[Nicolas06]

Another example of some portrait (full image & 100% crop)

K3, FA77, ISO100, f/4.5, 1/800s, handled of course

---

Last edited by Nicolas06; 12-21-2014 at 03:25 PM.

[emalvick]

My own thought would be the ultimate comparison would be to compare images framed the same, with the same lens, same exposure settings, and no cropping (ideally with a tripod)... BUT, the comparison would be making the same size print from each image.

Digital comparisons of the images from each camera would be compromised because of cropping or resizing. On print, you would get an indication of how much of a difference there really is (or how perceptible it is). Of course there is still a resize component depending on the printer, but I think comparisons that rely on the resolution or dpi of a monitor are not the best way to make comparisons. I'd suspect as you print larger, the difference would become more noticeable without taking into account viewing distance (i.e. if you try to see the difference). Of course other things like focus and noise factor in as well. If someone would give me a K5ii and a K3, I'll do the comparison myself

[Nicolas06]

My own thought would be the ultimate comparison would be to compare images framed the same, with the same lens, same exposure settings, and no cropping (ideally with a tripod)... BUT, the comparison would be making the same size print from each image.

Digital comparisons of the images from each camera would be compromised because of cropping or resizing. On print, you would get an indication of how much of a difference there really is (or how perceptible it is). Of course there is still a resize component depending on the printer, but I think comparisons that rely on the resolution or dpi of a monitor are not the best way to make comparisons. I'd suspect as you print larger, the difference would become more noticeable without taking into account viewing distance (i.e. if you try to see the difference). Of course other things like focus and noise factor in as well. If someone would give me a K5ii and a K3, I'll do the comparison myself

If you try with a print and you do not crop, you'd need at least 20"x30" shoot so it start to be visible ("only" arround 150dpi for the K5 shoot). I would say it would really show at 40x30 or 60x40, and that's if you look closely. Honestly, I would not sugest a K3 for that if you do just a few print of this size.

What the digit crops show from the K3 is that one is completely able to take perfectly sharp handled shoot with a K3 without any visible shake. Some were doubtfull of this. They also show that even if you take very high isos shoot, the K3 keep the additionnal sharpness. This is not just something that you have at isos 100.

If one like to reframe his shoots or want to get the most out of his tele lenses, his cropping habits are gona severly reduced the number of pixels remaining and the blur introduced by the low pass filter. This is where the K3 additional sharpness would be of help.

I stll think the better AF improvements are a more important feature (and also greatly contribute to perfectly focussed images with perfectly sharp subjects) but as it was the main question of this thread, yes I think you gain from having more than 16MP. Not much for printing directly as this would show only in huge prints, but more if you tend to reframe/crop your shoots. The question for everybody there is to know if he is really doing that. I mean, everybody have different habits and way of doing things, and there nothing bad with that. If you don't crop and don't do huge print on a regular basis, you just don't need more than 16MP anyway, I mean, you might be safe with 10MP already !

Me I tend to do the contrary, I tend to reframe, and I tend to change ratios even something to get some 1:3. I mean this is not because the sensor was design with 3:2 in mind that this is the only valid creative format to use. So this come in very handy !

[Rondec]

Words, words and more words ....

I took some photos today. Not sure how much photos on a dreary day resized to 1000 pixels are, but here they are:

K3



K5 II



K3



K5 II



The first set was shot with the FA 31 limited, the second with the DA *16-50. I apologize for any inconsistencies in the settings, etc that make this comparison useless. I hate doing these sorts of things because people tend to pounce on any differences, but its the best I can do.

I will say as well that Flickr tends to do its own compression and that messes with the photos even more...

---

Last edited by Rondec; 10-07-2014 at 04:37 PM.

[Nicolas06]

The first set was shot with the FA 31 limited, the second with the DA *16-50. I apologize for any inconsistencies in the settings, etc that make this comparison useless. I hate doing these sorts of things because people tend to pounce on any differences, but its the best I can do.

I will say as well that Flickr tends to do its own compression and that messes with the photos even more...

You didn't include crops here and your flickrs settings limit to see your pictures at no more than 2000x1325. That 2.6MP and it will not be possible to see the difference you get from having a 24MP sensor !

The biggest difference that remains are the lighting conditions and color rendering.

[Rondec]

You didn't include crops here and your flickrs settings limit to see your pictures at no more than 2000x1325. That 2.6MP and it will not be possible to see the difference you get from having a 24MP sensor !

The biggest difference that remains are the lighting conditions and color rendering.

I went ahead and uploaded full resolution jpegs of these files, if that is helpful.

[kh1234567890]

I went ahead and uploaded full resolution jpegs of these files, if that is helpful.

Brilliant ! That's what I was hoping to see.

Comparing these in FastStone at 80:100% or 100:125% shows that there is really very little in it. The K-3 has a slight advantage but you have to look for it. It is really down to lens/mount/sensor alignment and focus tolerances level.

Many thanks for that.

[stevebrot]

Brilliant ! That's what I was hoping to see.

