[Alex645]

I totally agree on that, but then why does the pixel count on the sensors keeps rising? Purely for marketing reasons? Or does the professional world really benefit from that? I guess for some applications it makes sense but is it really so common?

I think there are many reasons why MP keeps rising other than marketing. Other reasons include economic as obsolescence helps sell new tech or for competition and also the scientific, military, and intelligence (spy) fields.

We also see it happening on screen going from HD to Retina to 4K etc, from CRT to plasma to LCD to LED to OLED and now ULED. Each improvement is bigger, better, and initially more expensive. The only field that Iʻm aware of that has sort of gone backward is audio recording. Music compression in the for example in the form of mp3 files has really dumbed down the listener for the sake of small audio files for iPods, streaming, mobile devices, etc. In that case, I think most everyone benefitted except the audiophile.

[normhead]

I think there are many reasons why MP keeps rising other than marketing. Other reasons include economic as obsolescence helps sell new tech or for competition and also the scientific, military, and intelligence (spy) fields.

We also see it happening on screen going from HD to Retina to 4K etc, from CRT to plasma to LCD to LED to OLED and now ULED. Each improvement is bigger, better, and initially more expensive. The only field that Iʻm aware of that has sort of gone backward is audio recording. Music compression in the for example in the form of mp3 files has really dumbed down the listener for the sake of small audio files for iPods, streaming, mobile devices, etc. In that case, I think most everyone benefitted except the audiophile.

Young audiophile perhaps. When 16 I heard things well beyond what an average teenager could. At 70, I don't hear things any teenager can hear. When your hearing is limited to 10 mhz or less, you don't need great audio equipment. For me it was always about a nice clean high end. Back in the day, my Yamaha equipment sounded awesome my Sony equipment sounded like mud vibrating. I can plug in the same two amps now and not hear a difference. I simply don't have the same ability to appreciate high end audio equipment I did when I was younger.

Although it still irritates me that my wife thinks her internet speakers sound just as good as my Yamy surround sound system with NS-30x speakers. She says the richness of the sound distracts her.

[CDW]

Young audiophile perhaps. When 16 I heard things well beyond what an average teenager could. At 70, I don't hear things any teenager can hear. When your hearing is limited to 10 mhz or less, you don't need great audio equipment. For me it was always about a nice clean high end.

Wow, if you can hear to 10 megs, I want some of your meds!

---------- Post added 02-06-18 at 06:36 AM ----------

We also see it happening on screen going from HD to Retina to 4K etc, from CRT to plasma to LCD to LED to OLED and now ULED. Each improvement is bigger, better, and initially more expensive. The only field that Iʻm aware of that has sort of gone backward is audio recording. Music compression in the for example in the form of mp3 files has really dumbed down the listener for the sake of small audio files for iPods, streaming, mobile devices, etc. In that case, I think most everyone benefitted except the audiophile.

ULED isn't necessarily an improvement beyond OLED, except perhaps in panel manufacturing costs. It's essentially a technique of using full panel LED back lighting in combination with a QLED type of display to enhance contrast. An OLED panel can turn off individual sites and deliver a true black, something an QLED or ULED panel is unable to do.

I agree with you regarding the audio quality of MP3s but high resolution audio files with sampling rates of 96-192 kHz at 24 bits, can take audio quality far beyond vinyl or 16 bit CD. So there has been some improvement in that arena, as well.

[tuco]

Hi there,

... medium format film does not produce higher resolution of image than 35mm one (minus the lens resolution factor), as both formats are using the same film material thus same 'pixel density' but higher 'pixel count' on medium format, it is how large the image we reproduced on paper/screen matters.

Since medium and large format film also have more area to define the image, they also have better tonal scale and look more life-like for the same size print

[Bob 256]

Hi there,

I am not pro this is just my understanding...........

.........just want to clarify, we don't mix "pixel count" and "resolution", larger sensor naturally has higher pixel counts but NOT pixel density (resolution), and medium format film does not produce higher resolution of image than 35mm one (minus the lens resolution factor), as both formats are using the same film material thus same 'pixel density' but higher 'pixel count' on medium format, it is how large the image we reproduced on paper/screen matters.

