[wll]

Very interesting article.

I did not know about "T" stops, and I did not know each lens light transmissions are different even though they may be set at the same "F" stop !

wll

[MRRiley]

Neither Pentax News nor Pentax Rumor... Moved to Gen Photo!

Mike
PF Moderation Team

[thibs]

I'd say it is fast enough and moreover, it was designed for digital cameras so I would assume that the light rays are directed so that we get the full benefit of the f/1.4 on a digital camera.

It could be interesting to compare the DA* 55/1.4 with a K 55/1.4! I don't have either so I can't do it

Ole, there's no K55/1.4. There are K55/1.8 and K55/2.

[northcoastgreg]

Fascinating article. I was particularly intrigued by the following passage.

By the way, this type of light loss is the main reason that modern lenses designed for digital photography place great emphasis on the direction of light rays behind the lens. It is also an important reason why Medium Format backs (often used with camera movements such as tilt, rise/fall and shift) and the Leica M9 (with lenses that produce mostly oblique light rays) use CCD sensors with a very different structure, rather than CMOS sensors.

I had previously read about the issue of light direction and digital sensors, but I knew nothing about the supposed difference in this respect between CCD and CMOS sensors. Because a number of us shooting Pentax are using older "film" lenses, this raises some interesting questions. Are there any noticeable, "real world" differences between how CCD sensors function with Pentax film lenses and CMOS sensors function? Would someone upgrading from, say, the K200D to the K-5, notice a degredation, however slight, in IQ when using older film lenses with the K-5's CMOS sensor?

Also, on the whole light transmission issue: occasionally I've run across, on the web, complaints that Pentax cameras tend to "underexpose." A common explanation for this is that Pentax doesn't want its users to blow highlights; but could the real explanation be that it's because of light transmission issues via the lens? And if so, wouldn't this confirm that Pentax is not in fact guilty of "gaming the system" by fiddling with ISO?

[Eruditass]

Also, on the whole light transmission issue: occasionally I've run across, on the web, complaints that Pentax cameras tend to "underexpose." A common explanation for this is that Pentax doesn't want its users to blow highlights; but could the real explanation be that it's because of light transmission issues via the lens? And if so, wouldn't this confirm that Pentax is not in fact guilty of "gaming the system" by fiddling with ISO?

While DxO agrees with you (according to their ISO sensitivity tests), there's no reason for Pentax not to select 1/3rd stop higher when metering.

[cateto]

After reading this article in Luminous Landscape, I posted this in my bog:

Are CCD-based cameras the best option for high quality, high aperture lenses?
I found today a very interesting article posted by the well known site Luminous Landscape, entitled "An Open Letter to the Major Camera Manufacturers". It is based on confidential data provided by DxO Labs, and to summarize, they say that, due to light loss occurring with the current CMOS sensors (derived from their structure, which requires light rays coming in a completely perpendicular way for full effect), the camera manufacturers have implemented an automated change in ISO when the lens aperture is very high (which means when there is very high light loss, because such aperture allows many oblique rays to reach the sensor). And this is non openly communicated by the camera manufacturers; in other words, there is an ISO response being manipulated without information to the photographer.

This is certainly interesting (or maybe, even scandalous?), and they suggest that, under such conditions, it might have more sense to invest in lenses which lack such high apertures (which means, in general, cheaper lenses), and use regularly non maximum apertures.

But it has also other implications: in fact, the above described problems are coming from the massive utilization of CMOS sensors; their main advantage is that they have a great sensitivity to low light, but in fact part of this advantage comes, most probably, from this non reported implementation of the automated change in ISO for high apertures.

And the main implication, the one that interests me more, is that those problems do not affect cameras based on CCD sensors, instead of CMOS sensors; the CCD sensor structure does not suffer from light loss, because they are also receiving oblique rays of light (and not only the perpendicular ones, as CMOS sensors), which in turn means that high aperture lenses can be fully exploited by these cameras without any manipulation in the ISO response of the sensor. And this sounds very nice to me, since I have a Pentax K10D, which sensor is of the CCD type. The articule in Luminous Landscape mentions that other cameras, as medium format and also the Leica M9, are also utilizing CCD sensors for this very reason. But there were other popular cameras, as the Nikon D200, based also on the same Sony CCD sensor as the K10D.

So, what was thus far considered as a "problem" of older cameras based on CCD sensors, might become in fact a big advantage when using high quality lenses which allow loads of light to pass across them and reach the sensor: they are getting all that light, regardless of being straight or oblique, and using it for the exposure WITHOUT manipulating the ISO settings of the sensor. These are big news to me, and I'm fairly convinced it could be the same for other users of "obsolete CCD-based cameras".

[sjwaldron]

it might have more sense to invest in lenses which lack such high apertures (which means, in general, cheaper lenses), and use regularly non maximum apertures.

... problems do not affect cameras based on CCD sensors, instead of CMOS sensors; the CCD sensor structure does not suffer from light loss, because they are also receiving oblique rays of light (and not only the perpendicular ones, as CMOS sensors) ... I have a Pentax K10D ...