Comparing these in FastStone at 80:100% or 100:125% shows that there is really very little in it. The K-3 has a slight advantage but you have to look for it. It is really down to lens/mount/sensor alignment and focus tolerances level.

Many thanks for that.

I don't have FastStone and have no idea what those numbers mean. I took the originals for the two barn photos and did a simple upsample in PSP X6 of the K-5 II image and did a quick comparison. The first thing I noticed is that I could not initially find the plane of focus in either image. The second thing I noticed is that the photo is not particularly sharp. I found the apparent plane of focus in the foreground. Here is the side-by-side of the full-resolution crop at or near the plane of focus:




There are tons of potential factors influencing the comparison. To be honest, the difference is more striking than I expected.


Steve

[Nicolas06]

None of theses pictures are sharp to be honest. The K5-II is sharper because at least a small part of the foreground is in focus. For the second image the exposure bias is -3.3EV. The sharpness setting in PP is set to "hard" whatever it might mean.

With all due respect, I think there a problem with thoses photos and without even counting the K3, a FA31 or DA16-50 should yield better results on a K5 than that:
- First the focussing should be put on the subject. I would have naively used the center AF point. It if was the case I would look if I happen to have frontfocus problems.
- Second for a landscape like this, I would have used more f/8 to get the max deph of field I could.
- Third I think photos on a cloudy day do not help to show sharpness.
- Fourth, I'am not sure I understand the -3.3EV exposure bias on the K3 soy bean photo shoot.
- Five I do not get how stevebrot manage to get much darker crops in his comparisons than the image are looking in the flickr.

---------- Post added 10-08-14 at 09:32 AM ----------

It also funny for me to analyse human psychology: kh want to have a point that you can't get sharpn enough image handled to really benefit of the K3 sensor resolution or even K5 so that it is plain useless to have more pixels. When he see some K3 shoots are quite sharp, there nothing said, no comment. But when there 2 shoots, both K5 & K3, both not very sharp, must softer than the K3 shoots I provided, then he confirm that was exactly what he wanted and that it show his point. One could say, yes it is not the same scene. Agree but the difference is very visible anyway !

Well from kh own arguments, there many things that can fail in a shoot and make it soft. I agree. I do think that many can be solved even when shooting handled.

One just need 1 shoot to proove that it is possible to get sharp shoots and it has been done. A few blurred shoots mean nothing. It could be the argument if that sharp look is plain luck and soft look is the normal output but then I be very, very lucky. I'am not against it. But this look unlikely to me.

As for Rondec, I ask for you opinion on yours shoots! Are there typically the sharpness you get of your shoots? You don't get ofen sharper shoots? If not, and if you are interrested, I think you have possibility to improve a lot there. In all due respect.

[p38arover]

Well I would like one. But for my average work?

How about you?

I'm afraid your "average" work is far, far superior to my best work!

So a camera better than my K-5 wouldn't help me. Having said that, I'd have a K-3.

A 645Z might be a bit bulky for your sports photography.

[kh1234567890]

I don't have FastStone and have no idea what those numbers mean. I took the originals for the two barn photos and did a simple upsample in PSP X6 of the K-5 II image and did a quick comparison. The first thing I noticed is that I could not initially find the plane of focus in either image. The second thing I noticed is that the photo is not particularly sharp. I found the apparent plane of focus in the foreground. Here is the side-by-side of the full-resolution crop at or near the plane of focus:
There are tons of potential factors influencing the comparison. To be honest, the difference is more striking than I expected.
Steve

FastStone is basically just a good free image viewer. Amongst other things it will do side-by-side comparisons and each image in the comparison can be individually very quickly up/down scaled (using Lancosz or Lancosz2 algorithm). The numbers are just the up/down scale ratios I used to visually compare the shots.

The idea was to get a ballpark feel of how much difference there was, using the same lens and aperture etc. settings. It is pointless to forensically analyse the shots and I'm sure that Rondec would appreciate if people did not.

[Rondec]

None of theses pictures are sharp to be honest. The K5-II is sharper because at least a small part of the foreground is in focus. For the second image the exposure bias is -3.3EV. The sharpness setting in PP is set to "hard" whatever it might mean.

With all due respect, I think there a problem with thoses photos and without even counting the K3, a FA31 or DA16-50 should yield better results on a K5 than that:
- First the focussing should be put on the subject. I would have naively used the center AF point. It if was the case I would look if I happen to have frontfocus problems.
- Second for a landscape like this, I would have used more f/8 to get the max deph of field I could.
- Third I think photos on a cloudy day do not help to show sharpness.
- Fourth, I'am not sure I understand the -3.3EV exposure bias on the K3 soy bean photo shoot.
- Five I do not get how stevebrot manage to get much darker crops in his comparisons than the image are looking in the flickr.

