Any Advantage to Full Frame?

KSV said:

Flashlight said:

Jack Hogan said:

Flashlight said:

I guess you made a mistake here as it counters your story; FX has less resolution in your example.

Click to expand...

Ah, but that's exactly the point that many miss and/or misunderstand when evaluating equivalent final-sized iimages. What counts is the resolution on the final image (line pairs or widths per picture height) not the resolution on the sensor (line pairs per mm).

Lenstip.com says that a FF setup such as the above will result in MTF50 at 44 lp/mm on the sensor. That means 2112 lw/ph on the final image because:

Final Image Spatial Resolution = 44 lp/mm x 24 mm/ph x 2 lw/lp = 2112 lw/ph

That's the figure that counts. There is no economical, Equivalent DX setup that will achieve 2112 lw/ph in large part because of the smaller format's smaller sensor. To do it it would have to show MTF50 at 66 lp/mm on the sensor. Never seen it. If there is such a DX setup let me know and I will run out and buy it.

There are many gray areas when comparing IQ from different formats and technologies. Linear spatial resolution is not one of them. A larger format will typically easily outperform a smaller one (as far as FF vs DX and mFT are concerned easily=cheaply). That's one of the reasons why Ansel Adams shot 4x5 and 8x10 - and so would I if I only could reasonably justify it for my amateurish landscape uses

Click to expand...

If all that is true can you explain why there seems to be almost no difference in resolution between these two samples?



--
Philip

Click to expand...

610 looking better to me, though margin is small. And BTW what lenses was used?

Click to expand...

D610 looks to have a little more resolution but also a little more moire. But realize you're looking from about 25 cm at something like a 1.5 meter (5 foot) wide print.

According to the exif a Sigma Macro 70mm F2.8 EX DG was used on both cameras and both images shot with f/8.

--
Philip

Simples. f/8.

MisterHairy said:

Simples. f/8.

Click to expand...

Not sure that makes much difference. The part of the image shown was on the right side of the picture (dark red columns in the image below):







--
Philip

Is that data for fx or dx? Low res camera or high res?

unless the data pertains to the cameras in the example, it may not be as informative as you feel.

[edit: I see it's canon full frame on a 5D2.]

Still, good for the 610 to show slightly better resolution and detail than the denser 7100 sensor.

Flashlight said:

If all that is true can you explain why there seems to be almost no difference in resolution between these two samples?



--
Philip

Click to expand...

And this where the mtf charts and the theories stop and the rubber meets the road. Which one can show more detail and how much noise? Very close, even with the strong AA filter on the 610.

The D600/610 sensor generally is a thing of beauty. They got everything right. Tractable all the way to ISO 6400, a lot more than the D7100, which tends to get a "pigskin" texture at about ISO 800. The D610 features flat out gorgeous color and ultra low noise as well as whopping shadow pulling ability mia on the D7100. If they had left off the AA filter, I suspect the D610 would showing its heels for detail. Anyone know if the upcoming D750 has an AA filter?

My reason for going FF is silly.... I wanted to get back to shooting the way I did when I had film cameras. I was tired of having to calculate equivalencies. I've been taking pictures for over 53 years and now with the D810, I feel like I've finally got the full functional equivalency to my previous film days. I do portraits with an 85mm. My normal field of view is the 50mm. The 35mm is a great swiss-army-knife of lenses for a day trip. I can get ultra wide angle lenses if I want to. I'm so darned 'old school' that if I could get it, I'd want a manufacturer supplied split microprism screen and return fully manual lenses.

I still prefer to use an old HP-41c calculator over the newer ones... it just goes back to what I was accustomed to.

MisterHairy said:

Still, good for the 610 to show slightly better resolution and detail than the denser 7100 sensor.

Click to expand...

They moved the camera closer for the FX shot so pixel density is not a factor, the 24MP pixels cover the scene in the same wway. Here are the links to the files:

Well, they don't. The 610 looks better so it's not the same.