So, what was thus far considered as a "problem" of older cameras based on CCD sensors, might become in fact a big advantage when using high quality lenses ... These are big news to me, and I'm fairly convinced it could be the same for other users of "obsolete CCD-based cameras".

I came to similar ideas on my blog as well.

However, I wouldn't say with such certainty that CCD sensors would fair better with wide aperture glass not designed for digital compared to CMOS ones.

I also have a K10D (and K-7 which has a CMOS) and would like to test with the FA50mm 1.4 (wish I had a f1.2 lens!), but I don't know how one would tell if Pentax is increasing ISO slightly on the CMOS sensor camera without changing the value in EXIF...

Does anyone have any ideas on how to test or understand the results form each camera? it would seemingly be like comparing apples to oranges.

[ovim]

Does anyone have any ideas on how to test or understand the results form each camera? it would seemingly be like comparing apples to oranges.

If I understood this right the ISO-boosting in canikon happens when the lens tells the body it has a very large aperture. If you use M/K/M42 lens then the body doesn't know what the aperture is. Or stop the f-stop communication by covering the A-contacts from your FA50?

[Digitalis]

If I understood this right the ISO-boosting in canikon happens when the lens tells the body it has a very large aperture. If you use M/K/M42 lens then the body doesn't know what the aperture is

well there is the answer right there, because of the legacy compatibility that Pentax has opted for, it makes it difficult for pentax to bump the ISO figures.

I have a pentax 50mm f/1.2 in both A and K versions and there is no difference between the exposures on each lens, on either the K10D or the K-7.

[sjwaldron]

If I understood this right the ISO-boosting in canikon happens when the lens tells the body it has a very large aperture. If you use M/K/M42 lens then the body doesn't know what the aperture is. Or stop the f-stop communication by covering the A-contacts from your FA50?

Interesting.

I tested my FA 50mm f1.4 and a Chinon 55mm f1.7 on a K10D/K-7 with results on the blog. That took way to long without any conclusions.... It was kind of fun though.

[TeamPentax]

Very interesting, I didn't know that the camera would change the iso to get the right exposure, I had always thought once you set it, it stayed that way.
Thats no good, I'd like to see some replys from the camera companies, I bet they really love DxOmark.com

Any one have a list of Pentax camera's that do this ? I don't intend to shoot with my ist Ds for ever, would like to upgrade at some point, if it could be called an upgrade if I lose certain controls......One to keep an eye on.

[ytterbium]

I think this issue is exaggerated. It is a just an engineering solution to produce correct exposure.
Not some kind of evil conspiracy to cheat the consumers.
Similar to white balance or other post processing and hardware settings to produce desired results.

Maybe it should have been explained by manufacturers. Given the option to turn it off, or just provide "absolute" RAW files (direct dump from ADC).
Actually it would be nice if detailed sensor description and characteristics important to photographers were available.
It is very limited information to type just CMOS/CCD about the heart of camera.
It is know that there are numerous types of technologies (micro lenses - none, centered, offset, coating types, hot filters - absorbing or interference, anti-aliasing filters, back and front illuminated sensors).
Still, most people wouldn't care anyway.

Any one have a list of Pentax camera's that do this ? I don't intend to shoot with my ist Ds for ever, would like to upgrade at some point, if it could be called an upgrade if I lose certain controls......One to keep an eye on.

I expect your older cameras doing this even more than newer ones (worse microlenses etc). Look at sensor design. Which ones are more vulnerable to light falloff.

[Jodokast96]

As said previously, I wouldn't worry about any of the Pentax cameras. If you take a look at the DxO data, every Pentax camera is closer to the stated ISO than any other camera out there, with the exception of the Nikon D3000.

[uccemebug]

The two stated problems are:
1) CMOS sensors have a problem capturing non-perpendicular light such as in out of focus areas of a photo
2) The ultimately obtained image is already manipulated before it's even written to file

The former would seem the easier to test, especially if it's severe enough to cause a visible change in the image. The impact is either apparent or it's not. Using the same fast lens wide open on two bodies (one with a CCD sensor, one with a CMOS) should determine whether there's any visual impact.

The second one is certainly interesting but if we've gotten this far without any out cry, perhaps it just doesn't matter.

[sjwaldron]

As said previously, I wouldn't worry about any of the Pentax cameras. If you take a look at the DxO data, every Pentax camera is closer to the stated ISO than any other camera out there, with the exception of the Nikon D3000.

The two stated problems are:
1) CMOS sensors have a problem capturing non-perpendicular light such as in out of focus areas of a photo
2) The ultimately obtained image is already manipulated before it's even written to file

The former would seem the easier to test, especially if it's severe enough to cause a visible change in the image. The impact is either apparent or it's not. Using the same fast lens wide open on two bodies (one with a CCD sensor, one with a CMOS) should determine whether there's any visual impact.

I could not see any visual impact, but I've never been a pixel peeper (this experiment was fun to try though). The only differences I notice is likely just differences between the sensor designs themselves. Stretching out the K-7's data produces a jagged histogram for example (aka. the K10D's CCD seems to have smoother tones).


Test 1 (K-7, K10D, FA50 f1.4, Chinon 55mm f1.7)

and

Test 2 (K-7, K10D, FA50 with and without data connection) that I just finished a few minutes ago.