Pixel Density is GENIUS!

Started Jul 13, 2008 | Discussions
simpy Veteran Member • Posts: 3,090
Re: Pixel Density is GENIUS!

RRJackson wrote:

simpy wrote:

Again, why would pixel density be better than a simple sensor size?

But DP lists both. Are you upset that they're giving you too many
searchable options?

Well, they list the size in archaic x/y" terms. It would be helpful to list it as a surface measure (in cm²) as well. This also helps avoiding confusing linear size vs. area (such as in the MP calculation in the other post).

And, by Phil Askey's own admittance, the pixel density measure is intended to be a 'lower-is-better' selection criterion, and that's simply not the case in general.

Simon

 simpy's gear list:simpy's gear list
Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm F4.0-5.6 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8 Olympus Zuiko Digital ED 50mm 1:2.0 Macro +7 more
simpy Veteran Member • Posts: 3,090
Re: No really, it isn't

RRJackson wrote:

simpy wrote:

I presume you divided by 2.5 to get from 12MP to approx 5MP. However,
the pixel count scales with the surface area, which goes as the
square of the length.

Yeah, didn't occur to me until after I'd posted. Math is not
intuitive for me. Still, it seems like an argument could be made for
bigger sensors, bigger photosites even at the expense of resolution.
Although some digiicams seem to be binning now. I assume that's why I
keep seeing low-resolution/High-ISO modes popping up.

Well, there is a case for a low-res sensor in a compact camera, but that is mostly because of limited processing speed and storage capacity. Yesterday(?), Eric Fossum posted a link to a research paper that showed that for real world sensors from the same manufacturer, smaller pixels do not lead to decreased image quality (at the same final image size).

Simon

 simpy's gear list:simpy's gear list
Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm F4.0-5.6 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8 Olympus Zuiko Digital ED 50mm 1:2.0 Macro +7 more
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: No really, it isn't

Andrew dB wrote:

A compact with a decent sized sensor would be pretty nice. Even a
2/3 chip would be a big improvement but provided the pixel count
isn't so high as to make the camera unattractively slow, we don't
need to worry about pixel density.

I'm kind of surprised at the gradual disappearance of 2/3" chips in compacts. They're pretty much ubiquitous in broadcast video because of all the legacy glass from the saticon tube cameras that I used for years when I was starting out. I always kind of figured 2/3" sensors would end up being in pretty much everything just so the sensor manufacturers could concentrate on fewer production lines. Hasn't been the case, though. Of course, digicam resolutions have leapfrogged ahead of even HD video, so it isn't like they could share parts. It's a good sensor size, though. Small enough that it has plenty of DOF, but big enough that there's still some sense of selective focus if you need it.

RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: No really, it isn't

simpy wrote:

Well, there is a case for a low-res sensor in a compact camera, but
that is mostly because of limited processing speed and storage
capacity. Yesterday(?), Eric Fossum posted a link to a research paper
that showed that for real world sensors from the same manufacturer,
smaller pixels do not lead to decreased image quality (at the same
final image size).

No insult intended to the esteemed Mr. Fossum, but Micron is a sensor manufacturer and that paper read more like a press release for their new line of sensors than a serious comparison of sensor performance.

Obviously smaller and smaller photosites are managing to perform at levels that were unimaginable just a few years ago. However, in "real world" sensors the ones with the smallest photosites almost invariably have the worst "real world" performance. If having the smallest photosites you can pack on a sensor was such a huge draw the D3 would be an in-joke among photographers and they'd all be carrying around E-3 bodies like mine.

simpy Veteran Member • Posts: 3,090
Re: No really, it isn't

RRJackson wrote:

If having the
smallest photosites you can pack on a sensor was such a huge draw the
D3 would be an in-joke among photographers and they'd all be carrying
around E-3 bodies like mine.

The reason that doesn't happen is that it's first about sensor size (main noise determinant) and then about getting as many pixels as possible for maximum resolution. Note also that the D3@9fps is already moving around some 150MB per second [similar to the Canon 1Ds3@5fps].

To be perfectly honest, at any technological level there is a limit to the amount of pixels you can cram on a sensor before you do start to hurt noise performance, but this limit is a moving target and disappears as the sensors approach their theoretical maximum performance. Therefore, pixel density is not a useful measure on its own. Sensor size, pixel count and year of introduction are a much better indicator of performance.

