[biz-engineer]

There is a physical limit for dynamic rangeat a given iso though.

What parameter limits the dynamic range of a sensel?

[RockvilleBob]

A new advanced auto focus system in continuous mode would be enough for me to replace the K-3.

[jpzk]

A new advanced auto focus system in continuous mode would be enough for me to replace the K-3.

Kudos to that !

[falconeye]

What parameter limits the dynamic range of a sensel?

The ISO norm defines a maximum exposure (limited by fstop and exp.time) onto a unit surface of the sensor, for any given value of iso. Therefore, the number of photons captured is finite. Photon shot noise then defines the maximum dynamic range achievable at that iso.

Dynamic range is a function of spatial frequency (as is noise). Which is why DxO normalized it to the angular resolution of a pixel in an 8MP image. Therefore, if you ask per sensel, its dynamic range can be made larger by covering a larger fraction of the image. But that would not improve the dynamic range at the level of the image.

If you allow to lower iso as much as possibly (until the sensel saturates), then the dynamic range of a sensel is limited by its full well capacity. Which however, can be overcome by continuous digital integration.

Therefore, the DxO landscape score (the DR number) of an ideal sensor is unlimited. And its DxO overall score is infinite. If however, you artificially constrain iso to some number (say 25) and sensor size too (say 24x36mm), then all three DxO scores are finite and the DxO overall score would be finite too. I once wanted to compute it, DxO marks for the ideal ISO25 full frame sensor ...

BTW, that's my biggest problem with DxO mark: while the landscape score of an ideal full frame sensor is infinite (no iso limit), the sports score is finite. Therefore in the future, DxO mark will overemphasize high light performance and underemphasize low light performance. Already today, DxO mark makes the difference between Canon and Sony appear larger than it actually is.

---

Last edited by falconeye; 11-14-2015 at 03:14 AM.

[Nicolas06]

To me we don't care if the DxO score need to be say updated to a new scale 1 day. That easy to do if needed.

This is more the problem in general to translate this scale to your own. And it depend a lot of your practice and objectives. You see to value high iso performance. It is important for some people, not much to other like studio shooters or landscape shooters.

[Rondec]

The ISO norm defines a maximum exposure (limited by fstop and exp.time) onto a unit surface of the sensor, for any given value of iso. Therefore, the number of photons captured is finite. Photon shot noise then defines the maximum dynamic range achievable at that iso.

Dynamic range is a function of spatial frequency (as is noise). Which is why DxO normalized it to the angular resolution of a pixel in an 8MP image. Therefore, if you ask per sensel, its dynamic range can be made larger by covering a larger fraction of the image. But that would not improve the dynamic range at the level of the image.

If you allow to lower iso as much as possibly (until the sensel saturates), then the dynamic range of a sensel is limited by its full well capacity. Which however, can be overcome by continuous digital integration.

Therefore, the DxO landscape score (the DR number) of an ideal sensor is unlimited. And its DxO overall score is infinite. If however, you artificially constrain iso to some number (say 25) and sensor size too (say 24x36mm), then all three DxO scores are finite and the DxO overall score would be finite too. I once wanted to compute it, DxO marks for the ideal ISO25 full frame sensor ...

BTW, that's my biggest problem with DxO mark: while the landscape score of an ideal full frame sensor is infinite (no iso limit), the sports score is finite. Therefore in the future, DxO mark will overemphasize high light performance and underemphasize low light performance. Already today, DxO mark makes the difference between Canon and Sony appear larger than it actually is.

I dislike putting a single score on anything, be it lens or sensor. Better, in my opinion, to publish the break down -- color depth, dynamic range, and sport iso scores and then let people compare themselves. A K3's score of 80 is pretty close to a 6D's score of 82, but when you break it down, the K3 has better DR at low iso, while the Canon has better SNR throughout the range and better DR over iso 400. I personally use low iso a lot more than high iso, so having better iso 1600 probably isn't that important to me, but for many people high iso is the more important side of the equation and having one more EV of dynamic range at iso 100 is immaterial.

I guess on the internet people have to rank everything, even sensors.

[RockvilleBob]

Kudos to that !

Indeed! And enough to switch brands if the K-3 replacement doesn't exceed AF capabilities of Nikon and Canon. Life is too short and there are too many fun things to do.

[biz-engineer]

If you allow to lower iso as much as possibly (until the sensel saturates), then the dynamic range of a sensel is limited by its full well capacity. Which however, can be overcome by continuous digital integration.

The problem is that a better dynamic range achieved at lower ISO is possible but it is a tradeoff, rather than a general performance improvement of sensor tech. I guess the only way to get more DR with the same ISO is to have a higher supply voltage available on the sensor signal conditioning and for A to D conversion. With current state of the art, I'm not sure if APSC improvement would be cheaper than using a larger sensor, and I'm not even considering the benefit of larger sensor regarding the MTF potential. I guess if there would be improvement done with APSC sensor would be with regards to UHD video 4k , 8k, rather than any significant improvement in dynamic range.

