[Scootatheschool1990]

I'm reading up on various sensors and technical aspects that sensors have. Basically how they work and their features they can possibly have. Restrictions of sensors etc... I'm curious if the k3 has micro lenses on its photosites. does the k5? How does someone go about finding this information? I haven't ever seen pentax or another company release this specific information before. Also, how do I find out how large each photosites is on the sensor? I'd like to compare it to cameras like the d800e and 6d. Any information on this specific information would be helpful.

---

Last edited by Scootatheschool1990; 11-11-2013 at 10:14 PM.

[Adam]

I'm reading up on various sensors and technical aspects that sensors have. Basically how they work and their features they can possibly have. Restrictions of sensors etc... I'm curious if the k3 has micro lenses on its photosites. does the k5? How does someone go about finding this information? I haven't ever seen pentax or another company release this specific information before. Also, how do I find out how large each photosites is on the sensor? I'd like to compare it to cameras like the d800e and 6d. Panty information on this specific information would be helpful.

The pixel pitch on the K-3 should be similar to what the D7100 has, if not identical. That's 3.9 micrometers, down from the K-5's 4.8 micrometers.

FWIW based on the testing I've done so far, the K-3's sensor is superior to the K-5's in overall IQ. The results will be posted in our in-depth review on Monday

[7samurai]

Adam, are you going to have recommended settings? They worked really well for me the last time and saved me a ton of trial and error.

[Scootatheschool1990]

The pixel pitch on the K-3 should be similar to what the D7100 has, if not identical. That's 3.9 micrometers, down from the K-5's 4.8 micrometers.

FWIW based on the testing I've done so far, the K-3's sensor is superior to the K-5's in overall IQ. The results will be posted in our in-depth review on Monday

Adam, I'm looking to find all the specs on the sensor. Do you know where there's a chart or information on this information? Where did you find out that the k3 had photosites that large?

Thanks!

[bdery]

I'm curious if the k3 has micro lenses on its photosites

It's a CMOS so yes, 99% chances that there are microlenses. A CMOS has part of its electronic on the pixel, so the actual light gathering area is smaller than with a CCD. The microlens covers the whole pixel, redirecting the light on the photosite itself, improving performances.

[rawr]

how do I find out how large each photosites is on the sensor? I'd like to compare it to cameras like the d800e and 6d

Have you looked at dxomark.com? They would have a lot of the basic specifications, as well as some discussion of photosites etc. And of course they let you compare various camera sensors.

[RGlasel]

Panty information

I think this is the wrong site for that information! bdery is right about the microlenses, any CMOS sensor with decent sensitivity has them, including the K-5 and K-3. Along with smaller microsites (light sensitive regions of metal oxide - the MO in CMOS) the new sensor in the K-3 doesn't have an optical anti-aliasing filter like the K-5 does. Using the camera's built-in shake reduction and/or software instead of glass to eliminate moire patterns should result in a better image, but any improvement will be minor compared to improvements from the increased pixel density. in the end, human visual resolution is much lower than that of the sensor, but our perception of colour, dynamic range and depth is so superior, that we need software (computer and human) to transform the data supplied by the sensor into an image we recognize as an accurate representation of what our brains saw at that particular place and time. More data should result in better images, everything else being equal. But there are limits imposed by our eyes and brains, and the improvement gained by increasing the number of pixels is not linear. Going from 8 MP to 12 MP was more of an improvement than going from 16 MP to 24 MP.

d800e and 6d

What you are really asking is how does the K-3 compare to the most advanced full-frame DLSRs. I would think the optical differences between APS-C and FF cameras would outweigh the differences between the sensors themselves, because electronics can't do anything with light that doesn't come through the lens. I will say that changing the sensor form factor comes with a cost in terms of how the camera is constructed and the demands placed on the camera's image processor, but there are ways to compensate for those factors in order to produce images of similar quality. It appears that the K-3 sensor is state of the art for the APS-C form factor, for now, and any advantages of FF cameras over the K-3 are not dependent on their larger sensors. Considering that the debate over FF vs. APS-C wasn't entirely settled when the K-5 was the APS-C standard, and the K-3 is a significant improvement over the K-5, tells me that your choice should be over which form factor or which model of camera works better for your photography, not over a comparison of sensor specs.

[lister6520]

It's a CMOS so yes, 99% chances that there are microlenses. A CMOS has part of its electronic on the pixel, so the actual light gathering area is smaller than with a CCD. The microlens covers the whole pixel, redirecting the light on the photosite itself, improving performances.

If I'm not mistaken the K-3's sensor is back illuminated so the electronics would be behind the photosensitive part and the photosensitive area can thus fill the whole square.

However even so AFAIK microlenses are still used (even in CCD sensors). I'm not sure why but I think it has to do with ensuring that the light that goes through a particular colour of the bayer filter hits the right photo sensor even if the light is coming in at an angle.

[bdery]

If I'm not mistaken the K-3's sensor is back illuminated so the electronics would be behind the photosensitive part and the photosensitive area can thus fill the whole square.

I haven't read anything to that effect, it's possible but from what I've learned of CMOS the photosite still won't fill the whole pixel.

However even so AFAIK microlenses are still used (even in CCD sensors). I'm not sure why but I think it has to do with ensuring that the light that goes through a particular colour of the bayer filter hits the right photo sensor even if the light is coming in at an angle.

