Sony's new sensor...

My analogy is correct, but I don't feel there is anything that I could say to change your thoughts. The 'bucket' is a classic analogy and I was so-hoping that it would help - but.....

Perhaps putting words into action might be best. Taking two CURRENT cameras and doing a pixel vs pixel test. Nikon D3 vs 12 mp digicam, go out in the evening, turn off noise reduction - then shoot and compare.

John, I strongly feel that after the point of resolution is reached where mp's are concerned, anything else is a marketing ploy to snooker individuals who actually believe that more mp's are better. Recently, my neice was searching for a camera - and I recommended one to her; she shared this info with her brother who shook his head and told her, "NO NO NO. Get the one with the most megapixels."

Pixels are great - and are necessary, and as everyone has agreed - a certain amount are necessary in order to produce a great photo. But too many equal unused pixels which are incapable of being resolved. :-(

Sky
 
but I find that thread interesting because the OP is getting pretty good results from his camera but he thinks that the results are "horrible". This is the response, I think, of an inexperienced photographer who is seeing soft unsharpened images right out of his Camera.

If you work with that shot in post, you can get a very detailed sharp punchy image.
To an experienced photographer it's nice output, I think.

So here is the dilemma that modern camera manufacturers find themselves in. They can let the camera have good basic images like the W300 but the first time camera users think that the shots are "horrible". If they punch up the shots so that the new users like them, then the experienced users find artifacts and problems in the processing.

Many experienced photographers are prejudiced against high megapixel cameras, so they won't even give the output of the new cameras a second glance. But no one here is like that, I'm confident.
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
but I don't think it's correct. Big pixels are not like a big bucket of water. Only the surface of the water counts.

It's more like a big mirror, and a mirror that has been broken into 100 million pieces. The big mirror is being observed by a machine that can only see light or no light. If the big mirror reflects light then the machine registers the number 1. If there is not enough light reflected by the big mirror then the machine registers 0.

The fragmented mirror is much more sensitive to the light in the room. Some pixels will register 1 and some will register 0. The machine can make a much more meaningful image from this input.

Big pixels are ignorant. They can't understand very much. Small pixels are big pixels who have been to university. They are much more capable of seeing what is going on and reporting it back to us.

Noise is not much of a factor in a big pixel because its report is not very sophisticated. Noise is part of the 100 million things that the fractured mirror is reporting. But there is the same amount of noise in the big mirror as in the fractured mirror, because the mirror doesn't make the noise.
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
I like ARTCIC's theory better, after all the big bucket has a lager
surface, we are not counting it's depth.
The more signal, the less noise produced per pixel..

Keep on making memories..
 
David,

If 20 years ago one would come and tell me that in 20 years I can throw away my film cameras because there will be digital cameras that will work without a film and have even better quality and less noise - I would have assumed that he run away from a psychiatric hospital.

A tiny 100MP sensor with virtually no noise sounds completely crazy and impossible, for the moment. With the way technology advanced, I wouldn't be surprised if in 10 years it will already be history.

Cheers
Moti

--
ARPS - ABPPA
My new website here:
http://www.pixpix.be
 
If we take out the seats from the bus we can fit 60 people in and we
don't use any more gas. The 60 people don't represent more light,
but they represent more variety and detail.
Good metaphor and perfectly to the point. You just forgot to mention that if you put 60 people in a bus without seats they will start complaining and they will make so much noise that you'd hardly be able to see any details...
LOL I like it.

Cheers
Moti

--
ARPS - ABPPA
My new website here:
http://www.pixpix.be
 
a comparison between lets say the Nikon D3 and Sony 900 ..........maybe that will end all debates about more Megapixels or not.

for me less is still more or better what ever u call it.

in the point and shoot section is more megapixels of the latest generation really better if it comes to quality and noise department. ?

just my two cnts.
--
  • living in harmony with nature and other beings...will create an better world for all * marti58 -2006
http://www.flickr.com/photos/marti58/
http://www.flickr.com/groups/worldwidefriendship/
 
Yes I see light or no I don't see light.

A large pixel is less likely to be triggered by noise but when it does give noise, for example when underexposed, the noise is big and blotchy.

A very small pixel might be more likely to make a mistake and say there was light, when there really wasn't. But that mistake would be tiny and unimportant in all the correct information around it. When underexposed, the character of the noise recorded by a small pixel sensor, tends to be less unpleasant.
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
I'm definitely no expert on this topic, but after reading this thread, I'm surprised no one has mentioned high ISO. It seems that generally speaking at low ISO's there is so much signal that it practically doesn't matter the size of your pixel sites. Cameras at ISO 80 and 100 will generate noise free images. However, boost the ISO to 400, 800, or beyond and noise will appear. The reason we increase ISO is because the amount of light is too low. So we use the sensor to increase the "gain" on the light signal. This doesn't actually increase the amount of light, just how the circuitry processes what is there. In doing so more and more noise is introduced into the picture that we see. The smaller the pixels are the less light will hit any given pixel. This causes the camera to increase the noise more on cameras with more and smaller pixels than with less and larger pixels. I don't know how clear this is, but if it's not perhaps one of our other fine members who has more technical expertise can clarify. I hope this adds something to our discussion!

