[vrolok]

I have been surfing the net on the search of info on the K-3 and I have come across the Techradar review at:

Pentax K-3 review | Digital slrs/hybrids Reviews | TechRadar

I am a wee confused by: "The biggest problem with this camera is the chromatic aberration issue, which was not solved by switching lens or cameras." And "We were so concerned about this that we asked for a second sample of the camera to compare and found the same issue in both models, so we can only assume that this is a general problem with the camera."

Ergh... since when CA is not dependent on the lens exclusively?
Seriously, does anyone have CA issues with the K-3?

Cheers

[JinDesu]

http://cdn0.mos.techradar.futurecdn.net/art/cameras/Pentax/K-3/sample%20images/IMGP0438.JPG

That's one image that they mention seeing the "CA"

[vrolok]

If I have read it well, this is the one and only image with this issue, which btw is clearly visible indeed.

[kh1234567890]

Ergh... since when CA is not dependent on the lens exclusively?

Perhaps they mean purple fringing ?

BTW, their lens is a bit decentered, or is the DA18-135 always as bad as that ?

The EXIF shows that they had both distortion and CA corrections on.

[Na Horuk]

Well thats weird. CA depends on the lens. I guess if the camera has higher resolution, the CA will also be higher resolution, so a 24MP camera might "show more of it" than a 6MP camera of same sensor size. But this is not a "fault".
Next up is the lens. Not sure whether they were using the 18-55mm or 18-135mm lens, but that sample photo looks pretty bad. However, it shows next to no CA! Even in the extreme edges, at 18mm, shooting up at the sky. But what it does show is purple fringing - something that digital cameras always pick up, especially with such high contrast edges like tree branches against the sky. I'm sure any photographer that took a photo like that has seen it, regardless of camera and lens. The only way to avoid it is by watching the angle and exposure (and of course PP). But Purple fringing is not CA. Its a different phenomenon. And I doubt the K-3 is any worse at it than a K-5 or D800.

The only thing that I find odd in that sample photo is that the edges are so poor at f9 and 18mm. But I think the photo also caught handshake blur, because even the in-focus part seems to be fuzzy. The reviewers need to read up on types of CA and other optical phenomenons like fringing.

[rawr]

The level of CA is not bad. A few pixels width here and there. I've seen a lot worse Remember, pixel peeping a 24MP image...

CA is primarily a lens issue. Maybe they are criticising the poor capabilities of the CA correction option in the camera firmware.

The lens does look soft all over the edges. But as they were probably holding the camera above their head pointing upwards, and shooting at only 1/125, it's hard to completely blame the softness of the image on the lens. No doubt there is some motion blur in there too.

[stevebrot]

CA is ENTIRELY a lens issue and can be both detected and measured on the optical bench. I find it strange that a competent reviewer would fault the camera body for having more CA than other cameras tested. The one example photo they included did show very mild LaCA that might not have been visible on a lower resolution sensor. Perhaps that is what the reviewer was referring to?


Steve

---------- Post added 06-12-14 at 10:25 AM ----------

Not sure whether they were using the 18-55mm or 18-135mm lens,

The maker notes in the Exif for the back-lit tree example indicate the DA 18-135 at 18mm. That explains the softness.


Steve

[kh1234567890]

The maker notes in the Exif for the back-lit tree example indicate the DA 18-135 at 18mm. That explains the softness.

The DA18-135 is really as cr*p as that even at f9 ?

I'm glad that I was never tempted to get one.

As for 24Mp camera showing more CA - this is only some 25% more resolution than a K-5, <-----> instead of <---->. Not all that much.

[stevebrot]

As for 24Mp camera showing more CA - this is only some 25% more resolution than a K-5, <-----> instead of <---->. Not all that much.

The % depends on how you do the calculation.* That being said, I don't know that a comparison with the K-3 is that important. I didn't see that the review said what camera the K-3 was worse than. Perhaps the entire camera universe?


Steve

* examples:

relative increase difference, (24-16)/16 * 100 = 50% "K-3 has 50% higher resolution than the K-5"
relative increase, (24/16) * 100 = 150% "K-3 has 150% of the K-5's resolution"
relative decrease, (16/24) * 100 = 67% "K-5 has 67% of the K-3's resolution"
relative decrease difference, (24-16)/24 * 100 = 33% "The K-5 has 33% lower resolution than the K-3"

The % depends on how you do the calculation.* That being said, I don't know that a comparison with the K-3 is that important. I didn't see that the review said what camera the K-3 was worse than. Perhaps the entire camera universe?


Steve

* examples:

relative increase difference, (24-16)/16 * 100 = 50% "K-3 has 50% higher resolution than the K-5"
relative increase, (24/16) * 100 = 150% "K-3 has 150% of the K-5's resolution"
relative decrease, (16/24) * 100 = 67% "K-5 has 67% of the K-3's resolution"
relative decrease difference, (24-16)/24 * 100 = 33% "The K-5 has 33% lower resolution than the K-3"

You are comparing the megapixels which is not the resolution. The resolution is pixels per inch or yard or whatever. So the increase in resolution is approxemately 20 to 22%. 6016x4000 vs 4992x3284 where 6016/4992 = 20.5% and 4000/3284 = 21.8%

[stevebrot]

The resolution is pixels per inch or yard or whatever.

Ok, expressing the pixels as a linear array:

100 * (6016 pixels per row * 4000 rows)/(4992 pixels per row * 3284 rows) =

150% increase or a 50% difference

...or are you saying that only the first row counts? That is a very short photograph.


Steve

[kh1234567890]

Ok, expressing the pixels as a linear array:

100 * (6016 pixels per row * 4000 rows)/(4992 pixels per row * 3284 rows) =

150% increase or a 50% difference

...or are you saying that only the first row counts? That is a very short photograph.

Steve

It is the linear resolution that matters as far as any increased visibility of CA etc. goes. The K-3 sensor is a touch smaller so the pixel spacing is about 3.9um for the K-3 vs 4.8um for the K-5. Less than 25% difference in whichever direction you measure.

Ok, expressing the pixels as a linear array:

100 * (6016 pixels per row * 4000 rows)/(4992 pixels per row * 3284 rows) =

150% increase or a 50% difference

...or are you saying that only the first row counts? That is a very short photograph.


Steve

"Yes". Resolution = picture elements per a length i.e pixels per inch.

Let me put it this way: You have 4 pixels in a square inch, 2x2. Now you double the amount in each direction, i.e 4x4=16 in the same area. The picture goes from 4 pixels to 16 pixels. I now can see twice as much detail in each direction, therefor the resolution has increased 100%, not 300% as per your math.

I had some issues wrapping my head around this to start with but it makes sense.

[kh1234567890]

I now can see twice as much resolution in each direction, therefor the resolution has increased 200%

Actually when something doubles in size or whatever then it is a 100% increase ...

Actually when something doubles in size or whatever then it is a 100% increase ...

Correct. Edit made...