Relative Pixel Density - don't expect miracles from 40D

[upnorthfar]

This has been very interesting topic. From the very obvious knowledge how 2Mp more is hurting the image quality we have moved to the conclusion that perhaps it should get 30 million more pixels for the best image quality.

Now I just wait... Maybe the next 1Ds will have 70 million pixels and boldly comes down in fps speed? Who needs another fast low light camera. =)

 

[Anastigmat]

If I was not clear I would like to add that I realize Fred Briggs
also think 8MP is unlikely in the new model.

I think that even 10mp might not be enough. The D200 replacement, the high amateur Sony Alpha and the next Pentax may all have the same 12mp 1.5x Sony CCD chip. These cameras are expected to arrive this fall. With three different models having more megapixels than the 40D, sales of the 40D will likely be similar to current 30D sales figures.

 

[Hardnose]

To simplify your discussion as I see it.

I can either A) create a detailed mosaic in my small 8x8 ft bathroom using small tiles

or B) create a cruder mosaic in my small 8x8 ft bathroom using larger tiles.

If I had a larger bathroom I could fit more in, step back and the large tiles would look better but then even more smaller tiles would allow more variety of detail and I wouldn't have to step back as far.

Isn't that the point? we could talk about the quality of the tiles?

I have a mk2n and a 20D, both around 8mp. The mk2n is tonally smoother great with a 24-70 but I use the pixel density of the 20D for my extreme wide angle and long telephoto work where I want those small details to fall on their own pixels and not be averaged/shared on a larger pixel.

I always experienced more noise with small pixels, that's always been their drawback. On a good bright day a sharp 20D image is stunning but when light is low I'll use the 1dn to smooth out the noise/take out the tonal grit.

I'd hope the new 40D would have more pixels but less tonal grit and noise. The fact that it probably will have just the same crappy focus probably negates all sensor progress anyway.

 

[Hardnose]

I get what you're saying.

 

[John Sheehy]

This has been very interesting topic. From the very obvious knowledge
how 2Mp more is hurting the image quality we have moved to the
conclusion that perhaps it should get 30 million more pixels for the
best image quality.

It is not obvious that the extra 2 MP is hurting image quality. The XTi is a Rebel, and as such has a lower budget for research and components. It was Canon's first 10MP APS sensor. Regardless, it has less read noise than the 30D at ISOs 100 and 200.

Now I just wait... Maybe the next 1Ds will have 70 million pixels and
boldly comes down in fps speed? Who needs another fast low light
camera. =)

If readout is parallelized, then the only bottleneck is going to be writing to the card. The camera can use something like linear DNG for RAW output at lower resolutions, or lossy compression can be used as an option to reduce write times and storage consumption for those that don't want the full res. The lower resolutions will be sharp without aliasing, and will sample away demosaicing artifacts.

The biggest naysayers are always the people with the least imagination. "It won't work; case closed".

--
John

 

[digital_ray_of_light]

Certainly the pixels don't care, but they are bound to the sensor.
Sensors are designed to deliver the best possible picture at the
lowest possible price at their size. Pixels are dependent on that. So
this dependence of pixels on the sensor size/design means that you
can't compare pixels outside the context of a sensor, thus the
experiment does not answer the question asked. Pixels don't exist
outside the sensor.

I can't find any sense in what you say, or see the point of your
perspective. You sound like an apologist for industrial lethargy.
Big pixels are a compromise, not an ideal. We can see that by
comparing crops of the same physical size on the sensor. What don't
you understand about this?

When someone resorts to name-calling, it's usually a sign that they run out of arguments And I'm certainly not an apologist for the industry. I've criticized plenty of companies plenty of times (when I thought they deserved it)

Let me summarize this even more: You have decoupled the pixels from the sensor and based on this you made this comparison and arrived at your conclusion.

What I am stating is that you cannot decouple the pixels from the sensor, thus the ensuing comparison while very interesting, it is not valid.

--

Comprehensive 2007 speculation and predictions: http://1001noisycameras.blogspot.com

 

[coldbivy]

Certainly the pixels don't care, but they are bound to the sensor.
Sensors are designed to deliver the best possible picture at the
lowest possible price at their size. Pixels are dependent on that. So
this dependence of pixels on the sensor size/design means that you
can't compare pixels outside the context of a sensor, thus the
experiment does not answer the question asked. Pixels don't exist
outside the sensor.