Resolution is actually a measure of how many pixels there are in a specific direction in a given picture (horizontal or vertical). Because it can be measured in X and Y directions, there can be two different resolution figures. These resolution figures might be pixels per mm or maybe "dpi" for a printed image. They can also be stated as the total horizontal or vertical pixels in a complete image (e.g., 1200 x 1600 pixels). 36.4 Mpixels is a measure of the total pixels in an image which relates to resolution but isn't "resolution".

When UHD television came out, the industry mistakenly (on purpose) stated that UHD had four times the resolution of HD. UHD has four times the pixels but only twice the resolution of HD. Pixel count goes up with the square of resolution. Double resolution and the pixel count will increase by 4. Increasing the size of the sensor doesn't necessarily imply that the resolution has increased.

Taking the K-01 and the K-1 as examples (they both have very similar sensor structures), the pixel pitch is the same for each but the K-1 has extra pixels, making it larger. They both have identical pixel density (because the pitch is the same in both). For a given lens, both should have identical resolution for a cropped image (cropped smaller than the APS-C frame). The K-1 can have higher image resolution because when the image is enlarged to cover the K-1 sensor (with a longer focal length lens), there are now more pixels in that same image. Now the cropping principle doesn't apply.

Think of building a square brick patio with dimensions of 12 feet by 8 feet. That's the APS-C K-01 sensor. So many people can sit on it. To get more seating space, enlarge your patio to 18 by 12 feet. That's the K-1 sensor. More bricks and larger surface (2.25 times as large an area). A single person still covers the same number of bricks, but more people can fit in this patio. If the people ate a lot and became 1.5 times their original dimensions, then the patio would only seat as many as the smaller patio did. But, because the people are larger, they cover more bricks per person than before so each person is "resolved" better.

I'm done

[geo444]

...Taking the K-01 and the K-1 as examples (they both have very similar sensor structures)...

.
Yes, and if you choose 2 Sensors closer in time with very similar Pixel Pitch and Technology...
... then, you could verify one more thing :
their Global Sensitivity ( as measured in ISO by DxoMark ) are proportional to their Area
- the condition is : all other things equal -

Taking the Nikon D5100 APSc 15,5 x 23,5 ( ISO 1183 ) and the Nikon D800 FF 24 x 36 ( ISO 2853 )

so 1183 x 24 x 36 / 15,5 / 23,5 = 2806 >>> a result that's very close to 2853 !

[lotech]

Smaller the pixel pitch so higher pixel count on the same area, thus higher pixel density/resolution, it's just another way to call it. One thing I don't like about digit is the name/terms keep 'inventing' everyday, causing confusion especially to newbies.

Pixel Pitch is the distance between the centres of each adjoining pixel. The smaller the pitch, the closer they are packed together.
In terms of quality, a bigger pitch is better, since adjoining pixels interfere (which causes noise), and the more space between them the easier it is to deal with heat (which causes noise), and the easier it is to wire up without interference (which causes noise). Hence why you can easily squeeze 10+ MP into a sliver of a cellphone sensor, but its noise performance is awful.

In an ideal world you would have 100+MP in a sensor the size of a sheet of A4. But then you also want to carry the camera, and pay for it, and producing lenses that can resolve well in a huge image circle would be phenomenally expensive and heavy. So in the real world there is a compromise to be had in which, if size and cost wins, you end up with a cellphone or compact; if quality wins you end up with medium format with larger pixel pitch and therefore relatively modest pixel counts.

[normhead]

Resolution is actually a measure of how many pixels there are in a specific direction in a given picture (horizontal or vertical).

Lumix FZ-1000 (one inch sensor) easily out resolves a K-5 at 100 ISO or a Canon 5D at 12 MP. Your theory needs a little work.

Resolution is usually measured in lw/ph, (Line Widths per picure hieght)
A

Think of building a square brick patio with dimensions of 12 feet by 8 feet. That's the APS-C K-01 sensor. So many people can sit on it. To get more seating space, enlarge your patio to 18 by 12 feet. That's the K-1 sensor. More bricks and larger surface (2.25 times as large an area). A single person still covers the same number of bricks, but more people can fit in this patio. If the people ate a lot and became 1.5 times their original dimensions, then the patio would only seat as many as the smaller patio did. But, because the people are larger, they cover more bricks per person than before so each person is "resolved" better.

