snubbr.com

sensor size vs mp
I'd appreciate it if someone could help me with a few questions..

I am aware that image quality is closely correlated with the size of sensor. I am looking for a camera around $400, and found out that the sensor size around this price bracket are almost all 1/2.5" with a few 1/1.8". But I am not sure how # of mp affects image quality with the same size sensor. My questions are:.

- With sensor size both being 1/1.8", does a 12mp or a 8mp camera deliver better image quality?.

- If I purchase a 12mp camera and set it to 8mp, does it deliver better/same/worse image quality comparing to the 8mp one.?.

Thank you.

Ren T...

Comments (41)

Mmotp wrote:.

- With sensor size both being 1/1.8", does a 12mp or a 8mp cameradeliver better image quality?.

Everything else (sensor size, sensor technology, noise reduction algorithms and settings, lens) being equal, the 8mp camera would be better. That jump to 12mp only proides 20% or so (SQRT(12/8)) more linear resolution, but 8mp is probably already too many pixels for that sensor size, in terms of noise..

Now if you put a 35mm-full-frame-size sensor in the camera, 12mp might work!..

Comment #1

Mmotp wrote:.

I'd appreciate it if someone could help me with a few questions..

I am aware that image quality is closely correlated with the size ofsensor. I am looking for a camera around $400, and found out that thesensor size around this price bracket are almost all 1/2.5" with afew 1/1.8". But I am not sure how # of mp affects image quality withthe same size sensor. My questions are:.

- With sensor size both being 1/1.8", does a 12mp or a 8mp cameradeliver better image quality?.

- If I purchase a 12mp camera and set it to 8mp, does it deliverbetter/same/worse image quality comparing to the 8mp one.?.

Thank you.

Ren T..

1) 8mp obviously, since the size of the pixels is physically larger..

2) Worse. Buying a 12mp camera and setting it to 8mp only makes the camera resize the 12mp picture it takes down to 8mp before saving it. You not only get the disadvantage for smaller pixels but you also introduce extra error during the resize process. For maximum quality you should always shoot at the native resolution of your camera.....

Comment #2

I am having 8MP camera but most of the time I will make it as 5MP due to Memory constrain....

Now only I understood the fact if we take pic for 5MP using 8MP camera will results the lesser quality pic taken by 5MP cameraThanks for the information!!!!.

I have one more question.

Is there any relation bettwen sensor size and ISO??? Why I am asking this question is All the DSLR having bigger size of sensor and people are saying because of that sensor size DSLR having better quality picture in poor light..

I have compared the some DSLR camera with P&S(like S5IS ,fuiji9600 etc)camera All the features of P&S matches with DSLR only it's differing from sensor size..

-Thanks..

Comment #3

Prabakaran wrote:Is there any relation bettwen sensor size and ISO???.

There is a big relation between sensor size and usable ISO. Think of a photosite (a site on the surface of the sensor that collects data for one pixel ... or for part of one pixel) as a bucket set out to collect water..

(Sensor Area - Overhead)Photosite Size = Number of Photosites (Megapixels).

The bigger a bucket is, the more rain it can collect and the more accurate the bucket's "estimate" of rainfall will be. Same for photosites. The Sensor Area part of the above equation means that DSLRs can get away with more megapixels (for the same noise performance) or obtain better noise performance (for the same MPs) than small-sensor compacts..

Why I am askingthis question is All the DSLR having bigger size of sensor andpeople are saying because of that sensor size DSLR having betterquality picture in poor light..

Yes. That is one reason. The availability of prime lenses that are faster than any lens usually seen on compact cameras (e.g., 50mm f/1.8), and the option to use external tilt/bounce flashes are two others...

Comment #4

I've heard it described that ISO 400 on a APS-C sized DSLR is just about equal to ISO 100 on a compact as far as noise is concerned. I don't know if that is true...

Comment #5

Mmotp wrote:.

I have my questions answered. Thank you  .

This was your first post on dpr. Why do you think the two guys that answered your questions were correct? They both may be idiots... .

Well, relax...they gave you the right answers, but my point is that if you believe the first two guys who step up, you will at some point get BAD advice..

How about a 3rd opinion? Here are the "rules"....

1.a. Buy a camera with as large a sensor as you can afford.1.b. Buy a camera with as few pixels as you can stand..

To make "rule" 1.b. clear, many people are sensitive or embarrassed to tell people that their camera has ONLY X MP. These people want to impress their friends (and strangers) MORE than they want to take high IQ pix..

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #6

BA baracus wrote:.

I've heard it described that ISO 400 on a APS-C sized DSLR is justabout equal to ISO 100 on a compact as far as noise is concerned. Idon't know if that is true..

The APS-C sensor has about 9 times more area than a 1/2.5" sensor. The photon noise comparison would therefore be about 3:1 (square_root of photon count) for an ISO comparison of about 300:100..

I got my numbers from dpreview's glossary..

Comment #7

Chuxter wrote:.

How about a 3rd opinion? Here are the "rules"....

1.a. Buy a camera with as large a sensor as you can afford..

Yep..

1.b. Buy a camera with as few pixels as you can stand..

Nope..

