I can think of that might is to deep-freeze your camera; digital camera sensors produce less noise when they're very cold."

 

I'm taking this comment somewhat seriously. Will try to keep the camera cool at all times. Looking forward for winter shooting!

"If the camera captures a 5MP image and then downsizes it to 3MP, it will be averaging pixels. When you average pixels you may tend to lower random noise."

 

You can try this and A-B two images. However, I wouldn't expect a demonstrable difference in noise, all other things being equal, between a 3MP setting and a 5MP setting.

 

"Now the other issue is that perhaps the shot to shot time does not improve if the camera captures at 5MP regardless and then resize to 3MP."

 

Robert, I have an Oly C-5060. If I shoot on TIFF, the camera locks up longer digesting an image than if I shoot on RAW. And if I shoot in RAW, the camera locks up longer digesting images than if I shoot in JPEG. I don't notice dramatically-shorter write times shooting in high-rez JPEG v. shooting in medium-rez JPEG.

And, BTW, if noise is an issue for you, you might read this thread:

 

http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00950t

I always look for a compromise. I find that SHQ is good enough for what I do. Gives good quality, workable file size,and prints and e mails just fine. I forget,but as I recall the image size averages about 4 MB on my 5 MB C 5050. And I shoot at ISO 64 as often as I can get away with it. Noise has never been an issue. And I crop a durned lot. Also,nature of subject matter seems to get into the noise issue,doesn't it. Things get to be complicated.<p>

Atkins has a point about experimentation,with your model camera.Good luck.

>I was wondering if by setting the camera to shoot at 3MP will

>it shorten the time between shots.

 

Possibly, although not guaranteed. It could also increase it due to the resizing algorithm.

 

>Also, will my images be less noisy since the sensor is bigger

>relative to pixel count?

 

Possibly, although I wouldn't necessarily count on the difference being obvious (especially as the odd resize could introduce noise of its own, and because the noise that would disappear might be of the non-objectionable kind.

I had a Panasonic Lumix camera that had significantly lower noise levels at 2 Megapixels than at 4. (And others who posted with the same camera had the same observation.) I'm not sure why, though.

That's because light is a random process (photons are emitted randomly and uncorrelated), and because most sensors saturate around the same number of photons per unit of surface (it's of the order of 1000 photons per square micron if I remember correctly). But shooting at a lower resolution you increse the sensor surface that is available for each destination pixel, i.e. you increase the number of photons that will be measured for each pixel, i.e. you decrease the noise (of course, assuming that there are no other sources of noise in the process of using fewer pixels).

 

That is the primary reason why DSLR sensors are less noisy than small-sensor point-and-shoots.

When Iuse my 6MP DSLR with very small resolution I can shoot more frames continous.

For noise reduction I'd choose high resolution to have enough pixel I can dump by downsizing after using "neat image".

I don't know why you?r camera performance isn't perfect. Whith digital I see no problem clicking away some 20 frames while counting seconds.

Thank you all for your input. It's been enlightening. I certainly could've experimented and answered my own question for this specific camera (Minolta A1) but knowledge obtained this way would've been pretty static: (yes, it's faster or less noisier if you do it this way or that way...etc.)

I really posted this question in order to understand digital photography better. Specifically the dynamics regarding sensor size/quality, pixel count, lens size/quality and ISO settings. What would be the order of importance? I suspect that pixel count is the least important, yet manufactures keep pushing pixel count on casual photographers like me.

Please keep your comments, ideas and knowledge coming.

 

Best Regards,

 

Rob

<i>I give it a try...</i><br>

If you consider a sensor A, and a sensor B with 4x more MP than A, <i>at the same technology level and same sensor size</i>:<p>

 

<b>A. Photon Noise</b><br>

- at the Photosite level: Signal(A) = 4x Signal(B)<br>

- Photon-Noise(A)=sqrt(Signal(A))<br>

- Photon-Noise(B)=sqrt(Signal(B)) = sqrt(Signal(A)/4)=Photon-Noise(A)/2<br>

=> Signal(B)/Photon-Noise(B) = 1/2 x Signal(A)/Photon-Noise(A)<p>

 