Simon

 simpy's gear list:simpy's gear list
Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm F4.0-5.6 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8 Olympus Zuiko Digital ED 50mm 1:2.0 Macro +7 more
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: No really, it isn't

simpy wrote:

The reason that doesn't happen is that it's first about sensor size
(main noise determinant)

Well, only as it relates to how many photosites you've crammed on the sensor. Using equivalent sensor technology, a 2/3" VGA sensor could undoubtedly outperform a 24mm x 36mm 50 megapixel sensor. Unless you decide to start binning or using some other form of exotic mojo on the uber-sensor the chip running at 640 x 480 is going to have some pretty long legs. If there was only some metric for comparison...something like pixel density would be pretty handy for ballpark estimations.

simpy Veteran Member • Posts: 3,090
binning isn't 'exotic mojo'

RRJackson wrote:

simpy wrote:

The reason that doesn't happen is that it's first about sensor size
(main noise determinant)

Well, only as it relates to how many photosites you've crammed on the
sensor. Using equivalent sensor technology, a 2/3" VGA sensor could
undoubtedly outperform a 24mm x 36mm 50 megapixel sensor. Unless you
decide to start binning or using some other form of exotic mojo on
the uber-sensor the chip running at 640 x 480 is going to have some
pretty long legs. If there was only some metric for
comparison...something like pixel density would be pretty handy for
ballpark estimations.

When you compare sensors at an equal output size, you're obviously binning pixels together (if the resolution of both chips is high enough). However, that's hardly 'exotic mojo'. What do you think your raw converter of choice does when you produce an 800x600 image?

 simpy's gear list:simpy's gear list
Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm F4.0-5.6 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8 Olympus Zuiko Digital ED 50mm 1:2.0 Macro +7 more
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: binning isn't 'exotic mojo'

simpy wrote:

When you compare sensors at an equal output size, you're obviously
binning pixels together (if the resolution of both chips is high
enough). However, that's hardly 'exotic mojo'. What do you think your
raw converter of choice does when you produce an 800x600 image?

So then you'd be changing its relative pixel density.

simpy Veteran Member • Posts: 3,090
almost there...

RRJackson wrote:

simpy wrote:

When you compare sensors at an equal output size, you're obviously
binning pixels together (if the resolution of both chips is high
enough). However, that's hardly 'exotic mojo'. What do you think your
raw converter of choice does when you produce an 800x600 image?

So then you'd be changing its relative pixel density.

If you like... We're getting somewhere here. If you agree that this is allowed, what use is it to know the actual pixel density? If you know the pixel count you have the maximum resolution (a hard limit) and if you know the sensor size you have a handle on the noise performance for a given output size.

Simon

 simpy's gear list:simpy's gear list
Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm F4.0-5.6 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8 Olympus Zuiko Digital ED 50mm 1:2.0 Macro +7 more
BJN
BJN Veteran Member • Posts: 5,087
No basis in science?

Just how do you justify that statement?

Indeed, this is a rule-of-thumb metric since there are handful of different sensor technologies that don't have the same inherent photosite area to noise ratio, but in all current sensor technologies the greater the photosite area, the greater the number of photons per photosite vs. the inherent sensor noise. If there were sensors on the market that had obvious performance advantages over others at the same photosite size, the rule-of-thumb wouldn't be useful. That's not today's reality.

And even if there were substantial differences in sensor noise performance at given photosite sizes, the factor would still be useful so that you could compare cameras within a given sensor size class. Sensors with a clear advantage in their density class would be sought out by buyers - a necessary feedback from the marketplace to tell manufacturers that sheer megapixels aren't the only factor for market success.

-- hide signature --

BJ Nicholls
SLC, UT

 BJN's gear list:BJN's gear list
Panasonic Lumix DMC-GM5
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: almost there...

simpy wrote:

If you like... We're getting somewhere here. If you agree that this
is allowed, what use is it to know the actual pixel density? If you
know the pixel count you have the maximum resolution (a hard limit)
and if you know the sensor size you have a handle on the noise
performance for a given output size.

Simon

No use at all in a theoretical world where all sensors tens of millions of teeny-tiny pixels that respond any level of binning on-demand. In terms of what's actually available for purchase right now, it's pretty handy data. Even though there do seem to be a couple of compacts now that have reduced-resolution binning modes for high ISO use.

Andrew dB Contributing Member • Posts: 970
Re: No basis in science?

BJN wrote:

Just how do you justify that statement?

It's justified because a claim has been made by the staff at dpreview that bigger pixels are better REGARDLESS OF SENSOR SIZE without the proper analysis to back it up. The evidence presented is weak to say the least and relies on tests that don't compare like with like.

Indeed, this is a rule-of-thumb metric since there are handful of
different sensor technologies that don't have the same inherent
photosite area to noise ratio, but in all current sensor technologies
the greater the photosite area, the greater the number of photons per
photosite vs. the inherent sensor noise. If there were sensors on the
market that had obvious performance advantages over others at the
same photosite size, the rule-of-thumb wouldn't be useful. That's not
today's reality.

Detailed analysis by several posters has shown that smaller pixels have the same or better efficiency at collecting light when compared with larger pixels and frequently have much lower read noise. The end result is that those small pixels are not disadvantaged in the way that would be expected were noise levels to be invariant with pixel size.

And even if there were substantial differences in sensor noise
performance at given photosite sizes, the factor would still be
useful so that you could compare cameras within a given sensor size
class. Sensors with a clear advantage in their density class would be
sought out by buyers - a necessary feedback from the marketplace to
tell manufacturers that sheer megapixels aren't the only factor for
market success.