To make "rule" 1.b. clear, many people are sensitive or embarrassedto tell people that their camera has ONLY X MP. These people want toimpress their friends (and strangers) MORE than they want to takehigh IQ pix..

With higher MP you have the choice of more detail or low noise. With less MP you get low noise but don't get the choice of more detail..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #8

Mmotp wrote:.

- With sensor size both being 1/1.8", does a 12mp or a 8mp cameradeliver better image quality?.

All other things being equal, you get more detail and more noise..

- If I purchase a 12mp camera and set it to 8mp, does it deliverbetter/same/worse image quality comparing to the 8mp one.?.

All other things being equal, you get the same detail and noise..

Thank you.

You're welcome..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #9

I dont want to beat this to death but... ill use generic sizes to make it simpler for me. if I take a 1cm sensor that is 10mp, thats 3333x3333... simple. each pixel is 300 um is size. if now I take a 2cm sensor that is 10mp, I still have 3333x3333 pixels but now they are 600um in size.

The larger sensor has a "bigger bucket" to collect light. so you get much more light compared to the background noise level (signal to noise ratio). of course there are other factor that can change things a bit, like pixel spacing, but for the most part you can rely on this for a general understanding of image quality vs sensor size/resolution..

Comment #10

Nickleback wrote:.

Chuxter wrote:.

How about a 3rd opinion? Here are the "rules"....

1.a. Buy a camera with as large a sensor as you can afford..

Yep..

1.b. Buy a camera with as few pixels as you can stand..

Nope..

To make "rule" 1.b. clear, many people are sensitive or embarrassedto tell people that their camera has ONLY X MP. These people want toimpress their friends (and strangers) MORE than they want to takehigh IQ pix..

With higher MP you have the choice of more detail or low noise. Withless MP you get low noise but don't get the choice of more detail..

It's clear that you can't stand low MP. That's fine. Each of us has their own threshold. The "rule" is still a good one..

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #11

Chuxter wrote:.

It's clear that you can't stand low MP..

I'm fine with low MP. Given your cameras it seems that you aren't..

The "rule" is still a good one..

Nope. We've discussed this before. Go compare noise and detail of D40 and D300 in one of the still-life shots at the same ISO. For noise look at the color checker, for detail look at the knurls on the salt & pepper shakers, or the strings on the fiddle on the bottle, the pattern on the wine vinegar label, or any other fine detail you'd like to pick out..

Http://www.imaging-resource.com/IMCOMP/COMPS01.HTM.

Or just click on these links for an ISO 1600 comparison:.

Http://www.imaging-resource.com/PRODS/D300/D300hSLI1600.HTMhttp://www.imaging-resource.com/PRODS/ND40/ND40hSLI1600.HTM.

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #12

PhotonFiend wrote:.

If I take a 1cm sensor that is 10mp, thats3333x3333... simple. each pixel is 300 um is size. if now I take a2cm sensor that is 10mp, I still have 3333x3333 pixels but now theyare 600um in size. that is 4 times the area..

It's the sensor size that counts, not the size of the individual pixels. The 2cm sensor has less noise compared to the 1cm sensor because it is larger..

If you changed your example to 40mp on the 2cm sensor, you'd have double the (linear) resolution and the same noise per pixel compared to the 1cm sensor. NR the 40mp image to get effective 10MP resolution and you'll have lower noise than the 1cm sensor at similar resolution. All else being equal (sensor technology, etc) bigger sensor and higher resolution win out..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #13

Nickleback wrote:.

It's the sensor size that counts, not the size of the individualpixels. The 2cm sensor has less noise compared to the 1cm sensorbecause it is larger..

Only if the resolution is the same I think, the 2cm sensor has less noise becaue it has a larger noise sample to subtract. but a higher resolution sensor will also get a better(more acurate) noise sample. now I think my head is going to explode. obviously to get more MP in say a full frame sensor you have to make the pixels smaller so size does matter ..

Comment #14

PhotonFiend wrote:.

Nickleback wrote:.

It's the sensor size that counts, not the size of the individualpixels. The 2cm sensor has less noise compared to the 1cm sensorbecause it is larger..

Only if the resolution is the same I think.

Assume the larger sensor has the same size pixels as the smaller sensor. You'll get the same noise per pixel, but the larger sensor will simply have more of them, giving you higher resolution. Or NR the image from the larger sensor. You'll get lower noise, and resolution similar to the smaller sensor..

Or have the larger sensor with the same resolution as the smaller sensor, therefore the pixels on the larger sensor are... larger. You'll get lower noise than the smaller sensor, but the same resolution..

So to sum it up... larger sensors give lower noise. Period, that's it, done..

However, on any particular sensor size, having more pixels gives you the choice of high resolution or low noise, less pixels takes away that choice and gives you only low noise..

Now I think my head is going to explode..

Careful!.

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #15

Nickleback wrote:.

Chuxter wrote:.

It's clear that you can't stand low MP..

I'm fine with low MP. Given your cameras it seems that you aren't..