Assuming 8 bit capture, each pixel site on an FF sensor will contain a number between 0 and 255 in each channel, RGB and luminance. An image captured with an APS-c sensor will be the same.. After the image is captured, you can't tell from sensor size or resolution which sensor size was used to capture the image, it's just a bunch of numbers.

RGB values of 187,165,199 are exactly the same on whatever format they were captured.There is no FF magic that produces better numbers. And regardless of the size of the sensor, they will enlarge in exactly the same way. There is no "it's not enlarged as much" advantage to the FF numbers. Th imaging software can't even tell one from the other.

You guys are discussing noise as if it's always visible. It's not. Which camera produces the noisier image only matters after noise becomes visible. So, up to 640 ISO my K-3 produces the same amount of noise practically, as my K-1. You may be able to use really heavy contrast to make the noise more visible but essentially, you don't see it. So statements like the smaller sensor creates more noise is meaningful only after 640 ISO on a K-3, 800 ISO on a K-5 or 1600 ISO on a K-P/ Up to those points there is no noise advantage to a larger sensor in practical terms.

There are two effects of higher pixel pitch.
The negative is visible noise kicks in at lower ISOs, like around 800 ISO on my K-3 instead of 3200 ISO on my K-1.
The positieve is if you are a wildlife or macro guy, you get 50% more magnification with the smaller sensor.

I'm sure you guys are sceptical but look at some actual resolution measurements.

In the first one two, the lowest two in MP but both full frames, these two FFs present the least amount of information as measured (notice measured, this is not theory, this is not what happens when someone applies their biases and thinks will happen, this is measured (in a lab) in lw/ph
Canon 6D Canon 6D Review - Exposure
- 2400 lw/ph


Canon 5D mkII Canon 5D Mark III Review - Exposure
2400~2500 kw/ph


A one inch sensor should produce less resolution right? Not so fast.

Panasonic FZ1000 Panasonic FZ1000 Review - Exposure
~2,500 to ~2,600


And the Pentax K-3 kicks all their butts.
Pentax K-3II - Pentax K-3 II Review - Exposure
~2,700 to ~2,750 lines of strong detail

Nikon D750 Nikon D750 Review - Exposure
resolution, ~2,850 to ~2,900


OK stop here for a second the difference between a 24 MP K-3 APS-c and a 24 MP D750 is at 100 ISO 150 lw/ph. That's about a 5% difference in resolution. Honestly, that's not visible to the human eye in all but a tiny number of cases.. They are functionally the same. The FF advantage is at higher ISOs like above ISO 800 where the larger light collecting surface produces less noise. Up to that point, the APS-c camera will do pretty much everything the FF will do and produce the same IQ. The FZ 1000 will be the same, but is only good at 200 ISO or below, noise starts to impact IQ after 200 ISO.

Pentax K-1 Pentax K-1 Review - Exposure
~3,350 lines of strong detail


You're going to say if you an FF is better absolutist that the K-1 will produce higher IQ images, and there may be circumstances where that's true.
Shortly after getting the K-1 my wife and I went out to a nice landscape location with her K-5, my K-1 and the DFA 28-105. We took the same scene, K-1 at ~50mm, K-5 at ~35mm. Should see a difference right?

We didn't. We put both images up on my wife's 4K monitor and flipped back and forth. There was no visible difference between the two images even though technically there was an 25% difference according to test chart measurements.

There are a pile of things you need to add to your knowledge base here.

At high 100 ISO, small sensors provide the same performance at much less cost and weight.
Crop sensors of the same MP produce much more magnification and resolution on the subject than an FF with the same lens.
APS-c sensors shot at 800 ISO or less are just as good as FF sensors.

The FF high ISO advantage is tempered by a couple of factors.
- not every image needs resolution, in fact one of the most arrogant conceit I see pushed in these types of discussions is "I need the extra resolution". Most people don't.
- high ISO images may be better on FF than high ISO on smaller sensors, but not better than low ISO smaller sensor images.
-high ISO images for the most part have too low dynamic range to be effective images. The fact that the FF images may look better than the smaller sensor images doesn't always help. The APS-c image looks bad, the FF images look not as bad, but still bad.