[Nicolas06]

I dislike putting a single score on anything, be it lens or sensor. Better, in my opinion, to publish the break down -- color depth, dynamic range, and sport iso scores and then let people compare themselves. A K3's score of 80 is pretty close to a 6D's score of 82, but when you break it down, the K3 has better DR at low iso, while the Canon has better SNR throughout the range and better DR over iso 400. I personally use low iso a lot more than high iso, so having better iso 1600 probably isn't that important to me, but for many people high iso is the more important side of the equation and having one more EV of dynamic range at iso 100 is immaterial.

I guess on the internet people have to rank everything, even sensors.

You can't really blame them they give you a 3 level view: an overall score, a second level with 3 scores and a third level with all the measurements.

Beside I don't think people rank only on the internet. It is more a need for simplification we have everywhere. There so many thing in one life... And you need to choose in a blink out of many factors. Many time you are far from being an expert on the subject so you can't really take better decision with more information.

There lot of value from a source of information that you can trust and work well and that can help you drive your decision when you don't have the time or knowledge to go deeper.

Hey for most people, even the one DxO score of the sensor is too much detail. This is just one aspect of the camera. The viewfinder, AF performance, sensor resolution, the scenes mode, the ergonomics, the size/weight, the lense echosystem. That's too much to get when you begin.

In the end, that's why marketing work so well. People don't have time, and they don't master everything. They need to take shortcut. So you exploit that for your on adventage so they choose your product and not another one.

And that's why people want reviews with if possible a single number at the end like 3 stars or 85%. That's because they want to double check but maybe don't have the time to go into the deep details.

The latest thing to understand is that it doesn't matter in the end to have always the best compromize. you are going to fail. What count is that the thing is able to do it job well enough and that's all. For all the Pentax Forum guy that go into every detail, there 100 peoples out there that just want a camera that take great photos.

[normhead]

The rumours ar 15.5 EV dynamic range for that sensor. That would be over the top crazy.

[Rondec]

You can't really blame them they give you a 3 level view: an overall score, a second level with 3 scores and a third level with all the measurements.

Beside I don't think people rank only on the internet. It is more a need for simplification we have everywhere. There so many thing in one life... And you need to choose in a blink out of many factors. Many time you are far from being an expert on the subject so you can't really take better decision with more information.

There lot of value from a source of information that you can trust and work well and that can help you drive your decision when you don't have the time or knowledge to go deeper.

Hey for most people, even the one DxO score of the sensor is too much detail. This is just one aspect of the camera. The viewfinder, AF performance, sensor resolution, the scenes mode, the ergonomics, the size/weight, the lense echosystem. That's too much to get when you begin.

In the end, that's why marketing work so well. People don't have time, and they don't master everything. They need to take shortcut. So you exploit that for your on adventage so they choose your product and not another one.

And that's why people want reviews with if possible a single number at the end like 3 stars or 85%. That's because they want to double check but maybe don't have the time to go into the deep details.

The latest thing to understand is that it doesn't matter in the end to have always the best compromize. you are going to fail. What count is that the thing is able to do it job well enough and that's all. For all the Pentax Forum guy that go into every detail, there 100 peoples out there that just want a camera that take great photos.

That's fine. People just make too big a deal about small differences in the scores. The D810 and A7r II scored almost the same and it isn't till you get to the break down that you see the real differences between the sensors and could see which one might work better in a given situation. The D7000 and K5 scored the same, but people acted like the fact the K5 had a tiny bit more DR at base iso pushed it over the top, when the difference was probably just sensor/test variation and not really visible in the real world.

I like DXO Mark, I just don't like when people abuse their test results.

[reh321]

Last flagship just been released. It is called K3-II. There no reason to release another one so soon. The next one is not before next summer so the K3-II has 1 year to sell. But the K50 (= same body as K30) is really basic now and K-S2 has been released for quite some time... So they could release an update to one of 2.

They could be based on K-S2, put an improved 24MP sensor inside and a few other features. The most interresting would be a better AF at K3 level, but honestly that would remove the interrest to buy a K3 or K3-II. Outside of special features that really the improved AF that make the most difference with the flagship.

There are several things I don't know:

(1) how many production lines does Pentax have that can build DSLR cameras?

(2) how intensively are these lines working?

(3) how long does it take to switch a line from one product to another

A year ago I thought the Pentax plan was to sell three lines of APS-C cameras, and they were in the middle of switching from
K-500 : K-50 : K-3
to
K-S1 : K-S2 : K-3ii
but even today places like B&H are still selling everything on that list except the K-500. Are they selling the K-50 and K-3 out of inventory? Even if some of the products are being sold out of inventory, having products like that on the market has to limit what they can introduce, because prices of products at one level have to limit prices of products at higher level. It seems to me that they need to "rationalize" their product line before they add even more products to it.