It's impossible to tell. The issue is that at the time of purchase, there most often isn't an option to buy a comparable lower MP camera. Sony didn't ask potential buyers if they wanted an 8MP or a 10MP sensor in the R1...they just took the sensor from the Nikon D2X and removed a few pixels. Simarly, I didn't see an option for the D300 that gave me fewer pixels!.

The "rule" is still a good one..

Nope. We've discussed this before..

I know. You still don't see that it's worded VERY carefully, so that it's almost impossible to fault...if you are reasonable..

Go compare noise and detail ofD40 and D300 in one of the still-life shots at the same ISO. Fornoise look at the color checker, for detail look at the knurls on thesalt & pepper shakers, or the strings on the fiddle on the bottle,the pattern on the wine vinegar label, or any other fine detail you'dlike to pick out..

That is your rationalle for WHY you can't stand low MP. It's fine for you to take that position. But you should not attempt to force your rationalle on others. There are equally valid reasons for preferring big photosites instead of many photosites...and equally impressive evidence..

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #16

Nickleback wrote:.

PhotonFiend wrote:.

If I take a 1cm sensor that is 10mp, thats3333x3333... simple. each pixel is 300 um is size. if now I take a2cm sensor that is 10mp, I still have 3333x3333 pixels but now theyare 600um in size. that is 4 times the area..

It's the sensor size that counts, not the size of the individualpixels. The 2cm sensor has less noise compared to the 1cm sensorbecause it is larger..

I don't want to start a long debate with you, but I want PhotonFiend to know that not everybody agrees with that opinion. It's my belief that it's the size of the bucket (photosite) that is important. BTW, Bayer sensor don't have "pixels"...they have photosites. The pixels are created in the little processor inside the camera (or the big processor in your computer) afterwards..

Also, I have never seen an example of lower noise by cropping/grouping pixels from a high-pixel-count camera. In theory it should work that way, but so far, nobody has shown conclusive proof that in practice it does. Do you have a link to good proof?.

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #17

Chuxter wrote:.

I know. You still don't see that it's worded VERY carefully, so thatit's almost impossible to fault...if you are reasonable..

The first rule is fine. The second rule, whether carefully written or not, is incorrect. Which makes it quite easy for a reasonable person to fault..

Go compare noise and detail ofD40 and D300 in one of the still-life shots at the same ISO..

That is your rationalle for WHY you can't stand low MP..

Nope, it's simply evidence than smaller pixels doesn't equal more noise. Size of the sensor is what counts (your rule #1), not size of the pixels (your rule #2)..

It's fine foryou to take that position. But you should not attempt to force yourrationalle on others..

It seems to me that oyu are forcing your position on others, despite evidence to the contrary..

There are equally valid reasons for preferringbig photosites instead of many photosites...and equally impressiveevidence..

Show me..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #18

Chuxter wrote:.

I don't want to start a long debate with you.

Me neither. We've had this discussion before, I showed ample evidence, you didn't respond to it or offer your own..

Also, I have never seen an example of lower noise bycropping/grouping pixels from a high-pixel-count camera. In theory itshould work that way, but so far, nobody has shown conclusive proofthat in practice it does. Do you have a link to good proof?.

I never said cropping/grouping, I said NR. I posted examples. Please respond..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #19

Nickleback wrote:.

Chuxter wrote:.

I don't want to start a long debate with you.

Me neither. We've had this discussion before, I showed ampleevidence, you didn't respond to it or offer your own..

Also, I have never seen an example of lower noise bycropping/grouping pixels from a high-pixel-count camera. In theory itshould work that way, but so far, nobody has shown conclusive proofthat in practice it does. Do you have a link to good proof?.

I never said cropping/grouping, I said NR. I posted examples.Please respond..

I may have misunderstood you when you said:.

"With higher MP you have the choice of more detail or low noise. With less MP you get low noise but don't get the choice of more detail.".

I didn't interpret that as referring to NR. My bad?.

Some sages contend that it's good to have lots of pixels because if low-noise is the goal, it's easy to simply apply a "binning" algorithm, which in essence is simply fewer, larger photosites (when using RAW data). I was saying that while in theory it should work that way, I hadn't seen evidence that in practice "binning" works to lower noise..

I'll contend that with NR, you never get the "choice of more detail". The early experiences with the H9 are my evidence. There are hundreds of posts on STF about the selective smearing (usually on big areas of grass) that the H9 occasionally produced. Here is one example:.

Http://www.flickr.com/photos/clicky/478805764/sizes/o/.

Note that some of the pix from this juncture (April 2007) have been removed...for example, the full-size picture where this 100% crop was extracted is gone..

The full thread might be interesting to you?.

Http://forums.dpreview.com/...forums/read.asp?forum=1009&message=23052588.

This was a rather bad time for STF. Several members got REALLY bent outa shape!.

The problems with the H9 (and other cameras of the same vintage, like the T100) were caused by Sony trying a new "smart" NR algorithm...it apparently looked for areas that would not be too badly affected and applied NR selectively to these areas. That IS a good idea, since good PS pilots do that all the time w/o serious problems. However, it must be a difficult task for a little uP to do quickly?.

I do think that the D300 does a rather good job of NR, but it has a MUCH more capable uP! None-the-less, even with the D300's sophisticated NR, it's obvious that detail is lost as noise is scrubbed..