I can't find any sense in what you say, or see the point of your
perspective. You sound like an apologist for industrial lethargy.
Big pixels are a compromise, not an ideal. We can see that by
comparing crops of the same physical size on the sensor. What don't
you understand about this?

When someone resorts to name-calling, it's usually a sign that they
run out of arguments And I'm certainly not an apologist for the
industry. I've criticized plenty of companies plenty of times (when I
thought they deserved it)

Let me summarize this even more: You have decoupled the pixels from
the sensor and based on this you made this comparison and arrived at
your conclusion.

What I am stating is that you cannot decouple the pixels from the
sensor, thus the ensuing comparison while very interesting, it is not
valid.

--
Comprehensive 2007 speculation and predictions:
http://1001noisycameras.blogspot.com

Then why don't you explain why you can't decouple the "pixels" from the sensor?The pixels are derived from a de-mosaicing algorithm applied to a limited number of photosites. The properties of these photosites seem to be independent of the total size of the sensor. So why would the noise exhibited by an individual pixel differ based on the total size of the sensor?

You made an assertion that appears to be wrong and haven't explained why you made it, instead just repeating the assertion as if that will eventually make it correct.

 

[carlk]

Your comparison is meanless and the conclusion is just plainly wrong.
Do you imply if Canon make 2MP 40D it will have best IQ of them all?

Why is that so many people on this thread seem unable to grasp that
low pixel density is not necessarily the same as low resolution. The
5D has the lowest pixel density of the cameras I listed , yet has the
second highest resolution - due to the large FF sensor.

Resolution is directly tied with total pixel counts. With the same sized sensors low pixel density equals low pixel counts, which means low resolution. If you want to compare sensor of different sizes then you should use pixel counts. More total pixels will give you higher resolution in print. 5D sure give more resolution than 30D only because it has more pixel counts. A 12MP xxD will have more resolution than 10MP xxD period.

I am simply observing that the sensors with the lowset pixel density
seem to have the highest image quality. However, I've never
suggested or even implied that you should achieve this by putting
ridiculously low numbers of pixels on a sensor.

Ridiculously low or just low, lower pixel density on the same sized sensors will result in LOWER resolution. Your original post has confused pixel counts with pixel density. You then were trying to extend the (wrong) logic to conclude the supposedly higher pixel density of 40D will hurt its IQ. You’re not the first one who thinks that way but it’s totally wrong and baseless.

Obviously to get the image quality associated with large pixels,
while retaining an adequate resolution, you need a bigger sensor in
order to fit in enough of these bigger pixels.

Of course Canon could also stop increasing the number of pixels to
unnecessarily high levels. That way improvements in sensor design
could actually translate into significantly improved DR and noise
performance instead of being primarily used up offset the losses due
to higher pixel density.

That is why I compared the 1D3 with 10MP on an APS-H sensor with the
rumoured 40D with 10MP on the smaller APS-C sensor and said that we
should not expect similar DR and high ISO performance from the
smaller sensor, even if it incorporated all the same improvements as
the 1D3.

That’s not what you said. You were comparing 40D to 20/D/30D and said higher pixel density of 40D will hurt its IQ.

Quote : -----“I can't see any way this can happen. I just hope the 40D doesn't end up as a 12MP 1.6 sensor.Sure there can be more agressive noise removal, but this will impact on ultimate IQ, which I believe is what happens with the 400D compared to the 20D/30D.”----

Matter of fact I really want to see Canon put 12MP in 40D if they can pull it off.

Nowhere did I suggest reducing resolution to very low levels would
improve image quality. In a previous post I did say that I thought
that 8MP was adequate resolution for this class of camera, and
forgoing the increase to 10MP could be worthwhile in terms of better
image quality. That is not to say that a 10MP 40D won't manage some
IQ improvements, just that they could be bigger at 8MP.

12MP is better than 10MP which is better than 8MP. The only question is whether you can use the higher resolution. There is little if any downside of higher MP. Cost may be but not IQ as you have implied.

Numbers don't lie but people often do.

I hope that the above comment is not accusing me of lying - because
if it is then I expect an apology.

People do lie. No apology.

Fred

 

[mfurman]

12MP is better than 10MP which is better than 8MP. The only question
is whether you can use the higher resolution. There is little if any
downside of higher MP. Cost may be but not IQ as you have implied.

