Pixel Density is GENIUS!

Started Jul 13, 2008 | Discussions
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: f30 - f50

Roland Karlsson wrote:

Yepp - it is.

The f30 is a kind of cult camera. You have to pay premium prices to
get one.

Things that work tend to end up like that.

But - the f50 is VERY close - and it has higher resolution.
I would say f50 gives more value for the money. And it is 25 MP/cm2.
And my guess is that this is much more than this site recommends.

Well, it's kind of a strange proposition. Fuji essentially said to DPR that they had a camera that they knew was noisier above ISO 400 than its predecessor, but if you take the images from that consumer-level camera and scale them to 6 megapixels in Photoshop the output will look almost as good as the discontinued model, essentially making up for the missing stop of sensitivity. Of course, the f50 is only rated up to 1600 where the f30 went to 3200, so there's an exposure gap there to make up for in software if you're shooting in very low light. Not that I think the output of either camera at peak ISO ratings is worth bothering with.

That line of reasoning might make more sense if you were selling someone a DSLR with doubled resolution. A lot of people might buy a 26 megapixel 5D with the understanding that the output would need to be scaled to 13 megapixels if you do any shooting in low light. Fuji should have probably included "Low Light" modes that automatically do the scaling, since the intended market isn't likely to have very sophisticated photo editing software.

It's just a very odd situation. If you read these forums very much there are a lot of comments about how higher megapixel cameras allow people to crop their shots. There's this sort of mindset that resolution allows more choices. I've seen the same thing on forums talking about 4K video where the users are talking about shooting an interview and just cropping cutaways out of the larger frame. Which is obviously a terrible idea on a lot of levels, but if you give someone a noisy 12 megapixel digital camera is it going to end up being used to crop really noisy portraits out of group shots? A very strange tack on the concept of consumer-level electronics, IMO.

Roland Karlsson Forum Pro • Posts: 28,230
Re: f30 - f50

RRJackson wrote:

...

You can have different views I suppose.

My view is that the f50 is virtually as good camera as the f30 - with regard to noise. And then you also have a potential for more resolution. So - too me it sounds like f50 is a better camera - and it has twice the pixel density.

But this very site has a campaign against high pixel density. So - following the advice here - it is much better to get the f30 - and the manufacturers should really take a step back and decrease the number of pixels.

This opinion sounds like pure faith and superstition to me.

There might well be an upper limit on pixel density with current technology. But the f30/f50 example hints at the limit being up there somewhere - above 25 MP/cm2.

-- hide signature --

Roland

 Roland Karlsson's gear list:Roland Karlsson's gear list
Sigma DP3 Merrill Sigma dp2 Quattro Sony RX100 III Pentax K-3 Pentax K-1 +14 more
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: f30 - f50

Roland Karlsson wrote:

My view is that the f50 is virtually as good camera as the f30 - with
regard to noise.

I don't own either one, so I can only look at the testing Phil's done.

But this very site has a campaign against high pixel density. So -
following the advice here - it is much better to get the f30 - and
the manufacturers should really take a step back and decrease the
number of pixels.

If they can make a 20 megapixel compact that shoots acceptably well at ISO 800 then build it. I don't have anything against high-resolution sensors. I just don't care for noisy cameras.

Of course, they could make these cameras a lot more useful if the lenses weren't all so slow. Even the F31 has an f/2.8-5 zoom. Is that really the best they can do? I understand they're having to design around cost and size constraints, but how about an f/2 prime in just one model? Olympus was able to put a 40mm f/1.7 lens in a compact camera 30 years ago, but nobody seems to be able to put a fast lens in a modern compact. A stop or so less sensitivity wouldn't even be worth mentioning if the lens was a stop faster.

Roland Karlsson Forum Pro • Posts: 28,230
Re: f30 - f50

RRJackson wrote:

If they can make a 20 megapixel compact that shoots acceptably well
at ISO 800 then build it. I don't have anything against
high-resolution sensors. I just don't care for noisy cameras.

You have said so several times - and I agree.

And the answer is - bigger sensors.

Of course, they could make these cameras a lot more useful if the
lenses weren't all so slow. Even the F31 has an f/2.8-5 zoom. Is that
really the best they can do? I understand they're having to design
around cost and size constraints, but how about an f/2 prime in just
one model? Olympus was able to put a 40mm f/1.7 lens in a compact
camera 30 years ago, but nobody seems to be able to put a fast lens
in a modern compact. A stop or so less sensitivity wouldn't even be
worth mentioning if the lens was a stop faster.

I have f2.0-2.5 on my Canon G2.

Now - as I understand - the f30 lens is very good.

-- hide signature --

Roland

 Roland Karlsson's gear list:Roland Karlsson's gear list
Sigma DP3 Merrill Sigma dp2 Quattro Sony RX100 III Pentax K-3 Pentax K-1 +14 more
John Sheehy Forum Pro • Posts: 21,861
Re: No really, it isn't

RRJackson wrote:

Roland Karlsson wrote:

Sure - the optics is a part of the equation. But - it does not add
noise - just unsharpness.