My bottom line is that I think your "With higher MP you have the choice of more detail or low noise" is a bad oversimplification at best. I think it would be better if revised to state, "With higher MP you have the choice of more noise or less detail.".

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #20

Chuxter wrote:.

"With higher MP you have the choice of more detail or low noise. Withless MP you get low noise but don't get the choice of more detail.".

I didn't interpret that as referring to NR. My bad?.

OK..

Some sages contend that it's good to have lots of pixels because iflow-noise is the goal, it's easy to simply apply a "binning"algorithm.

...which is the most simple (but not best) NR. Just like demosaic algorithms, there are many ways to do it..

I'll contend that with NR, you never get the "choice of more detail"..

You most certainly do. Assuming the lens can keep up, you get more detail with more MP. For instance, the 12mp D300 resolves roughly 40% more detail than the 6mp D40:.

Http://www.dpreview.com/reviews/nikond300/page31.asphttp://www.dpreview.com/reviews/nikond40/page24.asp.

The early experiences with the H9 are my evidence. There are hundredsof posts on STF about the selective smearing (usually on big areas ofgrass) that the H9 occasionally produced. Here is one example:.

So that's an example of a bad implementation. Doesn't change the point. Shoot raw and/or with NR disabled (if available) and use NR software on a computer..

I do think that the D300 does a rather good job of NR, but it has aMUCH more capable uP! None-the-less, even with the D300'ssophisticated NR, it's obvious that detail is lost as noise isscrubbed..

Did you look at the detail in the sample shots? The knurls on the salt & pepper shakers, the details (including the strings on the fiddle)? The D300 starts with more detail, and it's clear even with NR applied the D300 still maintains more detail..

I don't feel like cutting up those images, so let's just used the ISO 1600 crops on this site:.

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

One of these photos has noticeably less noise (check the black at the bottom) and a bit more detail (check the tree and fence). Blow it up to 200% if you like with the zoom in button. Guess which camera is which..

My bottom line is that I think your "With higher MP you have thechoice of more detail or low noise" is a bad oversimplification atbest..

My bottom line is that I think your "Buy a camera with as few pixels as you can stand" is a gross oversimplificaton..

I think it would be better if revised to state, "With higher MPyou have the choice of more noise or less detail.".

You most certainly have the choice of more detail. For example, the D300 gives more detail than the D40 at any ISO with NR disabled. It just has to, it's got twice as many pixels. Unless you use a really crappy lens on the D300 and a really good lens on the D40, you will get more detail on the D300. Period..

Of course with NR disabled you have more noise on the D300 than the D40. At low ISO the noise isn't objectionable, so you can go for max detail. At high ISO you might find the noise objectionable..

So turn high ISO NR on the D300. What happens? You trade off detail for less noise. But that's not too bad, as you started off with more detail. In the case of the D300 vs D40, the D300 has significantly less noise and a bit more detail at high ISO..

I expect if the D300 were tested against a camera with a newer 6mp sensor (if there were such a beast), the high ISO noise would be about the same..

Comparing two cameras, one with more MP than the other....

With more MP you get to choose more detail or low noise. With lower MP you don't get to choose more detail (you simply don't have it) but you do have low noise..

Higher MP wins out..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #21

I'll contend that with NR, you never get the "choice of more detail"..

You most certainly do. Assuming the lens can keep up, you get moredetail with more MP. For instance, the 12mp D300 resolves roughly40% more detail than the 6mp D40:.

Http://www.dpreview.com/reviews/nikond300/page31.asphttp://www.dpreview.com/reviews/nikond40/page24.asp.

Now we're getting near the crux of the situation. How much detail does a person need? I'd argue it depends on the output medium and size. A crt/lcd screen at 1440x? A 4"x print? An 8"x print? A 16"x print? Amount of cropping?.

I contend a 6MP SLR with minimal cropping is fine for an 8"x12" print..

In a practical sense buying a camera with much more resolution than you can use for your desired output is not optimal. The money is better spent on a good tripod or lens than a body with more capability than is necessary. IMHO...

Comment #22

Mrxdimension wrote:.

I'll contend that with NR, you never get the "choice of more detail"..

You most certainly do. Assuming the lens can keep up, you get moredetail with more MP. For instance, the 12mp D300 resolves roughly40% more detail than the 6mp D40:.

Http://www.dpreview.com/reviews/nikond300/page31.asphttp://www.dpreview.com/reviews/nikond40/page24.asp.

Now we're getting near the crux of the situation. How much detaildoes a person need?.

The more the better. But it doesn't really matter, the point is that noise is related to sensor size. At the same level of detail, regardless of how you got there, noise is the same..

I contend a 6MP SLR with minimal cropping is fine for an 8"x12" print..

You're on the edge. Even with no cropping you are at 250 ppi. Crop 20% off and you are at 200 ppi. Most print processors work best with 300 to 320 ppi. A more realistic number is to start with 400 ppi, so you have room to crop 20% off and still have 320 ppi..

400 ppi at 8x12 is 15mp..