--
Ben Ereshefsky
http://flickr.com/photos/42189183@N00/
 
John....your 'educated pixel' theory is sending me into stitches. :-)

Nonetheless, I suppose at the end of the day that it really doesn't matter what any of us think. It is what it is.

AND - I wanted you to know that I thought about you amid my recent trip. We came across these gorgeous water lilies. Unfortunately, they were very far away and I didn't get any close-ups. Sure were pretty, though.

Sky
 
noise is generated by the sensor, because more power is put through it to increase its gain. I just want to point out that the noise is generated by the sensor and not by the pixels.

We see noise in shadows and in uniform colours like a blue sky. Big pixels are dumb and this works in their favour at high ISOs. Small pixels are clever and show a lot of the noise that is being generated by the sensor. Because the background is uniform, it's fairly easy to clean up this noise if you wish to. The noise shown by small pixels is not as unattractive as the noise shown by big pixels.

However, some recent tests have shown small pixels to do better than large ones at ISO1600. That is, they record more detail, and the image is less damaged by noise. Modern sensors are overcoming the noise problems associated with turning up the gain.
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
for hijacking your thread with this pixel size debate.

I think that the kind of camera you describe might be very welcome at this point in time. Lloydy and Sky are using the new Fuji camera that has a 2/3rds sensor and a 14 times zoom. They both are very pleased with it.

I like Sony too, and if Sony brought something like this I'm sure it would have cool features and great output.

The F707 and the R1 were resolution kings when they came out. The excellent lens and the higher pixel count, (Canon's G3 was only 4 megapixels) gave them a much more detailed image than the cameras that came out at that time.

I think that the new a900 will be resolution king again, but the entry price for camera and lens is going to be up around $5000.
I think the kind of camera you are wishing for makes a lot more sense for me.
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
Did you have to sit all day and wait for that pose??
A really really good shot..
Keep on making memories...
 
I'm the worst kind of guy to listen to: opinionated without any deep
knowledge but this is what I think.
You know you don't have the knowledge. So why don't you go back to learn more and then make a more educated opinion?
Small sensors, like the one in the W300, are only capable of catching
enough light to give you an excellent image up to about 8 by 10 size.
If you look at their output at 20 x 30 size, it may not look so good,
because at that size, the random nature of light is starting to show.
Tiny pixels have an advantage when it comes to noise, because they
chop the lumps of noise into tiny pieces. So the noise from a sensor
with tiny pixels doesn't look so bad.
We've seen watercolour effects and other problems from sensors with
small pixels, but this is a teething problem. The manufacturers are
trying to learn how to work with this kind of output. I suspect that
the processing engines in the cameras will continue to improve.
--
John Dunn
Portraits:

http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
The V1 has a sensor smaller sensor than W300 but is perfectly capable of printing 13"x19", or even 20"x27", as many STFers can testify. You lose.
 
Thank you. :-)

This recent shot hopefully illustrates that one can do a lot with just a pocket full of pixels. So many times we can talk until our tonsils fall out - and actions really do have the ability to speak louder than words - although it is important for everyone to have an opinion and be able to share their personal views. Whether we agree or not, we can still learn.

We have a Nikon 3 pixel camera that I also love. Takes the cleanest shots - and is a whiz for infared (something that I've started to dabble into AND it is so much fun!).

Thank you again, you wonderful guys you!!

Sky
 
Yes I see light or no I don't see light.
A large pixel is less likely to be triggered by noise but when it
does give noise, for example when underexposed, the noise is big and
blotchy.
Hey a sensor is NOT a digital device ( 1 or 0 ) it is an ANALOGUE device and each PHOTOSITE = 1 DOT ( that is 1/3 rd of a pixel ) delivers a continuous voltage value according to the AMOUNT of light it receives, which in turn is amplified by an electronic circuit that also amplifies the ELECTRICAL noise CREATED by the photosites ==> thus the famous Signal to Noise ratio.

The larger the photosites, the better the S/N ratio is and lesser the noise will be amplified Vs the actual light received and thus providing better noisefree pictures.

There is no such thing in nature as OPTICAL noise, just plain light. Therefore the sensor components do not record noise, they produce it, and so do every elementary electronic circuits thereafter up to the point when the data are converted into 1's and 0's ( digital data).

Now only discussing the sensor qualities/defects completely obliterates all the SOFTWARE processing that takes place inside the camera and where the manufacturer tries to eliminate this noise the best he can,

The nearest actual true data you can obtain from a sensor is when using the raw data, and even so the way data is DIGITALY converted and recorded in those raw digital files already is a data manipulation that does emphasize some values Vs others (data are not recorded in a linear fashion).
I have not read the whole thread, maybe this has been stated before ....
--
JJ.
 

Keyboard shortcuts

Back
Top