Sometimes bad id bad.

So if you look at the advantages and disadvantages it's always a trade off. How much is that low light performance worth to you? Are you willing to carry a lot of extra weight, pay a lot more, and be at a serious disadvantage for macro and wildlife to get it?

There's just no free lunch. Most people in this type of discussion seriously over-estimate what full frame is going to bring to the table. If you are a wedding shooter and want images documenting events in less than perfect light, the FF high ISO quality is a needed feature. Fast glass is a needed feature. If you are shooting wildlife out doors in good light, an FX 1000 with it's 1000mm equivalent lens may give you better results at a fraction the weight.

A recent Pentax XG-1 1:1.2 sensor, 1000mm equivalent at 200 ISO


Where is this noise of which you speak? Practically, if it's not visible, it's not there.

Imagine what you'd have to carry instead of this, if you wanted to take this picture with an FF. And the crazy thing is, there is a pretty good chance, you would't get more IQ in your image for all that work. You'd carry 16 pounds instead fo a half pound, for 8 km, to get to this lake, and then back out again. Right tool for the job I always say, FF has as many disadvantages as any other format. There are times to take it, there are times to leave it home.

[Bob 256]

Lumix FZ-1000 (one inch sensor) easily out resolves a K-5 at 100 ISO or a Canon 5D at 12 MP. Your theory needs a little work.................

Perhaps I wasn’t clear. I wasn’t referring to a sensor - I was referring to any two images of same size. If you compare one 10 x 17.8 inch photo (1080 x 1920 pixels - HD quality) to a second identical photo 2160 x 3840 pixels - UHD quality), the second will have higher resolution than the first and be clearer. If the sensors which captured these two photos have those pixel counts,
it’s irrelevant what sizes they are, the second will have higher resolution because it contributes more pixels to the photo.

Total pixels and pixel density can be confusing. Two identical sized sensors can have different resolutions as in the above case if one (the UHD sensor) has a higher pixel density. If the sensor size is changed without changing the total pixel count, the pixel density changes but sensor resolution stays the same.

If you take a single sensor, you can get two different resolutions of a fly if you use different lenses so as to make the fly cover more or fewer pixels on the sensor (assuming you don’t image the fly beyond the edges of the sensor). By the same token, you could use two different sized sensors (each with the same pixel count) and get the same image resolution of our fly if you
switch lenses (keeping the “fly coverage” on the same number of pixels on each sensor).

One more. You could take a 36Mp sensor and get less image resolution than a 3Mp sensor if you used all of the pixels on the 3Mp sensor for the fly, and switched lenses to cover fewer pixels on the 36Mp sensor with the fly image.

All this assumes ideal infinite resolution lenses. In the real world, that's not the case and a lens's resolution can easily rule over that of the sensor (the lower winning). Comparing sensors can be tricky because two different lenses may be involved and the difference (or similarity) in image resolution may not be due to the sensors but rather the lenses. Sensor comparisons are best done with the same lens (a really good one) and the test scale (object) size adjusted for same image scale relative to the sensor size.

[stevebrot]

Pixels = resolution, simply put. Or, put another way, more data (pixels) is more data.

How dense they are clustered for capture*, display, or print is irrelevant from a resolution perspective. (Capture is normalized by degrees of arc for the subject so lineal density cancels out leaving only pixels.)

...and switched lenses to cover fewer pixels on the 36Mp sensor with the fly image.

You're going to have to post an example of this one. I don't have a single lens on my shelf that is able to do an on-sensor crop.


Steve

* Technically, being logical constructs, pixels only exist post capture and then only by analogy, but we won't go there...

[normhead]

it’s irrelevant what sizes they are, the second will have higher resolution because it contributes more pixels to the photo.

I guess you didn't read the part in my post where a K-5 image is indistinguishable from a K-1 image at 4000x2300, a test done with real cameras and real images. I'n not saying it will be like that every time but it will be sometimes.

The K-1 "contributed" more pixels to the photo, but the parameters as I defined them did not allow those pixels to make a difference to the final image. When you say this kind of thing, you have to define at exactly what size the difference will be noticeable. Just saying one will be better is clearly wrong. You have to say, there will be circumstance where it will be better, and define what they are.