In a practical sense buying a camera with much more resolution thanyou can use for your desired output is not optimal. The money isbetter spent on a good tripod or lens than a body with morecapability than is necessary. IMHO..

And I think a photography course is money better spent than a good tripod, lens or body. You are upgrading the most important part of the camera..

As always, it depends on how much money you want/have to spend..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #23

Nickleback wrote:.

Mrxdimension wrote:.

I'll contend that with NR, you never get the "choice of more detail"..

You most certainly do. Assuming the lens can keep up, you get moredetail with more MP. For instance, the 12mp D300 resolves roughly40% more detail than the 6mp D40:.

Http://www.dpreview.com/reviews/nikond300/page31.asphttp://www.dpreview.com/reviews/nikond40/page24.asp.

Now we're getting near the crux of the situation. How much detaildoes a person need?.

The more the better. But it doesn't really matter, the point is thatnoise is related to sensor size. At the same level of detail,regardless of how you got there, noise is the same..

I contend a 6MP SLR with minimal cropping is fine for an 8"x12" print..

You're on the edge. Even with no cropping you are at 250 ppi. Crop20% off and you are at 200 ppi. Most print processors work best with300 to 320 ppi. A more realistic number is to start with 400 ppi, soyou have room to crop 20% off and still have 320 ppi..

Maybe. I learned to shoot slides and crop in the camera at exposure time..

400 ppi at 8x12 is 15mp..

I have a 14+MP and a 6MP body. At 8x print it's difficult to tell the difference, at 11x print I can see the difference if I look closely..

In a practical sense buying a camera with much more resolution thanyou can use for your desired output is not optimal. The money isbetter spent on a good tripod or lens than a body with morecapability than is necessary. IMHO..

And I think a photography course is money better spent than a goodtripod, lens or body. You are upgrading the most important part ofthe camera..

As always, it depends on how much money you want/have to spend..

Dang. You got me on that one ..

Comment #24

Over the debate between Nickleback and Chuxter..

If I get this right, the optimum camera would be like a D3 full frame sensor (large sensor for Mr. Nickleback) with 1 pixel (no, not one megapixel, but a single pixel)..

That would be the ultimate, humongous photosite (bucket) size to satisfy Mr Chuxter...

Comment #25

Nickleback wrote:.

Chuxter wrote:.

Some sages contend that it's good to have lots of pixels because iflow-noise is the goal, it's easy to simply apply a "binning"algorithm.

...which is the most simple (but not best) NR. Just like demosaicalgorithms, there are many ways to do it..

If we had "binning" in cameras, it would deliver essentially the same IQ as sensors with 1/4 the number of pixels. IE, a 12 MP sensor would have 4 pixels "binned" before any processing other than demosaicing. It should look like a 3 MP image of comparable technology, vintage, and manufacture..

I'll contend that with NR, you never get the "choice of more detail"..

You most certainly do. Assuming the lens can keep up, you get moredetail with more MP. For instance, the 12mp D300 resolves roughly40% more detail than the 6mp D40:.

Nah, you didn't correctly read what I wrote. I said that there is NO noise reduction scheme that increases the detail. All NR reduces detail..

The early experiences with the H9 are my evidence. There are hundredsof posts on STF about the selective smearing (usually on big areas ofgrass) that the H9 occasionally produced. Here is one example:.

So that's an example of a bad implementation. Doesn't change thepoint. Shoot raw and/or with NR disabled (if available) and use NRsoftware on a computer..

I do. Yes, everybody except Sony and a guy named "AAK" agrees that Sony didn't do a good job with there early-2007 NR schemes. But their failure shows clearly what NR does...it blurs the image! If they hadn't tried to implement a selective NR scheme, it wouldn't be as obvious, thus a bad example..

I do think that the D300 does a rather good job of NR, but it has aMUCH more capable uP! None-the-less, even with the D300'ssophisticated NR, it's obvious that detail is lost as noise isscrubbed..

Did you look at the detail in the sample shots? The knurls on thesalt & pepper shakers, the details (including the strings on thefiddle)? The D300 starts with more detail, and it's clear even withNR applied the D300 still maintains more detail..

I don't feel like cutting up those images, so let's just used the ISO1600 crops on this site:.

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

One of these photos has noticeably less noise (check the black at thebottom) and a bit more detail (check the tree and fence). Blow it upto 200% if you like with the zoom in button. Guess which camera iswhich..

I can't guess. If I wanted to impress you, I'd just go look at the reviews. .

One appears to have more contrast...not show more detail. It also appears to have been sharpened more, which makes the noise more obvious..

My bottom line is that I think your "With higher MP you have thechoice of more detail or low noise" is a bad oversimplification atbest..

My bottom line is that I think your "Buy a camera with as few pixelsas you can stand" is a gross oversimplificaton..

Perhaps..

I think it would be better if revised to state, "With higher MPyou have the choice of more noise or less detail.".

You most certainly have the choice of more detail. For example, theD300 gives more detail than the D40 at any ISO with NR disabled. Itjust has to, it's got twice as many pixels. Unless you use a reallycrappy lens on the D300 and a really good lens on the D40, you willget more detail on the D300. Period..

