Reply
Thread Tools
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#41
That's a bogus site. They completely ignore sensor technology and basically made up an ideal pixel size.

We calculated how many pixels have space to be in a sensor if one pixel has a size of 3 µm which we believe to be a minimum size for good image quality at higher sensitivity levels (>ISO 400).
An engineering conclusion based on beliefs ? Hardly serious.
 

The Following 3 Users Say Thank You to attila77 For This Useful Post:
Posts: 203 | Thanked: 68 times | Joined on Oct 2009
#42
Originally Posted by titan View Post
ok, you're referring to video digital zoom, i.e. the output image has always much smaller resolution than the chip.
There are two approaches:

1.) take a fraction of the sensor image (the smaller the higher the zoom) and rescale it to video resolution (apparently, what N86 the does).
this should lead to a the lowest noise level w/o zoom, and maximum noise if you zoom to video resolution.
with even more zoom you get the digital zoom artefacts (interpolation).

2.) like 1., but take a fraction of the inner VGA-sized area of the sensor image

I don't think the N900 or N97 would use 2.)
Otherwise the video would always record the inner VGA region of a still image of the same scene. However, the video seems to be wide angle, if you don't zoom in:
http://www.nokiausers.net/N-Series/N...d-Natural.html
According to the documentation of the camera chip there is also a builtin resize function
and it is likely that the N900 would use it.

I think what I'm saying is that in actual reviews of the N86, people have reported that the video zoom works much better than in previous iterations of Nokia N series camera phones (and all other camera phones for that matter). I don't know all the technical aspects of how it works. I outlined what I understood above. It seems to me that the actual end results are more important. I'm also saying that, as Steve Litchfield reports at All About Symbian, to his knowledge the N86 is the only camera phone to use this improved method. The N97 doesn't have it, despite coming out at the same time, nor the N97-mini that came out after the N86. (Perhaps Nokia wants to keep a few things back to make it's flagship camera phone actually better as a camera?) So given that Nokia left it out of the N97 and N97-mini, it seems likely they will leave it out of the N900 as well. The 5MP sensor of the N900 and it's higher resolution video recording would also not work as well for this technique.
 
Posts: 946 | Thanked: 1,650 times | Joined on Oct 2009 @ Germany
#43
Originally Posted by attila77 View Post
That's a bogus site. They completely ignore sensor technology and basically made up an ideal pixel size.
An engineering conclusion based on beliefs ? Hardly serious.
nope. It's based on physical limitations (wavelength of visible light is ca. 400-700nm)
and the low quality optics of consumer cameras.
The thermal noise has already been reduced so much in previous generations
that the efficiency factor is about 80-90%.
Even if you take into account new technologies (SuperCCD, Exmor-R, Foveon)
there will always be a trade-off between noise and resolution.
 
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#44
Originally Posted by titan View Post
nope. It's based on physical limitations (wavelength of visible light is ca. 400-700nm)
Please, do the math. On the N900 (with a 1/2.5" sensor) it would become an issue at ~f/4 and above. The N900, however, has f/2.8 optics, so it's within that limit even with TWICE the pixels mentioned on the site.

The thermal noise has already been reduced so much in previous generations that the efficiency factor is about 80-90%.
Efficiency factor of what ? You can't do a blanket statement like that, when different sensor technologies result in vastly different signal to noise characteristics, and noise itself is a mixture of at least a dozen different factors, thermal noise is just one of which.

there will always be a trade-off between noise and resolution.
Correct. Except you CAN trade resolution for less noise, but you CANNOT trade low-noise for resolution.
 
Posts: 946 | Thanked: 1,650 times | Joined on Oct 2009 @ Germany
#45
Originally Posted by attila77 View Post
Please, do the math. On the N900 (with a 1/2.5" sensor) it would become an issue at ~f/4 and above. The N900, however, has f/2.8 optics, so it's within that limit even with TWICE the pixels mentioned on the site.
aperture size is independent of optics quality, which also limits the line resolution.
The N900 may have a higher-quality Carl-Zeiss lens and I don't want to defend their
specific MP recommendations based on 2007 technology but there are definitely
physical limitations. Even the Canon marketing dept has realized that and
reduced MP from 14.7 (G10) to 10 (G11) for a 1/1.7" CCD sensor.

Efficiency factor of what ? You can't do a blanket statement like that, when different sensor technologies result in vastly different signal to noise characteristics, and noise itself is a mixture of at least a dozen different factors, thermal noise is just one of which.
It took that number from a current c't magazine article, which compares new sensor technologies. It is the ratio of incoming photons to released electrons.
 