BUT if you resize the image from B to the size of A, you will sum signal from 4 pixels into 1, and smooth the noise accordingly => New-Photon-Noise(B)=sqrt(NoiseB1^2+NB2^2+NB3^2+NB4^2) = sqrt(4x Photon-Noise(B)^2) = sqrt (4x Photon-Noise(A)^2 /4) = Photon-Noise(A)<p>

 

<b>=> If you increase MP on a given sensor size, Photon-Noise will not be detrimental to Signal/Noise in the output image</b> (you just need to resize to revert to the same S/N... and probably not even needed once you don't see each pixel).<p>

 

<br>

<B>B. Constant Noise (Read Noise and others...)</b><br>

If you consider a constant noise, which doesn't depend on the size of Photosite, you will get: Signal(B)/Noise(B)=(Signal(A)/4)/Noise(A)= 1/4 x Signal(A)/Noise(A). If you resize B image to A size, New-Noise(B)=sqrt(4xNoise(B)^2)=2xNoise(A).<p>

 

<b>=> If you increase MP, constant Noise will be detrimental to Signal/noise in the output image, by a factor = sqrt(MP increase), here = sqrt(4)=2.</b><p>

 

<i>Some other types of noise will be neither constant, neither up to square(signal)</i>.<p>

 

<br>

I can't draw any conclusion because I don't know which type of noise is the real limitating factor in our pictures (it seems that 'Read-Noise' would be usually the bad one...). Any idea about which kind of noise does really show up in our pictures ?<p>

 

Without more data, I would tend to favor:<br>

<b>1. Larger Lens</b> (as an approximation of useful amount of light captured -ie:"Signal" - for a given useful Range)<br>

<b>2. Larger Sensor</b><br>

<b>3. Minimum MP</b> (above requirements, though...) - and not only for noise, but also for improved Dynamic Range, etc...<br>

Important: I would also favor a recent technology (sensor, electronics...) and ABOVE ALL: real tests of the camera by a good magazine/website.<p>

 

Olivier<br>

PS: as technology improves, we will get more and more MP without real adverse effects on quality (ie: Read/Dark noise will get lower, Saturation will get higher, etc...). So there are no reason to bash small photosites per se... however, there is today an unbalanced priority given to MP, which is probably detrimental to the overall quality of pictures for most needs.

Excellent post by Olivier.

 

A few pointers:

 

-Those numbers make the assumption that Signal(A) = 4x Signal(B), which isn't quite true - the fill factor is smaller for smaller pixels, i.e. resizing down a high-MP image will actually result in more noise than a "native" low-MP image.

 

-The tests at DPReview for e.g. the 1DII seem to show that at low ISOs the primary source of noise is constant, i.e. read noise between the amplifier and the ADC (which is a constant). At high ISOs the read noise between the sensor and the amplifier becomes significant (doubles for each stop), but in the middle of the range (ISO 200 - ISO 800) is looks like photon noise is the biggest contributor (sqrt(2) per stop). I'm too lazy to do it, but the curve for the 1DII seems smooth enough and has enough data points to actually try to compute the constrbution of those 3 sources of noise.

Thanks for the feedback, Jean-Baptiste. I agree.<br>

I found more data and continued the discussion in <a href="http://forums.dpreview.com/forums/read.asp?forum=1000&message=9985120">a thread on preview</a>, and before that, in a long and intensive <a href="http://www.photim.net/nci/discu.php3?code=20040806212918daleroux#O147">french discussion</a>.<p>

 

The picture shows the noise in a CCD sensor (source: a very detailled and interesting <a href="http://www.kodak.com/global/plugins/acrobat/en/digital/ccd/applicationNotes/noiseSources.pdf">doc</a> by Kodak). The conclusion is the same as yours: Read-Noise in low level (the most disturbing one) and Photon-Noise after...<p>

 

Other findings:<br>

- Read-Noise increases slightly less quickly than sqrt(photosite's area) => You have a slight advantage on Noise by choosing less MP over the same sensor size - for a given technology and generation.<br>

- Saturation increases noticeably more quickly than photosite's area => You have a noticeable advantage on Dynamic Range by choosing less MP over the same sensor size - for a given technology and generation.<p>

 

This is why I think that my previous point [3] still makes sense (but on Dynamic Range rather than on Noise).<br>

Of course, it depends on how much you value Dynamic-Range and Noise over Resolution (the current unbalanced emphasize on MP is not to my liking - and probably against what most people really need...).<p>

 

Olivier