Well, it somewhat depends on what you mean when you say "You have the choice of more detail". The D300 has more resolution than the D40, but neither has an option (other than sharpening and contrast) that allows you to choose more detail. They both sorta have the detail that they have! If you are saying that choosing the D300 over the D40 gives you more detail, then that's right. BUT, when you go to high ISO with the D300, NR kicks in and details get blurry quickly. I'm not saying that the D300 doesn't have great IQ...just that even at base 200 sensitivity, it has a noticeable amount of noise. MANY owners have griped about that noise!.

Of course with NR disabled you have more noise on the D300 than theD40. At low ISO the noise isn't objectionable, so you can go for maxdetail. At high ISO you might find the noise objectionable..

So turn high ISO NR on the D300. What happens? You trade off detailfor less noise. But that's not too bad, as you started off with moredetail. In the case of the D300 vs D40, the D300 has significantlyless noise and a bit more detail at high ISO..

Yes, but at low ISO, the D40 has basically no noise, whereas the D300's noise is quite visible..

I expect if the D300 were tested against a camera with a newer 6mpsensor (if there were such a beast), the high ISO noise would beabout the same..

Comparing two cameras, one with more MP than the other....

With more MP you get to choose more detail or low noise. With lowerMP you don't get to choose more detail (you simply don't have it) butyou do have low noise..

Higher MP wins out..

If your goal is ONLY low light pix, then the D300 is a good choice...about as good as the D40. But the D3 is king. It has bigger buckets!.

Bigger photosites win out. .

This discussion reminds me of the old small engine vs big engine debate. I'm a small engine kinda guy (emotionally), but I DO understand (logically) that you can't beat cubic inches..

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #26

Doug Haag wrote:.

Over the debate between Nickleback and Chuxter.If I get this right, the optimum camera would be like a D3 full framesensor (large sensor for Mr. Nickleback) with 1 pixel (no, not onemegapixel, but a single pixel).That would be the ultimate, humongous photosite (bucket) size tosatisfy Mr Chuxter..

LOL!.

Rule 1.b. states, "Buy a camera with as few pixels as you can stand.".

I may be odd, but I can't stand a 1-pixel camera. .

My threshold is around 5-6 MP. I can take good pix with that level of resolution. Many users of the Nikon D2H, which has what, 4 MP, are quite happy with the results. It has 4,000,000 really sensitive photosites! It's even better than the D3 for low light (would be even better with a modern processor and firmware to do a bit of NR)..

The whole point of Rule 1.b. is that EVERYBODY can have their own threshold for number of pixels, thus, they can make a different decision that the next guy..

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #27

I cut everything out, let's just down to it..

Comparing the D300 to the D40. Each has a DX sensor (i.e. same size). The D300 has 12mp, the D40 has 6mp..

Statement 1:.

The D300 has detail that you get from 12MP. The D40 has the detail you'd expect from 6MP. Therefore with the D300 you can choose to get more detail..

Statement 2:.

The D40 has lower noise than the D300. So NR the D300 image so that you end up with detail you'd expect from 6MP. Both images now have the same detail. But also, both images will have similar noise. In fact the D300 has lower noise, but I assume it's due to the D40 sensor being a generation or two older than the D300 sensor..

Got it?.

So, as I said, with higher MP you get to choose detail or low noise, with lower MP you only get low noise. That is, of course, in comparison to one another..

Thus "lowest MP you can stand" is backwards. You get more (not less!) with higher MP..

P.S. The D40 is the top image, the D300 is the bottom. The D300 image has much less noise and a little more detail..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #28

Chuxter wrote:.

If your goal is ONLY low light pix, then the D300 is a goodchoice...about as good as the D40. But the D3 is king. It has biggerbuckets!.

It has a bigger sensor!.

Bigger photosites win out. .

It would win out if it had 4mp, 12mp, 20mp or 40mp. It wins out because the sensor is bigger, not because of the size of the buckets..

This discussion reminds me of the old small engine vs big enginedebate. I'm a small engine kinda guy (emotionally), but I DOunderstand (logically) that you can't beat cubic inches..

You can beat cubic inches. Bigger engines weigh more. Look at almost any Lotus to see how to do it with a smaller engine. But it's a poor analogy to sensor size..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #29

Chuxter wrote:.

Many users of the Nikon D2H, which has what, 4 MP, arequite happy with the results. It has 4,000,000 really sensitivephotosites! It's even better than the D3 for low light.

Now I know for certain that you were dropped on your head..

Http://www.dpreview.com/reviews/NikonD2H/page19.asphttp://www.dpreview.com/reviews/nikond3/page18.asp.

I'll just post the luminance noise. You can check the reviews for RGB noise, but the trend is the same. The D2H luminance graph is on a grey patch, so I'll pick the grey patch graph from the D3 review too (black patch has less noise, of course)..

The D3 is about 2 stops better than the D2H. Look at the noise at any ISO on the D2H, then look at the same noise level on the D3 and check the ISO....

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

NR the D3 image so that it has similar resolution as a D2H image and you'll have a 4 stop noise advantage..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #30

I'm interested in this discussion... I was not aware that a higher MP image could be "downconverted" somehow in a way which reduces noise. Is there any special tools/techniques to do this or will a simple resize using any image editing software suffice?.