I understand what you are trying to say, it's just that my tests have shown it isn't always right. I'm saying with 7360x4912 looks the same as 4928x3264 when both images are reduced to 4000x2600. The fact that you are reducing both images means that the larger oversample of the larger file may make no difference to the reduced file.

The only way the larger file is going to be much better would be if you use a pixel size larger than the smaller file. If you actually have to expand the smaller file, i would expect the larger file to have more resolution, but that's not a test I've done.

Up to 20 x 13 if you use 200 pixels per inch, the images should be identical. And since it's hard to tell the difference between images printed at 150 DPI and 200 DPI... I really don't care about what should theoretically be true. If I want to know something I run a test, and what I turn up is often different from what people expect. So overall i question the value of these types of assumptions, unless you have actually run tests. The problem is that theoretical assumptions do not always produce the results you'd expect, because of the limits of human perception among other reasons. There may actually be a difference, but if you can't see it, it's not worth mentioning. And if you haven't run a few tests to see if the difference is perceptible, you really don't know anything real.

[Bob 256]

.............. I'm saying with 7360x4912 looks the same as 4928x3264 when both images are reduced to 4000x2600. The fact that you are reducing both images means that the larger oversample of the larger file may make no difference to the reduced file...................

No argument there. I didn't mention resampling anywhere in my discussion and was referring to an original photo utilizing all available sensor pixels (in the photo as well - a hypothetical in most cases) to the same from another camera with fewer pixels and that same image rescaled to fit that sensor. When images taken at two different native resolutions are resampled to a new resolution, most all I stated is simply not applicable.

It's also quite possible the lens used on a camera with fewer pixels has better angular resolution than the one used for the higher pixel count camera which makes the comparison of the two iffy at best concerning concerning sensor advantages in picture quality.

I think there is some confusion for many, in the terms used. There is sensor resolution (stated different ways: total pixels, pixel count in H&V directions, and pixel spacing), image resolution (which utilizes actual sensor pixels, or resampled pixels - the latter with some or little relation to the actual sensor pixels), lens resolution (which can be higher or lower than the sensor capability), and image quality (which depends on the three previous in combination).

If image resolution is a resampled version of the original sensor capability, you're entirely correct in saying two different sensors can deliver equal image quality (or the sensor with less resolution can actually give better image quality under the right circumstances).

I have no problem at all in carrying a different camera (even a very compact point & shoot) for different jobs, and I too have seen some outstanding images (image quality) from cameras with diminutive sensors. I'm not advocating a Bukoo pixel FF sensor for all cases, just stating that it has the potential for better images under the proper circumstances.

[normhead]

No argument there. I didn't mention resampling anywhere in my discussion and was referring to an original photo utilizing all available sensor pixels (in the photo as well - a hypothetical in most cases) to the same from another camera with fewer pixels and that same image rescaled to fit that sensor. When images taken at two different native resolutions are resampled to a new resolution, most all I stated is simply not applicable.

But in the real world, who does that? 99% of us resample 99% of our photos. (I just made that up.). There are no displays that can show un-resampled photos. Prints, no definitive work has been done, but what little has been done has shown that resolution is not the determining factor in whether or not people like prints. It would take more work than I'm willing to put into to sow that over 100 DPI resolution makes any difference at all to whether people enjoy the print. Some have taken that as low as 72 DPI, same result. It's not that you can't see a difference, it's that the difference you see usually doesn't increase the value of the print. I'm not sure who your point is relevant to.

It's very unlikely very many people ever actually see the improved IQ of their large format camera.

The thing that impresses me bout the K-1 is not the added pixels. It's 7360 across instead of 6016. That's only a 22% increase in the number of pixels in given direction. It's not at all critical.

The increase in Dynamic Range however, images like the absolutely depend on it.


I have not one image where I can say "I used every bit of the resolution on this one" I can find many where the DR was taxed, and I could have used a little more.Taken with my K20D none of the shadow detail would have been there in the above image.

[dubsider]

Earlier in the thread was discussed how, or if a lens could resolve at the pixel counts of the high MP cameras.. Where do you find that lens metric?