I am just wondering if you thought why Canon left pixel count at 10 Mp when designing the new (1.28 crop) 1D mkIII. In other words, what has Canon learned in the last few months that would allow them to have 12 Mpixels on a smaller sensor (with low noise of course)?

--
Michael

'People are crazy and times are strange, I'm locked in tight, I'm out of range, I used to care, but things have changed' - Bob Dylan

 

[carlk]

I was just hoping for higher MP 40D but don’t think Canon will offer anything higher than 10MP at this time. Not that they don’t know how to make it just that the cost will be too high. Not just cost of making the sensor. The larger file size will make it much harder to maintain the acceptable frame rate and processing speed. I think this is probably the reason 1d3 did not get anything more than 10MP. Remember it also got a boost of bit size and frame rate and needed two processors to handle that large amount of data already. They have to make compromises somewhere.

12MP is better than 10MP which is better than 8MP. The only question
is whether you can use the higher resolution. There is little if any
downside of higher MP. Cost may be but not IQ as you have implied.

I am just wondering if you thought why Canon left pixel count at 10
Mp when designing the new (1.28 crop) 1D mkIII. In other words, what
has Canon learned in the last few months that would allow them to
have 12 Mpixels on a smaller sensor (with low noise of course)?

--
Michael

'People are crazy and times are strange, I'm locked in tight, I'm
out of range, I used to care, but things have changed' - Bob Dylan

 

[Steve Hoffmann]

I think it is striking how closely this measure seems to corresponds
with the reported relative IQ (Image Quality) from these cameras.
This number also gives a good measure of how high a demand the
camera makes on lens resolution, with even the best lenses possibly
being asked for too much by very high pixel densities.

Whatever...My new XTi at 10mp and its densely packed sensor has as good ISO 1600 performance as my IDs MK II does....go figure...IMHO, IQ has more to do with the sensor design, anti-aliasing filter and firmware than it does with pixel density. constantly improving their engineering has allowed Canon to excel in IQ with all of their DSLR cameras.

see - http://www.sphoto.com/techinfo/dslrsensors/dslrsensors.htm
--
Steve Hoffmann
http://www.sphoto.com
Gallery and Digital Imaging Information

 

[John Sheehy]

Whatever...My new XTi at 10mp and its densely packed sensor has as
good ISO 1600 performance as my IDs MK II does....go figure...IMHO,
IQ has more to do with the sensor design, anti-aliasing filter and
firmware than it does with pixel density. constantly improving their
engineering has allowed Canon to excel in IQ with all of their DSLR
cameras.

The 1Dsmk2 seems to have a low quantum efficiency. The electrical and photon characteristics are probably about 2/3 stop worse than other Canons. The 1Dsmk2 only has ISO amplification at the photosites up to 800, like the 10D did, so 1600 is a push.

--
John

 

[John Sheehy]

Certainly the pixels don't care, but they are bound to the sensor.
Sensors are designed to deliver the best possible picture at the
lowest possible price at their size. Pixels are dependent on that. So
this dependence of pixels on the sensor size/design means that you
can't compare pixels outside the context of a sensor, thus the
experiment does not answer the question asked. Pixels don't exist
outside the sensor.

I can't find any sense in what you say, or see the point of your
perspective. You sound like an apologist for industrial lethargy.
Big pixels are a compromise, not an ideal. We can see that by
comparing crops of the same physical size on the sensor. What don't
you understand about this?

When someone resorts to name-calling, it's usually a sign that they
run out of arguments

To falsely accuse someone of "name calling" could be interpreted that way, as well.

And I'm certainly not an apologist for the
industry. I've criticized plenty of companies plenty of times (when I
thought they deserved it)

Let me summarize this even more: You have decoupled the pixels from
the sensor and based on this you made this comparison and arrived at
your conclusion.

You are assuming some kind of symbiosis that does not exist. Sensels have read noises, shot noises, sensor leakage noises, and filters, and these can be cropped or expanded maintaining the same characteristics (although parallelized readout may be needed to maintain readout speed without increasing read noise).

What I am stating is that you cannot decouple the pixels from the
sensor, thus the ensuing comparison while very interesting, it is not
valid.

I am stating that that doesn't make any practical sense; it sounds like mumbo-jumbo.

--
John

 

[hoppa]

You've convinced me... I'm going to go out and buy the oldest, largest sensor camera I can. I bet they go all the way down to 2 MP/sq. in or less! The IQ on those must be AMAZING!

