[normhead]

I've been working away again this afternoon, Today's chore reduce a D800 image to 4928 (the size of a K-5 IIs image, (reduce the image sizes and see where the FF advantage ends.

The first image at 1:1 was 1128 pixels wide

Here's the comparison..



At the K-5s native resolution there is still a clear edge to the D800. Look at the cloth fabric. You can see the threads in the D800 image.

Here's the image reduced to 800 pixels wide.



The FF resolution advantage has almost completely disappeared.

Here's the image at half size. Now there's absolutely no advantage to one over the other.



And here's the size the image would be printed at 300 DPI. I'm not sure at this size you your eyes could detect the difference if there is one.



So I would estimate that if you are going to reduce your D800 image to 3500 pixels or less you could have had just as good an image from a K-5.

There is no computer screen that I know of that goes to 3500 pixels... printing to 300 dpi, thats' 12 inches, or big enough for an 8x12.


I think it's fair to say that printed at 300 dpi ( or 16 inches wide) you wouldn't see any difference without a magnifying glass.

So essentially if you want a cut off point for which there is no advantage to the best FF made, on your computer, there will be no difference at up to 3500 pixel wide images.

On a print there will be no difference up to 16 inches @ 300 dpi.

I haven't done the printing yet to find out when the difference would be come notable printed at 300 dpi, I have no idea how much detail you can see @ 300 dpi, but I'm guessing it's very little, After 12 inches though there should be a difference if you're close to the image.

If you saw my guesstimate yesterday, you know I'm guessing at normal viewing distance, I'm not sure you'd see a difference up to 60 inches. But if you wanted to do the Gursky thing and have incredible detail, as much as possible, you'll get closer to it with an FF than you will with a K-5, but you won't get anywhere near his 5x7 Linholf with a digital back, or whatever it is he uses.

I think where you'd see a real difference between the two images would be with an 8 foot image printed at close to 100 DPI. There you would clearly see the difference. If you're going that large, it's going to be well worth the D800 for the improvement in image quality, if you want to see detail as you move closer to the print from a normal viewing distance. Somewhere between there (96 inches @ 100 dpi) and 12 inches @ 300 DPI theres a point where the difference will be noticeable. But I don't have enough experience to hazard a guess as to where that might be. My printer on prints to 19 inches, so it's pretty much a moot point.

This is a comparison of resolution only, as I pointed out earlier in another thread this is the most expensive image ever sold is not very high res. 250 million for this bad boy.



Whether or not you like the sharper image better is a whole different question and open to preference. It's quite possible to decide the K-5 is better for your shooting despite the lower resolution because you like the low res look. It's also quite possible to possible to prefer a D700 with 12 MP because of the narrower depth of field.

The difference between the K-5 is enough that if you wanted the big size crystal clear high res image with the possibility of the narrowest DoF, you made a mistake buying a K-5. But it also means that if you'll never plan to need an screen image larger than 3500 pixels and you never plan a print over a conservative 36 inches.. probably more like 48 on canvas.. you saved yourself a pile of money going Pentax. The extra resolution of the D800 is worth it, obviously only if you use it. And you have to go big to use it.

---

Last edited by normhead; 10-24-2012 at 06:28 PM.

[ramseybuckeye]

Nice comparison Norm, I had to zoom 2 times to see those threads. I guess I wonder how do you know it's not the lens? It would be cool if somehow the same lens could be mounted on both. I hope you got to drink the ale after the experiment.

[normhead]

Just a short list of things that could have compromised this comparison.

Lens focus, lens quality, exposure differences, as falconeye pointed out... these kinds of comparison are fraught with peril. That being said, to my mind a real world definition based on actual comparison of images for both cameras is still better than any theoretical position based on MTF or CA measurements or anything else. So many completely fool themselves into thinking they are seeing something important looking at the numbers. When in fact they are discussing minutiae that up to 3500 pixels, can't even be measured. It is quite possible that 99.9 % of Pentax shooters will never do a print large enough or own a screen with enough pixels to take advantage of the extra resolution a D800 has to offer.

Although "Panels for professional environments, such as medical use and air traffic control, support resolutions up to 4096 × 2160." The time when you can tell the difference between a D800 and a K-5 image on a computer display is already here in some environments.

and in 2012 Apple introduced a 2880 × 1800 display on the MacBook Pro.

So Apple is already getting close. For those with the latest and greatest, and biggest and best, it may be only a few more years before the D800 will be a "have to have item" to make the most of their equipment, just to run a slide show for their rich uncle. when they make a retina display for their 27 inch monitor, I would think for many of us, the APS-c will be history. At that point the monitors will out-perform the the theoretical limit of the APS-c sensor, as it exists today. And probably not long after that it will out perform the 36 Mp of FF.

Personally I don't think packing a smaller number of sensors into a tighter area is necessarily the answer. I'm hoping for 4x5 sensor on a tilt axis camera with a 4x5 live display out the back. Of course people will still carry APS-c cameras for snapshots and convenience.