Flashlight said:

Jack Hogan said:

Flashlight said:

I guess you made a mistake here as it counters your story; FX has less resolution in your example.

Click to expand...

Ah, but that's exactly the point that many miss and/or misunderstand when evaluating equivalent final-sized iimages. What counts is the resolution on the final image (line pairs or widths per picture height) not the resolution on the sensor (line pairs per mm).

Lenstip.com says that a FF setup such as the above will result in MTF50 at 44 lp/mm on the sensor. That means 2112 lw/ph on the final image because:

Final Image Spatial Resolution = 44 lp/mm x 24 mm/ph x 2 lw/lp = 2112 lw/ph

That's the figure that counts. There is no economical, Equivalent DX setup that will achieve 2112 lw/ph in large part because of the smaller format's smaller sensor. To do it it would have to show MTF50 at 66 lp/mm on the sensor. Never seen it. If there is such a DX setup let me know and I will run out and buy it.

There are many gray areas when comparing IQ from different formats and technologies. Linear spatial resolution is not one of them. A larger format will typically easily outperform a smaller one (as far as FF vs DX and mFT are concerned easily=cheaply). That's one of the reasons why Ansel Adams shot 4x5 and 8x10 - and so would I if I only could reasonably justify it for my amateurish landscape uses

Click to expand...

If all that is true can you explain why there seems to be almost no difference in resolution between these two samples?

Click to expand...

Part of using information from the net is understanding whether the data is useful for the application at hand based on the quality and reliability of the source.

The site you refer to is known not to be a good resource when comparing spatial resolution performance apples to apples. They admit to not having the time to focus cameras properly nor do they set them up Equivalently.

Lenstip.com and DxO are more scientific in their approach, try harder and give more reliable results. Lenstip's results are reproducible and easy to understand. Find me a decently priced lens whose MTF50 exceeds 66 lp/mm on a DX body at f/4 at lenstip.com and I'll go out and buy it.

But trust me, it does not exist for the reasons I mentioned above. Even with the FF bar set so low: the 50mm:1.8D is a $120 lens .

Jack

Jack Hogan said:

Part of using information from the net is understanding whether the data is useful for the application at hand based on the quality and reliability of the source.

The site you refer to is known not to be a good resource when comparing spatial resolution performance apples to apples. They admit to not having the time to focus cameras properly nor do they set them up Equivalently.

Click to expand...

Known by whom, I wonder? I'm pretty sure Dave Etchells would not agree that the IR methodology is as sloppy as you would have us believe.

Jack Hogan said:

Lenstip.com and DxO are more scientific in their approach, try harder and give more reliable results. Lenstip's results are reproducible and easy to understand. Find me a decently priced lens whose MTF50 exceeds 66 lp/mm on a DX body at f/4 at lenstip.com and I'll go out and buy it.

Click to expand...

I care not one whit for mtf charts, only what a given camera or lens can put up on my screen. Imaging Resource raws are the gold standard for determining camera body resolution. The IR raws show conclusively that there is little or no substantial difference in raw resolution at ISO 100 between the D7100 and the D610.

IMO full frame cameras tend to offer lenses of 24mm and wider as prime lenses. Prime lenses can be lighter than zoom lenses. Prime lenses often offer wider apertures, gathering more light, so especially usefull in the wide variety of low light photography situations and can generate more pleasing out of focus areas, bokeh. Non full frame cameras usually offer zooms at these wide focal lengths which are generally heavier and bulkier than a prime lens but you might need less of them!

Non full frame camera lenses offer a magnifying advantage i.e. a 100mm lens on a non full frame camera may give you the equivilent magnification of a 150mm lens etc So there are benefits to both camera types. Your choice may depend on whether you predominately take wide angle shots, low light shots, portraits or telephoto shots.