 

[digital_ray_of_light]

Certainly the pixels don't care, but they are bound to the sensor.
Sensors are designed to deliver the best possible picture at the
lowest possible price at their size. Pixels are dependent on that. So
this dependence of pixels on the sensor size/design means that you
can't compare pixels outside the context of a sensor, thus the
experiment does not answer the question asked. Pixels don't exist
outside the sensor.

I can't find any sense in what you say, or see the point of your
perspective. You sound like an apologist for industrial lethargy.
Big pixels are a compromise, not an ideal. We can see that by
comparing crops of the same physical size on the sensor. What don't
you understand about this?

When someone resorts to name-calling, it's usually a sign that they
run out of arguments And I'm certainly not an apologist for the
industry. I've criticized plenty of companies plenty of times (when I
thought they deserved it)

Let me summarize this even more: You have decoupled the pixels from
the sensor and based on this you made this comparison and arrived at
your conclusion.

What I am stating is that you cannot decouple the pixels from the
sensor, thus the ensuing comparison while very interesting, it is not
valid.

--
Comprehensive 2007 speculation and predictions:
http://1001noisycameras.blogspot.com

Then why don't you explain why you can't decouple the "pixels" from
the sensor?The pixels are derived from a de-mosaicing algorithm
applied to a limited number of photosites. The properties of these
photosites seem to be independent of the total size of the sensor. So
why would the noise exhibited by an individual pixel differ based on
the total size of the sensor?

But the pixels/photosites depend on the size and nature of the sensor. A tiny crop of a large sensor (which is the comparison here) would have never been designed or released as an independent sensor. The size of the sensor affects the design decisions/trade-offs which affect the photosites/pixels. That's what I meant by decoupling the pixels from the sensor..

If pixels were interchageable between existing sensors regardless of size, then I would have no objections, but they are not.

You made an assertion that appears to be wrong and haven't explained
why you made it, instead just repeating the assertion as if that will
eventually make it correct.

I explained it in previous posts, and in the paragraph above (I hope)

--

Comprehensive 2007 speculation and predictions: http://1001noisycameras.blogspot.com

 

[digital_ray_of_light]

Certainly the pixels don't care, but they are bound to the sensor.
Sensors are designed to deliver the best possible picture at the
lowest possible price at their size. Pixels are dependent on that. So
this dependence of pixels on the sensor size/design means that you
can't compare pixels outside the context of a sensor, thus the
experiment does not answer the question asked. Pixels don't exist
outside the sensor.

I can't find any sense in what you say, or see the point of your
perspective. You sound like an apologist for industrial lethargy.
Big pixels are a compromise, not an ideal. We can see that by
comparing crops of the same physical size on the sensor. What don't
you understand about this?

When someone resorts to name-calling, it's usually a sign that they
run out of arguments

To falsely accuse someone of "name calling" could be interpreted that
way, as well.

To be falsely accused of being an "industry apologist" could also be interpreted that way as well

And I'm certainly not an apologist for the
industry. I've criticized plenty of companies plenty of times (when I
thought they deserved it)

Let me summarize this even more: You have decoupled the pixels from
the sensor and based on this you made this comparison and arrived at
your conclusion.

You are assuming some kind of symbiosis that does not exist. Sensels
have read noises, shot noises, sensor leakage noises, and filters,
and these can be cropped or expanded maintaining the same
characteristics (although parallelized readout may be needed to
maintain readout speed without increasing read noise).

Okay, I think we are drilling closer to why I am objecting: My objection is not based on "organic symbiosis" but:

Does the sensor size/purpose not affect the designs and trade-offs of the photosites/pixels? If someone was designing a sensor that was the size of the crop (from the larger sensor) that you are comparing to (instead of APS-C size), can you say that in almost every such instance they would have designed it the same way with the same photosites/pixels as the larger sensor? If the answer to this mumbo-jumbo question is yes, then I don't have any objections

What I am stating is that you cannot decouple the pixels from the
sensor, thus the ensuing comparison while very interesting, it is not
valid.

I am stating that that doesn't make any practical sense; it sounds
like mumbo-jumbo.

I specialize in mumbo-jumbo!