On the other hand, maybe the effects of global warming will collapse the world economy and this will be it. These guys with the D800's wil be going around with cameras that can no longer be produced and folks will be offering them thousands more than they paid just to get their hands on one. The only thing that's for certain is whatever is going to happen, it will probably be way stranger than anything I could ever dream up.

[bossa]

You forgot the thumbnail comparisons Norm....

PS: There's nothing gained by seeing what it takes to make the images look the same on a web page on the internet and on a monitor set to 72dpi. It's where the D800 is different that is one of it's advantages. Whether that difference is important to you is another matter for you to decide of course. The K-5 has it's own set of advantages too, which is why I still own one along with my two D800E's.

---

Last edited by bossa; 10-24-2012 at 08:25 PM.

[Sankon]

The comparison is interesting to an extent, but the whole FF vs Crop debate Is not reducable ( no pun .. ) to comparing resolution only.

IMHO. YMMV etc :-)

Paul

[normhead]

Nice comparison Norm, I had to zoom 2 times to see those threads. I guess I wonder how do you know it's not the lens? It would be cool if somehow the same lens could be mounted on both. I hope you got to drink the ale after the experiment.

Thanks buckeye

You forgot the thumbnail comparisons Norm....

PS: There's nothing gained by seeing what it takes to make the images look the same on a web page on the internet and on a monitor set to 72dpi. It's where the D800 is different that is one of it's advantages. Whether that difference is important to you is another matter for you to decide of course. The K-5 has it's own set of advantages too, which is why I still own one along with my two D800E's.

As for the most part these images are only seen on images on web pages on the internet.. I'm not sure of what else you would compare.. or even what else would be relevant. I've already said I lack the capacity to do comparative prints. Digital cameras produce digital images, what else is there?

The comparison is interesting to an extent, but the whole FF vs Crop debate Is not reducable ( no pun .. ) to comparing resolution only.

IMHO. YMMV etc :-)

Paul

I didn't reduce it to resolution only, I provided portions of images where contrast, sharpness, you can compare anything you want. So again, I'm not sure what you are getting at.

Listen all I'm saying here is that if you reduce the Image size of a D800 image, you reduce the resolution, and that at about 35,000 pixels wide you've reduced resolution, which also translates into the detail in the image to a level that a K-5 can match. When you consider that a D800 starts off at 7000 pixels , then you have to reduce the image size by half to fall to K-5 levels, since the sensor is half the size, that would make perfect sense. Take your data, throw away half of it and voila, you get the same image you would have taken with a sensor half the size. It seems consistent to me.

There's another way of looking at this. 16 MP, or an image 5000 pixels wide is the theoretical limit of APS-c. This is pretty much confirmed by the similar comparisons I showed in which I showed there is no increase in image quality to be had by making a 24 MP APS-c sensor as has been put in the D3200.

SO lets assume that in reducing the d800 image from 7000- to 500 pixels, we've produced the perfect 16 Mp image. It definitely has more detail than the K-5 16 MP image.

Because we can look at the D800 image the senso most likely to produce the best 16 MP image after being reduced to 16 MP, and by noticing the images don't match until 35,000 pixels, we can assume a camera/lens efficiency of about 70% when comparing maximum theoretical resolution to maximum practical resolution. If I was going to make any serious use of this information I would go back and try and determine where the exact point is that the images become identical.

If you can assume the same level of performance for a D800, which is at the theoretical max for FF, then you would assume that you will have a functional resolution of 70% of 7000 - which would be 49,000 pixels. Which curiously enough would translate to 16 MP.

SO it would appear that a D800 is a functional 16 MP camera, and the K-5 is a functional 8 MP camera. which would mean a 6D with a 20 MP sensor could conceivably produce the same IQ as a 36 MP D800. This comparison is so full of dangerous assumptions, including equal sensor capacity and ignoring the value of over-sampling, the conclusions are suspect, for even the least rigorous of us. I suggest them only as a hint as to where further testing might go. The notion can be challenged by taking a D600 image and enlarging in photoshop , and comparing it to a D800 to see if in fact there is more detail in the D800. Or if in fact there is detail captured by the D800 image missing from the D600 image. That would show that the D800 is actually using the extra MP , or doesn't, one way or the other. What's been done on this page, would suggest that it won't.

But the assumptions are so week, I'm not going to actually make a prediction as to which way this is going to go. But next time I feel like messing around, I'll give it a look,

By the way, one fo the major factors I haven't taken into consideration is the level of jpeg reduction... I haven't found away to download a raw file and my internet is so slow it's rare I even get the whole full sized jpeg.

[normhead]

OK I couldn't resist...

I checked out the D800 against the, not as much difference as between the K-5 and D800 but the 645 D the clear winner. Pentax really needs to push this format to 55-60 mm to take advantage of the size of the sensor.