Personally I go against the modern trend of wanting a smaller camera. I love a big camera with big buttons and a large LCD screen as I have big hands and ageing eyes. I take mostly wide angle photos so on both counts it's a full frame with the best, brightest wide angle lens I can afford. The disadvantage is that bigger usually means heavier which is fine indoors or in a studio etc but not so good for lugging around on hikes or on holiday etc. A good smart phone with charger is a good option here.

I hope that is a bit helpful in adding to the mix of replies from the other responders.

Excuse me for my lack of knowledge here, but I just don't understand focal length. If we take a 50 mm lens and use it on a FF camera, we get a specific field of view - 47 degrees. On a DX camera the same lens has a 31 degree field of view. The equivalent of a 75mm lens FOV. ..... What does magnification mean? The focal length hasn't changed. What has changed is the field of view. If we take say a D7100 with a 24MP sensor, we have less of a FOV captured on the smaller sensor (16x24) with a higher density of pixels, which, I guess, is capturing more resolution because of the smaller FOV onto 24MP, but the focal length is the same. So is magnification the fact that a smaller FOV is captured on the same number of pixels (whereas in the case of a FF camera with 24MP there is a larger FOV on the same number of pixels and thus a lower pixel density over the FOV)? .... Because when I'm looking through a telescope the longer the focal length the bigger the image. Thus magnification. ... So in the case of cameras, reach is a combination of sensor size and optics or am I missing something like focal plane distance to the lens on different camera bodies?

Jack Hogan said:

Part of using information from the net is understanding whether the data is useful for the application at hand based on the quality and reliability of the source.

The site you refer to is known not to be a good resource when comparing spatial resolution performance apples to apples. They admit to not having the time to focus cameras properly nor do they set them up Equivalently.

Lenstip.com and DxO are more scientific in their approach, try harder and give more reliable results. Lenstip's results are reproducible and easy to understand. Find me a decently priced lens whose MTF50 exceeds 66 lp/mm on a DX body at f/4 at lenstip.com and I'll go out and buy it.

But trust me, it does not exist for the reasons I mentioned above. Even with the FF bar set so low: the 50mm:1.8D is a $120 lens .

Jack

Click to expand...

Assuming that the pictures discussed are otherwise fully comparable (no disrespect for IR meant) the pair of the samples for Fx vs Dx is best taken from the centre to avoid problems related to focus plane/curvature changes and using different part of the lens (effectively resolution field mapping). I believe the differences shown under current settings (f8, Sigma 70/2.8 macro) in this particular regard are quite minimal, though.

Leaving the possible or real differences among the sites in this regard aside, doing a comparison where the sensor resolution is an objective (rather than the camera’s one) one has to compare equal values by removing variances to get the idea of the difference. In this case we need to be compensating for the AA filter effect by touch of advanced sharpening (modern deconvolution versions of RL algorithm preferable to masking). And obviously best or simpler is comparing the sensors without AA filter.

On a side note, the differences in quality or nature of the detail compared between AA filter/deconvolution vs bare sensor, however marginal, is another topic.



--
Hynek

Hey Reilly,

Reilly Diefenbach said:

Jack Hogan said:

Part of using information from the net is understanding whether the data is useful for the application at hand based on the quality and reliability of the source.

The site you refer to is known not to be a good resource when comparing spatial resolution performance apples to apples. They admit to not having the time to focus cameras properly nor do they set them up Equivalently.

Click to expand...

Known by whom, I wonder? I'm pretty sure Dave Etchells would not agree that the IR methodology is as sloppy as you would have us believe.

Click to expand...

by Dave himself or whoever responded to the question asking for clarification in their methodology section: they said clearlly that they do not have the time to focus peak every image they take. I can confirm that , having taken a close, quantitative look at some of their raw captures.