I think even with your explanation however I am not completely convinced. I suspect that often a manufacturer will cram more pixels onto a sensor than they probably should in order to boast a high "megapixels" number on the camera (since it seems this is what the average Joe looks at when buying). By going over the top in this way they are introducing more noise into the picture since each detector on the sensor receives a much lower number of light photons than it would with a sensor of the same size but lower MP. The amount of background "noise" that the sensor picks up from among other things, it's own electrical functioning is the same... Which means that each receptor on the sensor gets less photons, but the same amount of noise... This would surely result in a noisier image than a lower MP sensor, which I think we all agree on.

Wouldn't that essentially not only combine the number of photons that multiple receptors had captured, but also combine the amount of noise that multiple receptors had captured?..

Comment #31

Am spining faster and faster ... ohhhhhhhhhhhhhhhhhh.

Do I need to know all this just to take a decent picture .

Sorry ...  just being funny.

Nickleback wrote:.

Chuxter wrote:.

Many users of the Nikon D2H, which has what, 4 MP, arequite happy with the results. It has 4,000,000 really sensitivephotosites! It's even better than the D3 for low light.

Now I know for certain that you were dropped on your head..

Http://www.dpreview.com/reviews/NikonD2H/page19.asphttp://www.dpreview.com/reviews/nikond3/page18.asp.

I'll just post the luminance noise. You can check the reviews forRGB noise, but the trend is the same. The D2H luminance graph is ona grey patch, so I'll pick the grey patch graph from the D3 reviewtoo (black patch has less noise, of course)..

The D3 is about 2 stops better than the D2H. Look at the noise atany ISO on the D2H, then look at the same noise level on the D3 andcheck the ISO....

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window.

NR the D3 image so that it has similar resolution as a D2H image andyou'll have a 4 stop noise advantage..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed.

Be a torch bearer ...S3 IS..

Comment #32

Zadam wrote:.

I'm interested in this discussion... I was not aware that a higherMP image could be "downconverted" somehow in a way which reducesnoise. Is there any special tools/techniques to do this or will asimple resize using any image editing software suffice?.

Resizing is the simplest noise reduction algorithm, but not the best. With better NR algorithm the image is not downconverted, but the smearing of detail reduces effective resolution. Apply enough NR to bring the detail down to the level of a lower resolution sensor and you will have similar noise..

I think even with your explanation however I am not completelyconvinced. I suspect that often a manufacturer will cram more pixelsonto a sensor than they probably should in order to boast a high"megapixels" number on the camera (since it seems this is what theaverage Joe looks at when buying)..

I don't see that as an issue, at least with the pixel sizes on DSLRs. If you aren't worried about the noise (for example, at low ISO), you get more detail than a lower MP sensor. If you are worried about the noise (for example, at high ISO), apply NR and get similar detail and noise as a lower MP sensor..

It's a win-tie (higher MP) vs a lose-tie (lower MP)..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #33

Peacewarrior wrote:.

Do I need to know all this just to take a decent picture .

Sorry ...  just being funny.

The most important part of a camera is the nut behind the viewfinder. Upgrade that nut and your camera, whatever it is, will take better pictures..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #34

Zadam wrote:.

Now you say to reduce thesize of the higher MP image... Wouldn't that essentially not onlycombine the number of photons that multiple receptors had captured,but also combine the amount of noise that multiple receptors hadcaptured?.

Hi Zadam, I can answer your question here. By downsizing (i.e. combining a few pixels to get an average pixel), we are not adding the noise together, but we are averaging out the noise. It is like we are taking the mean of the noise. Since noise is usually random with 0 mean, the more we average, the more it will get closer to 0..

As a result, if you take a photo with 1/1.8" 12MP camera and downsized it to 6MP, the downsized photo is not necessary noisier than the same photo from a 1/1.8" 6MP camera on pixel levels...

Comment #35

Nickleback wrote:.

Chuxter wrote:.

If your goal is ONLY low light pix, then the D300 is a goodchoice...about as good as the D40. But the D3 is king. It has biggerbuckets!.

It has a bigger sensor!.

...with bigger photosites (buckets). How can you be so sure that it's not the bigger photosites that makes it the low-light king?.

Bigger photosites win out. .

It would win out if it had 4mp, 12mp, 20mp or 40mp..

At 15+ MP, the D3 photosites will be smaller than the ones on the D40 and noise will be higher. At 40 MP, the D3 would have tiny photosites (about 2/3 the size of the ones on the D300)!.

It wins outbecause the sensor is bigger, not because of the size of the buckets..

Isn't this somewhat of a chicken vs egg point?.

This discussion reminds me of the old small engine vs big enginedebate. I'm a small engine kinda guy (emotionally), but I DOunderstand (logically) that you can't beat cubic inches..

You can beat cubic inches. Bigger engines weigh more. Look atalmost any Lotus to see how to do it with a smaller engine..

As I said, I'm a small engine kinda guy, so you are preaching to the choir. But on Monday morning, I want a bigger small engine. .

BTW, Loti are light weight because Chapman only designed them to go around the block once before needing repairs. .