[Nicolas06]

The problem is that a better dynamic range achieved at lower ISO is possible but it is a tradeoff, rather than a general performance improvement of sensor tech. I guess the only way to get more DR with the same ISO is to have a higher supply voltage available on the sensor signal conditioning and for A to D conversion. With current state of the art, I'm not sure if APSC improvement would be cheaper than using a larger sensor, and I'm not even considering the benefit of larger sensor regarding the MTF potential. I guess if there would be improvement done with APSC sensor would be with regards to UHD video 4k , 8k, rather than any significant improvement in dynamic range.

Not really you just need to give counting capability to sensor. When the photosite is full it can simply discharge and increase a counter to the number of time it discharged.

There prototypes doing just that as there as many othe prototypes for other improvements.

But as usual the innovation doesn't comes from FF, MF, APSC or m4/3 but from smartphones tiny sensor because theses guys get 350 millions unit sold a year, not 15 millions. BSI didn't come to FF first you now. Q add for years and smartphones too.

Some are thinking of quantum sensor likely a fancy name but they get more dynamic range and better low light. Again they don't target the DSLR market. Too small, too complex.

Today in fact you can already get more dynanmic range with a tripod and still subject with many shoots. But a sensor with just capacity for lower base iso and an integrated counter is much more conveniant. It allows to take moving subjects in day light not as the sunny rule say with f/16 1/100s iso 100 could say f/4 1/100s, iso 5.

That's could give 4EV more of dynamic range, like a 3HDR shoot but in shoot with no risk for alignment issues, ghosts or whatever.

And that without counting that if you allow say iso 1 many people will be able to take their long exposure shoot in daylight just by playing with apperture and isos.

[normhead]

I like DXO Mark, I just don't like when people abuse their test results.

Exactly, if you do your homework and understand the test results and what they mean, there's a lot to learn over there. But there are a lot of people who make unwarranted assumptions based on not understanding the results. And bottom line, as much as I like to mailgn DxO for not making their results easier to understand... it's not really their fault. Anyone who quote any stat they don't understand is at fault. It's quite common in the scientific community for researchers to publish data that can be quite easily misinterpreted by lay people. DxO are competent researchers. It's up to the reader to get a handle on how they comply their data and understand exactly what a DxO score of 82 means. The raw score itself tells you very little, and a score of 82 can be arrived at with different levels of strength in each category. You still have to research the camera.

[Nicolas06]

There are several things I don't know:

(1) how many production lines does Pentax have that can build DSLR cameras?

(2) how intensively are these lines working?

(3) how long does it take to switch a line from one product to another

A year ago I thought the Pentax plan was to sell three lines of APS-C cameras, and they were in the middle of switching from
K-500 : K-50 : K-3
to
K-S1 : K-S2 : K-3ii
but even today places like B&H are still selling everything on that list except the K-500. Are they selling the K-50 and K-3 out of inventory? Even if some of the products are being sold out of inventory, having products like that on the market has to limit what they can introduce, because prices of products at one level have to limit prices of products at higher level. It seems to me that they need to "rationalize" their product line before they add even more products to it.

I don't think this is something related to production capacity. This is more logistics, marketing and cost.

You can release new model more often with fewer changes. The benefit of that is that is there a new trend/feature whatever you can deliver it soonner to consumers. Let's call it 'time to market'. You can also ask more for the novelty.

But each new model require you engineers to redesign things a bit to work to make things work. You need to advertize it, to debug it and handle returns. You also need to ensure your seller have the model in inventory to be able to sell it and that the old model doesn't compete too much with the new model. That's like K50 vs K30 as basically theses where the same camera, the wise buying would buy the cheaper one. Not good. You'll also want reviewers to have the camera in hands, say it is great, get theses reviews writen and so on. That's lot of work.

On the opposite, the more you wait between 2 models, the more other manufacturers are likely to have release a new model with more features, the more your product look old and the less you'll be able to sell it.

For the past years Pentax/Ricoh had say arround a 3 year release cycle between 2 really different flagship. Like K5 vs K3... end of 2010 vs end of 2013. They do a refresh at mid term. They did that with the K3-II this summer. I call it a revision or a fake release. Sure there a few new feature but that basically the same gear an an excuse to bump prices, make some buzz and prompt users to consider an updgrade instead of waiting.

So the next APSC flagship would be say for fall 2016. Perfect time as the biggest sales are for Christmas.

But there also say mid level/entry level DSLRs. K-S2 is quite good but K50 start to be old and I don't know if KS1 sell that well. So make sence to see a new entry level or mid level or both cameras.

This is not that they can't make a new APSC flagship when they want... This is just that maybe this is not the right time for them.