Sure, but part of his presentation concerned the overall image
quality and he talked a bit in his post about the sharpness of the
image.

Exactly how sharp do you think anti-aliased pixels from the sharpest lens could be at 289%?

Part of the problem with demonstrations such as mine is that I am assuming a base level of knowledge of what the givens are in RAW imaging, a base knowledge that the vast majority of readers lack. RAW data is VERY simple, but it is different, and you need to get used to the differences.

Of course, the noise is another issue that really needs to be
tested using color charts as well as some real three-dimensional
targets under several temperatures of lighting at a variety of
contrast levels to see what's really happening and when.

Again, the RAW data is very simple. Barring unequal noise reduction, there is no way the ratio of noise between these two images can change with any homogeneous change of subject, white balance, contrast, etc, as they are both completely in the read-noise dominated tonal ranges.

I could do a color checker, but color checkers are suspect, because they contain no detail, so noise reduction is easy to hide, as well as the effects of sharpening or blurring, especially with the existence of edge-preserving algorithms. That's why I prefer doing scenes with some detail when comparing two cameras or pixel density's performance. If you sharpen or blur, it's easier to see, and the detail vs noise varies somewhat together.

These two cameras clearly have almost exactly the same color response, so barring IR or UV influence, change of light source is only going to change the two images in tandem.

The idea of testing

-- hide signature --

John

John Sheehy Forum Pro • Posts: 21,861
Re: skewed comparison

RRJackson wrote:

Well, if the choice was between a camera with output like the 1D Mark
II N and a camera that delivered a hundred megapixel image that
looked like the output from a compact, I'd take the 1D.

You've learned nothing in all this discourse, apparently. It is a given that a 100% pixel view of an APS-H sensor full of current P&S-sized pixels would look noisier than a 1D2. That does not mean "noiser" as far as the camera or image is concerned, though. I know for a fact that many P&S cameras have less noise per unit of sensor area than the 1D2. The 1D2 does what it does because of the sensor SIZE . The low pixel density is of no value; even at the optimized high ISOs, the 450D has about 1/2 stop less read noise per unit of area, and all current Canons have higher QE than the 1D2.

-- hide signature --

John

RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: f30 - f50

Roland Karlsson wrote:

And the answer is - bigger sensors.

Well, that would certainly look better on a table comparing pixel density.

I have f2.0-2.5 on my Canon G2.

Ooh, I wasn't aware of that lens. Why can't Canon still have optics that fast? The G9 has an f/2.8-4.8 lens.

RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: skewed comparison

John Sheehy wrote:

You've learned nothing in all this discourse, apparently. It is a
given that a 100% pixel view of an APS-H sensor full of current
P&S-sized pixels would look noisier than a 1D2.

Just as a heads-up, there are always going to be a lot of us ignoramuses who notice per-pixel noise and not the relativistic potential of comparable systems. While this may be a huge mistake, it's also a reality. Camera companies should release at least one model in their lineups with low per-pixel noise (that isn't simply a product of godawful NR) to quiet us down.

nickleback Forum Pro • Posts: 11,111
Re: f30 - f50

RRJackson wrote:

Roland Karlsson wrote:

And the answer is - bigger sensors.

Well, that would certainly look better on a table comparing pixel
density.

Unless the number of MP is not the same

I have f2.0-2.5 on my Canon G2.

Ooh, I wasn't aware of that lens. Why can't Canon still have optics
that fast? The G9 has an f/2.8-4.8 lens.

I'm sure the fact that the G2 lens is 7-21mm and the G9 lens is 7.4-44.4mm has nothing to do with it.

(hint: 21/2.5 = ???, 44.4/4.8 = ???)

-- hide signature --

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

Roland Karlsson Forum Pro • Posts: 28,230
Re: skewed comparison

RRJackson wrote:

Just as a heads-up, there are always going to be a lot of us
ignoramuses who notice per-pixel noise and not the relativistic
potential of comparable systems. While this may be a huge mistake,
it's also a reality. Camera companies should release at least one
model in their lineups with low per-pixel noise (that isn't simply a
product of godawful NR) to quiet us down.

hehe - I did not see THAT argument coming

Yeah - a 24 MP 2x2 binning machine that gives you 6 MP with nice pixel noise - and an ADVANCED mode that can be turned on in SECRET settings. Unfortunately the binning have to be made AFTER the Bayer CFA reconstruction. So ... the binning is really made in software ... but thats also a SECRET.

I dont know - is this really a good idea?

-- hide signature --

Roland

 Roland Karlsson's gear list:Roland Karlsson's gear list
Sigma DP3 Merrill Sigma dp2 Quattro Sony RX100 III Pentax K-3 Pentax K-1 +14 more
Keyboard shortcuts:
FForum MMy threads