But again, none of the proper analysis has been performed by dpreview to show that smaller pixels worsen image level noise and therefore back up their claims.

If the advice was merely that bigger pixels tend to be found on bigger sensors and bigger sensors produce better images (all things being equal) then no-one would complain but it still renders the information pretty useless when you can just look up sensor size by itself on a specification sheet.

Roland Karlsson Forum Pro • Posts: 28,233
Re: The exaggeration of the year?

This must be the greatest exaggeration of the year

It is absolutely not genius.

Actually - it is totally redundant.

It is of no use if we talk about sensors of different sizes.

Only if we keep the sensor size constant it starts to be meaningful - but - then we can use megapixels instead. So - it is redundant.

But thats not the most important reason why it sux. The most important reason is that DPReview has invented it for political reasons - to try to affect the market at stopping the megapixel race. Because they believe that the megapixel race creates cameras with worse image quality. But - they are wrong - it dont. There is no evidence that imagers with more pixels are worse performers - except maybe read speed. Few pixels is a good thing in fast cameras.

-- 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
simpy Veteran Member • Posts: 3,090
Re: almost there...

RRJackson wrote:

No use at all in a theoretical world where all sensors tens of
millions of teeny-tiny pixels that respond any level of binning
on-demand. In terms of what's actually available for purchase right
now, it's pretty handy data. Even though there do seem to be a couple
of compacts now that have reduced-resolution binning modes for high
ISO use.

While the limit is theoretical, the trend is certainly not. Pixel density will become ever less relevant over time. And to be honest, the evidence that pixel density by itself (so not sensor size) is negatively correlated with image quality (at fixed output size) right now is extremely thinly spread.

The political move by DPreview to make this a dominant parameter has, in my opinion, been a bad one, and certainly not a stroke of 'genius'.

I'll leave it at that, since we don't seem to be getting any further.

Simon

 simpy's gear list:simpy's gear list
Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm F4.0-5.6 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8 Olympus Zuiko Digital ED 50mm 1:2.0 Macro +7 more
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: The exaggeration of the year?

Roland Karlsson wrote:

But thats not the most important reason why it sux. The most
important reason is that DPReview has invented it for political
reasons - to try to affect the market at stopping the megapixel race.
Because they believe that the megapixel race creates cameras with
worse image quality.

I can't imagine how they could have arrived at that conclusion.

ejmartin Veteran Member • Posts: 6,274
Re: The exaggeration of the year?

RRJackson wrote:

Roland Karlsson wrote:

But thats not the most important reason why it sux. The most
important reason is that DPReview has invented it for political
reasons - to try to affect the market at stopping the megapixel race.
Because they believe that the megapixel race creates cameras with
worse image quality.

I can't imagine how they could have arrived at that conclusion.

Well, Phil isn't saying, but I'm guessing their reasoning is as follows:

http://forums.dpreview.com/forums/read.asp?forum=1000&message=28600443

-- hide signature --
RRJackson
OP RRJackson Senior Member • Posts: 2,555
Re: The exaggeration of the year?

ejmartin wrote:

Well, Phil isn't saying, but I'm guessing their reasoning is as follows:

http://forums.dpreview.com/forums/read.asp?forum=1000&message=28600443

Ah, the fallacy that just because smaller photosites generally capture noisier images the images actually contain more noise. Gotcha. We'd hate to see that kind of slanderous information propagated. Better to muse about what might be possible with scaling in Photoshop had there been enough light for the small photosites to actually capture an image.

Roland Karlsson Forum Pro • Posts: 28,233
Re: The exaggeration of the year?

RRJackson wrote:

Ah, the fallacy that just because smaller photosites generally
capture noisier images the images actually contain more noise.

Nope, the fallacy that just because smaller photosites generally
are noisier the images actually also are noisier.

-- 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: The exaggeration of the year?

Roland Karlsson wrote:

RRJackson wrote:

Ah, the fallacy that just because smaller photosites generally
capture noisier images the images actually contain more noise.

Nope, the fallacy that just because smaller photosites generally
are noisier the images actually also are noisier.

Which of course they are unless you scale the noise and the image down to compensate. Which is just peachy for times when you're shooting at 800 and you really want to share the image, but it's too noisy at native resolution. It's a completely useless tactic for the shot you might have been able to get at 6400, but all the teensy sensor saw was this:

simpy Veteran Member • Posts: 3,090
This is *not* about different sensor sizes

RRJackson wrote:

It's a completely useless tactic for the
shot you might have been able to get at 6400, but all the teensy
sensor saw was this:

Once more, this isn't about the 'teensy sensor'. It's about two equally 'teensy' sensors with differing pixel counts.

Simon

 simpy's gear list:simpy's gear list
Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm F4.0-5.6 Olympus M.Zuiko Digital 45mm F1.8 Olympus M.Zuiko Digital 17mm F1.8 Olympus Zuiko Digital ED 50mm 1:2.0 Macro +7 more
Keyboard shortcuts:
FForum MMy threads