But it's a poor analogy to sensor size..

If you have a better analogy, let's hear it....

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #36

Nickleback wrote:.

Chuxter wrote:.

Many users of the Nikon D2H, which has what, 4 MP, arequite happy with the results. It has 4,000,000 really sensitivephotosites! It's even better than the D3 for low light.

Now I know for certain that you were dropped on your head..

Hey, my momma mighta been clumsy, but she loved me (even after the accident)..

OK, you got me. But remember, the D2H and the D3 are not exactly comparable. The D2H is a 2003 vintage and it used Nikon's LBCAST technology (which wasn't a success). The D3 has MUCH better NR too..

Charlie DavisNikon 5700, Sony R1, Nikon D300HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/'Experience: Discovering that a claw hammer will bend nails.Epiphany: Discovering that a claw hammer is two tools...'..

Comment #37

Chuxter wrote:.

Nickleback wrote:.

Chuxter wrote:.

If your goal is ONLY low light pix, then the D300 is a goodchoice...about as good as the D40. But the D3 is king. It has biggerbuckets!.

It has a bigger sensor!.

...with bigger photosites (buckets). How can you be so sure that it'snot the bigger photosites that makes it the low-light king?.

Because noise scales with sensor size, not pixel size..

Did you look at the D300 vs D40 example?.

At 15+ MP, the D3 photosites will be smaller than the ones on the D40and noise will be higher..

...per pixel. However if you NR the image so that it has detail similar to what you get with 6mp, you'll have a lot less noise. The D3 has a 2-stop noise advantage over the D2H, despite having slightly smaller pixels. NR the D3 image to 4mp effective resolution and you'll have a 4-stop advantage..

Theoretically you should have a bit more than 1 stop advantage going from DX to FX, the likely reason the D3 enjoys a 4 stop noise advantage at the same level of detail is the D3 sensor is much more advanced than the D2H sensor..

The D3 has a bit more than 1 stop advantage over the D300 at the same level of detail (i.e. NR off on D300), which is what you'd expect from sensors of similar design..

At 40 MP, the D3 would have tiny photosites(about 2/3 the size of the ones on the D300)!.

And the D300 has pixels 1/2 the area of the D40, but still wins on noise at similar levels of detail..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #38

Chuxter wrote:.

OK, you got me. But remember, the D2H and the D3 are not exactlycomparable..

I never said they were! You did, based on what D2H shooters said. It's not surprising that D2H shooters say this, it's natural to rationalize ones choices, even more so with expensive choices..

The D2H is a 2003 vintage and it used Nikon's LBCASTtechnology (which wasn't a success). The D3 has MUCH better NR too..

The D3 kills the D2H without NR. Throw in NR and it slaughters it..

Seen in a fortune cookie:Fear is the darkroom where negatives are developed..

Comment #39

Thanks for that explanation, I never thought of it that way but it does make sense.....

Comment #40

Zadam wrote:.

Thanks for that explanation, I never thought of it that way but itdoes make sense....

You are welcome. That's why the pixel-to-pixel comparison of new 8MP 1/2.5" with old 3MP 1/2.5" doesn't make sense (but some still do so, edit: including the old me). The sensor technology has improve over the years that if you scale down this new 8MP to 3MP, it is much cleaner than the old 3MP..

But of course if we compare new 8MP 1/2.5" resized to 6MP with old legendary Fuji F31 (6MP 1/1.7" Super CCD), the Fuji F31 might still win most of the time because F31 not only have bigger sensor, but also different technology...

Comment #41

Click Here to View All...

Sponsored Amazon Deals:

1. Get big savings on Amazon warehouse deals.
2. Save up to 70% on Amazon Products.


This question was taken from a support group/message board and re-posted here so others can learn from it.

 

Categories: Home | Diet & Weight Management | Vitamins & Supplements | Herbs & Cleansing |

Sexual Health | Medifast Support | Nutrisystem Support | Medifast Questions |

Web Hosting | Web Hosts | Website Hosting | Hosting |

Web Hosting | GoDaddy | Digital Cameras | Best WebHosts |

Web Hosting FAQ | Web Hosts FAQ | Hosting FAQ | Hosting Group |

Hosting Questions | Camera Tips | Best Cameras To Buy | Best Cameras This Year |

Camera Q-A | Digital Cameras Q-A | Camera Forum | Nov 2010 - Cameras |

Oct 2010 - Cameras | Oct 2010 - DSLRs | Oct 2010 - Camera Tips | Sep 2010 - Cameras |

Sep 2010 - DSLRS | Sep 2010 - Camera Tips | Aug 2010 - Cameras | Aug 2010 - DSLR Tips |

Aug 2010 - Camera Tips | July 2010 - Cameras | July 2010 - Nikon Cameras | July 2010 - Canon Cameras |

July 2010 - Pentax Cameras | Medifast Recipes | Medifast Recipes Tips | Medifast Recipes Strategies |

Medifast Recipes Experiences | Medifast Recipes Group | Medifast Recipes Forum | Medifast Support Strategies |

Medifast Support Experiences |

 

(C) Copyright 2010 All rights reserved.