--

Comprehensive 2007 speculation and predictions: http://1001noisycameras.blogspot.com

 

[JimH]

I think his point was that the performance of these small sensors with very dense photosites proves that the technology exists to make larger sensors using the exact same design - just more of it

And since the goal of these camera companies is to win in the marketplace and make the most money for their stockholders, they'll do whatever it takes to achieve that.

Thus, if it is practical to do so, they can certainly just upscale the sensor from a P&S to be APS-C sized and give us a 40 megapixel APS-C sensor that has the same performance at the pixel level as the P&S has and of course, much better performance at the photograph level.

But there are always tradeoffs in any design. So to give us the most bang for our buck, they are using fewer but larger photosites in our DSLRs at this point.

Some people seem to think that we've long ago reached the peak of possible performance from these sensors so the only way to improve IQ is to make larger and larger sensors using relatively large photosites.

But the problem with that is that it's always vastly more expensive to produce huge ICs, so they end up being outrageously expensive to manufacture. And if we want really great performance, then we can't even use a standard DSLR-sized system. Instead, we must move up to even larger formats.

So to satisfy the demand for a reasonably priced and "standard sized" DSLR, we end up with things like the 20D/30D sensor which is amazingly good despite it's relatively high pixel density. Most folks thought that would be impossible to pull off, but here we have it. And now, people are saying that we can't possibly go further. I figure that again, those people will be proven wrong.

I don't think it's ever wise to bet on technology standing still.

--
Jim H.

 

[upnorthfar]

The "slow" 10Mp 400D is very nice compared to the fast but heavy and expensive 8Mp 30D. I didn't think twice which one to get.
Maybe also in future the 40D will be 10Mp and 500D will be 12Mp?

The relation between speed, resolution and price seems quite consistent. Who wants a xxxxD 1fps resolution king? I do at least for my second body.

Or is it the speed that holds back the resolution? Thoughts about the cost of it?
Would a plastic 1fps xxxxD be 16Mp with the price of a 8Mp xxD?

 

[Lawn Lends]

The ability of a camera to resolve details is part of the "IQ" in my book. Of course there are so many other parts of IQ, but you can't talk about the quality of an image produced by a camera without comparing the details resolved.

If you want to make a fair comparison of IQ, you need to apply different pixel densities to the SAME SIZE sensor and assume all increased processing requirements caused by denser sensors can be easily taken care of.

The size of the sensor is merely an economic decision by the company to increase or decrease yield of the wafer (or whatever they call it) and to increase or decrease the cost of other components which must deal with the data produced by the sensor.

There should be no reason why a company can't standardize the size of the sensor and vary the pixel densities for different cameras. Now that would be an interesting comparison.

 

[coldbivy]

Then why don't you explain why you can't decouple the "pixels" from
the sensor?The pixels are derived from a de-mosaicing algorithm
applied to a limited number of photosites. The properties of these
photosites seem to be independent of the total size of the sensor. So
why would the noise exhibited by an individual pixel differ based on
the total size of the sensor?

But the pixels/photosites depend on the size and nature of the
sensor. A tiny crop of a large sensor (which is the comparison here)
would have never been designed or released as an independent sensor.
The size of the sensor affects the design decisions/trade-offs which
affect the photosites/pixels. That's what I meant by decoupling the
pixels from the sensor..

If pixels were interchageable between existing sensors regardless of
size, then I would have no objections, but they are not.

They are. How do you suppose a CMOS photosite differes based on the size of the sensor? The only way that I can think of is that the pre-amplifier attached to the photodiode must drive a larger parasitic capacitance in a larger sensor because the read line is longer and may be sized differently, but may not be depending on the required readout speed. Either way, this could be offset at any practical read rate by increasing the parrallelism of the pixel reads. Anyway, there would be nothing to prevent such a photosite from being placed on a differently sized sensor if you wanted to live with a slightly reduced read rate. I don't see any other couplings of photosite design with sensor size. There are, however, design differences and optimizations to be made based on the size of the photosite itself.

No doubt there are design differences between the photosites used in the two sensors in the comparison, but that was not the author's point and was stated so.

I'm not saying you are entirely wrong- his example is far from perfect- but you are using many unproven, if not unfounded statements in your arguments. such as "But the pixels/photosites depend on the size and nature of the sensor" It isn't clear to this reader why the size matters or what you mean by nature. Do you mean CMOS vs CCD? Do you mean IC process? What do you mean? It feels like you are making holistic type statements as a logical argument.