Reilly Diefenbach said:

Reilly Diefenbach said:

Lenstip.com and DxO are more scientific in their approach, try harder and give more reliable results. Lenstip's results are reproducible and easy to understand. Find me a decently priced lens whose MTF50 exceeds 66 lp/mm on a DX body at f/4 at lenstip.com and I'll go out and buy it.

Click to expand...

I care not one whit for mtf charts, only what a given camera or lens can put up on my screen. Imaging Resource raws are the gold standard for determining camera body resolution. The IR raws show conclusively that there is little or no substantial difference in raw resolution at ISO 100 between the D7100 and the D610.

Click to expand...

When making hardware choices I prefer to deal with objective, quantitative data about the hardware unencumbered by subjective calls on processed images: otherwise the ensuing discussion tends to degenerate into a tedious at best 'you are blind - no you are blind' unending exchange. But since you mention it, and with all the disclaimers above, the D610 does look a little sharper in the example shown.

IR is good for many things for the average bloke. Comparing spatial resolution capabilities across cameras is not one of them in my experience.

Jack

HSway said:

... one has to compare equal values by removing variances to get the idea of the difference. In this case we need to be compensating for the AA filter effect by touch of advanced sharpening (modern deconvolution versions of RL algorithm preferable to masking).

Click to expand...

Sure, that introduces additional, potentially subjective variables but it is a valid approach.

HSway said:

And obviously best or simpler is comparing the sensors without AA filter.

Click to expand...

Sure, or both with

HSway said:

On a side note, the differences in quality or nature of the detail compared between AA filter/deconvolution vs bare sensor, however marginal, is another topic.

Click to expand...

Right. This is an interesting topic that I have some quantitative data on. When I have time I will start a thread on it in the Photographic Science and Technology Forum titled 'Why it's not a good idea to use deconvolution with a gaussian PSF liberally on AAless cameras'.

Jack

Sometimes the test chart doesn't show the differences. At base ISO, at a moderate aperture, using a largely 2 dimensional test target, and cameras with moderate resolutions, you will not see much if any difference.

The test chart used above is very instructive. The D610 uses an AA filter and shows a hint of moire in that sample. The D7100 has no AA filter, but doesn't seem worse off for it. What's happening can be partially explained by diffraction.

A 24MP DX sensor begins to be diffraction limited at f/7.4. You can just see start to see it softening the image.

See this link: http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm

A 24MP FX sensor begins to be diffraction limited at f/11.3

The DX sensor is closer to the edge where it will start returning less real resolution. Nikon has balanced things out by removing the AA filter, but as sensors become more densely packed, real world resolution will not scale as well in DX as in FX.

For example, for an FX sensor to be diffraction limited at f/7.4, it would have to have 56MP.

We don't have those sensors yet, but they will start to take us to the edge of practical limits. Such a density in DX would be limited at only f/4.9. Often f/5.6 to f/8 are needed to keep an adequate amount of the subject area in focus, so we will see diminishing returns from both, but we will see them more rapidly on DX.

It might not matter for common print sizes in terms of strict resolution, but the output will look different .

Jack Hogan said:

When making hardware choices I prefer to deal with objective, quantitative data about the hardware unencumbered by subjective calls on processed images: otherwise the ensuing discussion tends to degenerate into a tedious at best 'you are blind - no you are blind' unending exchange. But since you mention it, and with all the disclaimers above, the D610 does look a little sharper in the example shown.

IR is good for many things for the average bloke. Comparing spatial resolution capabilities across cameras is not one of them in my experience.

Jack

Click to expand...

DPR has the exact same result, so I think this average bloke will continue to rely on these and other actual pics.

Which is why I got the D800e :^)




Proportional Scale In Focus

Oh dear!

My use of magnification was just a simplistic term meaning the comparison of what you see through the lens; for example when you put a 100 mm focal length lens on a full frame camera, say a Nikon FX, you see an image of a certain size. When you put the same lens on say a Nikon DX camera the image you see will be bigger (magnified) by half.

Sorry for not being clearer.