It will help in that case, that is true. Digital sensors are more sensitive to the angle of incidence (this is why telecentric lenses are theoretically better).

[RGlasel]

electronics would be behind the photosensitive part

it's actually layers of different metal oxides that are exposed to X-rays to "burn" the circuitry. The photosensitive part is also part of the electronics. Think of a printed circuit board where all of the "wiring" is sensitive to light, but there is a grid over top that limits each "wire" to being exposed to light in only one spot (a pixel). To eliminate the shadow of the grid, little convex lenses are used to focus light falling on the area including the grid to just the hole in the grid. There is nothing that indicates a radically new design for CMOS sensors was used for the K-3.

[crewl1]

I think this is the wrong site for that information!

Good eye!

[lister6520]

it's actually layers of different metal oxides that are exposed to X-rays to "burn" the circuitry. The photosensitive part is also part of the electronics. Think of a printed circuit board where all of the "wiring" is sensitive to light, but there is a grid over top that limits each "wire" to being exposed to light in only one spot (a pixel). To eliminate the shadow of the grid, little convex lenses are used to focus light falling on the area including the grid to just the hole in the grid. There is nothing that indicates a radically new design for CMOS sensors was used for the K-3.

The issue of the wires on top obscuring part of the sensor is exactly the problem that is solved by the 'back lit' design. It allows the photosensitive area (photodiode) to be the first thing that light hits and all the electronics being behind it. It is quite complicated to manufacture as the chip has to be first etched from one side with the photosites on the substrate and the electronics on top (as with normal CMOS sensor) then they flip it upside down, build another substrate on top of the circuitry and then grind off the original substrate to expose the photodiodes,

The term 'back lit' is actually a bit confusing but in effect it means that the sensor is mounted 'the wrong way round' in the camera, but that is possible because they have removed the opaque part that was the back of the chip (the substrate) and turned that side towards the lens. All else being equal, 'back lighting' will almost double the sensitivity because of the more efficient light collection, equivalent to one stop ISO improvement.

I was quite sure that the K-3 sensor is back lit but Bdery has pointed out that it may not be so. Back lit CMOS was something extraordinary until a year or two ago but has suddenly become relatively common after Sony figured out a way to make them economically. It is certainly not a 'radically new design'. It is for sure used in various other cameras including the Pentax Q-7. Being the very latest Sony sensor, and given that Sony were the ones to pioneer the cost effective back lit CMOS sensor I still think it very likeley that the K-3's sensor is indeed backlit. I would almost be surprised if it weren't.

[Scootatheschool1990]

Have you looked at dxomark.com? They would have a lot of the basic specifications, as well as some discussion of photosites etc. And of course they let you compare various camera sensors.

Oh, very cool! I just checked it out. Found a new favorite website. Thank you!

[Scootatheschool1990]

I think this is the wrong site for that information!

HARDY HAR HAR! ha ha ha

bdery is right about the microlenses, any CMOS sensor with decent sensitivity has them, including the K-5 and K-3. Along with smaller microsites (light sensitive regions of metal oxide - the MO in CMOS) the new sensor in the K-3 doesn't have an optical anti-aliasing filter like the K-5 does. Using the camera's built-in shake reduction and/or software instead of glass to eliminate moire patterns should result in a better image, but any improvement will be minor compared to improvements from the increased pixel density. in the end, human visual resolution is much lower than that of the sensor, but our perception of colour, dynamic range and depth is so superior, that we need software (computer and human) to transform the data supplied by the sensor into an image we recognize as an accurate representation of what our brains saw at that particular place and time. More data should result in better images, everything else being equal. But there are limits imposed by our eyes and brains, and the improvement gained by increasing the number of pixels is not linear. Going from 8 MP to 12 MP was more of an improvement than going from 16 MP to 24 MP.


What you are really asking is how does the K-3 compare to the most advanced full-frame DLSRs. I would think the optical differences between APS-C and FF cameras would outweigh the differences between the sensors themselves, because electronics can't do anything with light that doesn't come through the lens. I will say that changing the sensor form factor comes with a cost in terms of how the camera is constructed and the demands placed on the camera's image processor, but there are ways to compensate for those factors in order to produce images of similar quality. It appears that the K-3 sensor is state of the art for the APS-C form factor, for now, and any advantages of FF cameras over the K-3 are not dependent on their larger sensors. Considering that the debate over FF vs. APS-C wasn't entirely settled when the K-5 was the APS-C standard, and the K-3 is a significant improvement over the K-5, tells me that your choice should be over which form factor or which model of camera works better for your photography, not over a comparison of sensor specs.

[Scootatheschool1990]

telecentric lenses are theoretically better

Telecentric?

it's actually layers of different metal oxides that are exposed to X-rays to "burn" the circuitry.

Interesting!

The issue of the wires on top obscuring part of the sensor is exactly the problem that is solved by the 'back lit' design. It allows the photosensitive area (photodiode) to be the first thing that light hits and all the electronics being behind it. It is quite complicated to manufacture as the chip has to be first etched from one side with the photosites on the substrate and the electronics on top (as with normal CMOS sensor) then they flip it upside down, build another substrate on top of the circuitry and then grind off the original substrate to expose the photodiodes,

This is what I'm talking about! Where did you learn all this information?