Posts: 3,319 | Thanked: 5,610 times | Joined on Aug 2008 @ Finland
#46
Originally Posted by titan View Post
aperture size is independent of optics quality, which also limits the line resolution.
The N900 may have a higher-quality Carl-Zeiss lens and I don't want to defend their
specific MP recommendations based on 2007 technology but there are definitely
physical limitations.
The fact that you CAN take good pictures in broad daylight suggests that the optics quality is just fine. If it was a lens limitation, the images would be no good regardless of lighting conditions.

Even the Canon marketing dept has realized that and reduced MP from 14.7 (G10) to 10 (G11) for a 1/1.7" CCD sensor.
That is unrelated to lens quality, it's the diffraction limit that is being overrun. With ~15Mpix per 1/1.7", the Canon is already diffraction limited at f/2.6. The lens is 2.8-4.5, so it's quite clear that nominal resolution images could not be recorded even if it had an optically IDEAL lens.

It took that number from a current c't magazine article, which compares new sensor technologies. It is the ratio of incoming photons to released electrons.
That's not thermal (?) noise. That's just one factor of many in the noise equation. What about readout noise ? What about A/D resolution/quality ? What about the bayer array ?

The bottom line is:
A lower megapixel number does not automatically result in higher quality images. Image quality is a result of the imaging system as a whole.

The anti-megapixel crusade is doing the same thing as the megapixel rush did - making it a story of megapixels ONLY. There is far too many factors involved than to be able to distill such a conclusion.
 
Posts: 946 | Thanked: 1,650 times | Joined on Oct 2009 @ Germany
#47
Originally Posted by attila77 View Post
The bottom line is:
A lower megapixel number does not automatically result in higher quality images. Image quality is a result of the imaging system as a whole.
correct. The purpose of that initiative was to educate the public that they should
not focus only on the MP count, which are already sufficient for most purposes,
but on other factors of the camera such as lower-noise, better optics, aperture and focal length range etc.
 
Posts: 59 | Thanked: 10 times | Joined on Sep 2009 @ Portugal
#48
Saturday, thanks to VDVsx, i was able to play a little bit with the N900.

I'm really pleased with the device (even in pre-prod unit) the only thing that scared me (and i must say a lot) were the limited options available on camera mode.

In picture:


you only have 2 modes (Widescreen 3.5 MPx / 4:3 5MPx)

You don't have all the modes available in the 5800, for exemple.


In Video capture mode (the worst)

You can't modify any value except Exposure and white balance!


You can't choose the size of the video, you can't activate the LED (to use at night) you can't record in B&W, sepia...


And another major flaw... the lens cover is IDENTICAL to the n97 one...




I hope this just a question of firmware and i hope that the issue will be solve in the commercial unit... I have some projects (professionals) for the mobile camera and it will be heart breaking not being able to fully use the wonderful N900.


By the way... Is it possible to somehow use an external mic in the N900 to capture sound wile video recording? Since the jack allow a mic (from the headset) isn't it possible to create someting who could use the mic pluged + the video camera?

Last edited by sinetype; 2009-11-16 at 11:23.
 

The Following User Says Thank You to sinetype For This Useful Post:
Helmuth's Avatar
Posts: 1,259 | Thanked: 1,341 times | Joined on Oct 2009 @ Germany
#49
Originally Posted by sinetype View Post
And another major flaw... the lens cover is IDENTICAL to the n97 one...
I'm glade to say it is not exactly the same. There iss a bit space between lense and shutter. So it shouldn't scratch the lense. (okay the other side, but this doesn't matter)

I prefer a simple but working lens cover instead of a missing lens cover. I wouldn't buy a Mobile without lens cover any more...
 

The Following 2 Users Say Thank You to Helmuth For This Useful Post:
Posts: 203 | Thanked: 68 times | Joined on Oct 2009
#50
The Nokia Blog has a review of the N900, which I posted a thread on in the N900 section. But it has this useful paragraph about the N900 camera, which confirms and elaborates on some of what sinetype said above:

The Nokia N900’s camera does not disappoint. It features a 5mp camera labeled with Carl Zeiss lens. It also packs dual LEDs flash, but don’t expect miracles when taking photos in dark restaurants and bars. By default, the N900 takes 3.5mp photos. It crops the top and bottom to give a widescreen aspect picture. You can change this to 5mp in options. It does not include all the camera options from previous Nseries cameras. It’s missing sequence, self-timer, color tone, contrast, and sharpness that are found in the Nokia N97.
See: http://thenokiablog.com/2009/11/13/nokia-n900-review/

So it seems the N900 camera is not really comparable to the N97 as has been assumed (including by myself) in some posts earlier in this thread. It's a little perplexing that features would be left out that have become common place on N series cameras. I can't think of a reason other than that perhaps Nokia really is rushing the N900 out the door.
 
Reply


 
Forum Jump


All times are GMT. The time now is 18:04.