And a comparison of the D800e and D600 with the D600 enlarged to D800 size.



My conclusion here would be that while the images are similar in resolution the D600 suffers from some rather extreme moire. As falc indicated, increasing the MP has been more effective at reducing moire than the AA filter on the D600.

[Winder]

So I would estimate that if you are going to reduce your D800 image to 3500 pixels or less you could have had just as good an image from a K-5.

There is no computer screen that I know of that goes to 3500 pixels... printing to 300 dpi, thats' 12 inches, or big enough for an 8x12.

By down sizing the studio shots of the D800 you are taking away the advantage the D800 has. For studio shots at base ISO an 5 year old Olympus E-3 will be more than enough. Take the D800 and the K-5 to at typical indoor event where 3200 is the minimum and then come home and print some A3+ images to compare.

There is no computer screen or printer that can replicate 14EV either, but I think everyone agrees that you are better off to have more to work with than less. You have a lot more flexibility and a better finished product.

Since you are more interested in real world results, go look at a gallery show shot on MF like the one Hassy put on at Photokina. Larger sensors have a real advantage for people who print big and are very critical of their work. Images that have that "3D" pop on a computer screen will often look rather plain when printed large. What looks crisp at screen resolution will look muddy the larger you go.

Standard viewing distance is fine for a lot of work, but if you go to a gallery with large prints on display, watch the people. When I have my work on display in a show I will stand back and watch. I want to see which prints draw the most interest. Which prints people look at the longest. Often what I feel is my strongest work is not the one that draws the most interest or the longest viewing times. But what I find is that the more interesting the print the closer people will slowly move towards it. If the print is strong people will be drawn in to see every detail. People will literally be 6" from a 20x30 inch print.

Good prints draw people in close. You don't want them to be disappointed when they get there.

[ducdao]

To some extends, this thread simply confirms that a Kia (you name the model) is as fast as a Ferrari on a jammed pack highway. Why are people still buying Ferrari? Asking is simply awnsering it.

I don't mean to say the K5 is a Kia. I think is a great camera with natural limitations.

[normhead]

By down sizing the studio shots of the D800 you are taking away the advantage the D800 has.

From PentaxForums.com: https://www.pentaxforums.com/forums/pentax-dslr-discussion/203193-last-post-h...#ixzz2AKfXlXH4

Exactly the point, if you are downsizing your images to 3500 pixels or less, the D800 has no advantage. If you need 3500 pixels or less, don't buy a D800. It will clog your hard drive and slow down your computer.

[lister6520]

Nice comparison Norm, I had to zoom 2 times to see those threads. I guess I wonder how do you know it's not the lens? It would be cool if somehow the same lens could be mounted on both. I hope you got to drink the ale after the experiment.

Should be quite easy - AFAIK Nikon can take screw mount lenses with an adapter and so can Pentax of course.

So all it needs is one good quality screw mount lens, one adapter for Nikon and one adapter for Pentax. You don't need auto exposure, auto focus or even auto diaphragm to shoot test pictures. That way you can use the exact same lens with both bodies.

[Pål Jensen]

At the K-5s native resolution there is still a clear edge to the D800. Look at the cloth fabric. You can see the threads in the D800 image.
.

Actually, the difference is less. The K-5 is focused on the bottle and the fabric is not in the plane of focus. The D800 image is focused on the bottle neck label and the cloth fabric is at or closer to the plane of focus....
The diffrences shown in this test is for practical purposes meningless. You may se bigger differences from K-5 to K-5 with different lenses, better techniques, better post processing or better printing.

[UnknownVT]

One has to pick an area that is capable of showing differences, if any.

A really silly example would be to compare shots of a plain blue sky -
then there would probably very little difference between 36Mp and even 6Mp......

ISO100 default NR JPGs from respective cameras on: Imaging-Resource.com
the 100% crops -


Pentax K-5IIS crop up-res/sampled to same size as Nikon D800E (100% crop as-is)


the difference would not be as significant when downsampled
same sized image downsampled to simulate a 300dpi 30x20 print -


Post #16 from another thread

Originally posted by ihasa
What this proves is that you downsample the results from a higher res camera to the same as a lower res camera, they look pretty similar.

we are talking about the difference between 36Mp (D800E) and 16Mp (K-5IIS)
- which is in theory 149% increase in linear resolution for the D800E over the K-5IIS
The K-5IIS is very good - but it is not miraculous.

[JinDesu]

To me, the advantage of the D800 is that I can crop. It has the pixels and the sensor size for me to use it as such. That is a huge advantage by itself.

However, why don't we try comparing a 18 or 20MP FF image vs the k-5IIs, if we're trying to claim FF advantage - or even better, a 28+MP APS-C sensor vs the D800.

[Buschmaster]

The comparison is somewhat helpful but more helpful is the price comparison:

The K-5 II is in my price range
The D800E is 3 times my price range.