---------- Post added 10-08-14 at 09:32 AM ----------

It also funny for me to analyse human psychology: kh want to have a point that you can't get sharpn enough image handled to really benefit of the K3 sensor resolution or even K5 so that it is plain useless to have more pixels. When he see some K3 shoots are quite sharp, there nothing said, no comment. But when there 2 shoots, both K5 & K3, both not very sharp, must softer than the K3 shoots I provided, then he confirm that was exactly what he wanted and that it show his point. One could say, yes it is not the same scene. Agree but the difference is very visible anyway !

Well from kh own arguments, there many things that can fail in a shoot and make it soft. I agree. I do think that many can be solved even when shooting handled.

One just need 1 shoot to proove that it is possible to get sharp shoots and it has been done. A few blurred shoots mean nothing. It could be the argument if that sharp look is plain luck and soft look is the normal output but then I be very, very lucky. I'am not against it. But this look unlikely to me.

As for Rondec, I ask for you opinion on yours shoots! Are there typically the sharpness you get of your shoots? You don't get ofen sharper shoots? If not, and if you are interrested, I think you have possibility to improve a lot there. In all due respect.

No. These aren't particularly sharp shots. I think there was some wind and even though I was on a tripod, I think there is some motion blur in the soybeans. The first shots I think are reasonably sharp. I usually post process photos quite a bit more to bring out clarity, etc but there wasn't any particular point in these, as they don't have any value -- I certainly wouldn't keep them normally, but kh said he wanted photos and he wanted them now and it was the best I could do with the weather the way it was here.

I have plenty of K3 photos and plenty of K5 II photos and honestly, there isn't a huge difference after processing. Yes, there is some and in a perfect world, you can print the K3 a little larger, but lack of an AA filter and an extra 8 megpixels doesn't make a huge a difference with most images. I still prefer the K3 due to speed of operation, better live view and other features.

[stevebrot]

The numbers are just the up/down scale ratios I used to visually compare the shots.

Cool. I would suggest that a downsize of the K-3 image for comparison purposes is a waste of time. The question, after all, is whether more pixels are worth it, not whether throwing pixels away looks the same.

In any case, the way I am reading your comments in this thread is that for your personal standards and needs any additional pixels are superfluous. In other words, you are happy with your most recent camera purchase. That is a good thing. I shot for seven years with my K10D and did not upgrade until the K-3 because the improvements in the K20D and K-7 did not justify the expense and the QA problems with the K-5 scared me off.


Steve

[normhead]

Cool. I would suggest that a downsize of the K-3 image for comparison purposes is a waste of time. The question, after all, is whether more pixels are worth it, not whether throwing pixels away looks the same.

In any case, the way I am reading your comments in this thread is that for your personal standards and needs any additional pixels are superfluous. In other words, you are happy with your most recent camera purchase. That is a good thing. I shot for seven years with my K10D and did not upgrade until the K-3 because the improvements in the K20D and K-7 did not justify the expense and the QA problems with the K-5 scared me off.


Steve

Well that's easy. NO the extra pixels are not worth it in any meaningful sense except for pixel peeping and bragging rights. People will try and imagine some impossible situation where it makes a difference, but that's just nonsense. It's not enough. But, 27 AF points is nice, I use then constantly, not of interest to single point spot focusers at all, but in landscape a lot of the time it saves me from having to go to manual focus. The 8 frames per second is nice for wildlife. Back when the D800 came out we had a rash of people claiming they needed more resolution.. but 16-24 Mp just isn't it. And now that we have 24 Mp APS-c, I'm not sure 36 Mp FF is it either.

Imagine the finest line visible on a 40 inch wide picture. That would be 1016 mm.
With the K-3 at 2500 lw/ph, if the printer is capable of priming that fine, each visible line would be .4064 mm

On a K5II2 at 2200 lw/ph it would be .4628mm

.4628-.4064=.0564

So people who are telling you it makes a difference are telling you they can see difference of 6/100 of a millimetre... ya right.

An inch is 25.4 mm. If you printer prints at 300 dpi the finest detail you can print is 25.4 /300= .08mm


Your printer can't print fine enough to print the difference between the two. .08 is already bigger than the difference between them. How does getting finer help?

I'm using 40 inches for my reference. For smaller prints it just gets more ridiculous.

There are instances where that .06 mm makes the image look better, and there may be instances where it makes the image just look more cluttered.

At this point the MP race is insane. People have to talk about "new technologies" to try and make a case, unless you're talking image sizes well over 40 inches.
People who talk about differences they can see at this kind of scale are just inflating their egos by messing with you.
Most of it comes down to me saying I can't see a difference and them saying "well you don't know what to look for", or "after you've seen thousands of images like me you begin to notice the difference"

They pretend they're talking about IQ, but what they are really saying is "I am so much more sophisticated than you I can see things you can't." They're lucky they can't be slapped upside the head over the internet. The comeback to that line of course is, "You're imagining that based on your pre knowledge of which is which" You can say that, but it doesn't sink in as much as being slapped upside the head might. The chances of any operant conditional taking hold are minimal, on the internet.

[dadipentak]

The chances of any operant conditional taking hold are minimal, on the internet.

Can I have my M&M, please.