Full Frame advantage real?

[Dan62326]

This has probably been discussed before so please excuse me if this is a repost.

I was thinking about the stuff Ken says in this page.

http://www.kenrockwell.com/tech/full-frame-advantage.htm

And I wanted to see how much of it was true in the real world.

After reading his "about me" page, I'm not sure what to think. He doesn't seem like he runs these comparisons correctly.

What I was wondering was if it is true what he says about poor quality lenses doing better on FF compared to the best 1.5 crop sensors. The discussion about medium format vs. 35mm seems logical. How much does this carry into the FF vs. APS sensors?

After seeing his D3 vs D300 shots, it seems like there's something wrong with the way he's testing these cameras or he's forming conclusions not knowing the true story.

http://www.kenrockwell.com/nikon/d3/sharpness-1.htm

So back to the essence of my original question. If I slap-on a 24-105 on a 5D, will it outperform a D300 with say a 24-70 F2.8? And the reason I mention the 24-105 is because that lens interests me and it shows poor MTF results near the long end of its zoom. Maybe I should say a D300 with a 70-200mm F2.8 to make the comparison at 105mm.

I was just reading an article on luminous landscape about FF and it said something about people that go for FF get it and those that don't, don't. I don't get it.

I just know the basic differences.

larger pixels gather more light, less noise, bla bla bla
then corners are less sharp

So for those that have FF, have you noticed a new level of quality in your images?

It seems to me that the AF system on a FF camera wouldn't have to be as accurate/precise compared to that of a smaller sensor camera. Is this true or not?

 

[gdanmitchell]

In general (though not every time) you can trust what you read on Luminous Landscape a lot more than what you read from Ken Rockwell...

Just sayin'

Dan

--
---
G Dan Mitchell
SF Bay Area
http://www.gdanmitchell.com/

 

[lovethejob]

It seems all very confusing.

However it is true that bigger pixel gives better image quality if there are enough that bigger pixels to form a high resolution image.
Higher pixel count does not auotmatically resulting better image quality.

The combination of the pixel number, sensor size, lens quality and image processing determines the final outcome.

That is the reason a 12MP P&S camera could not compete its image quality to an old EOS 1Ds with less pixels (11MP).

--
http://www.pbase.com/ltjiang

 

[joe mama]

Firstly, all of your questions are discussed in quite some depth here:

http://www.josephjamesphotography.com/equivalence/

So back to the essence of my original question. If I slap-on a 24-105
on a 5D, will it outperform a D300 with say a 24-70 F2.8?

Most likely, yes. But there are many variables. First of all, assuming the glass is sharp enough to resolve each individual pixel, then the sensor with more pixels will resolve more detail, regardless of the size of the sensor.

However, obviously, glass is not perfect, and it will not resolve every pixel equally as well, even if, on average , it is sharp enough to resolve all the pixels. The most notable example is the edge performance of the glass. Usually, the sharpness of FF glass falls off significantly in the extreme corners, even for the same aperture (which means the same DOF, not the same f-ratio) so cropped sensors will enjoy an advantage there, even though the FF sensor will enjoy an advantage in the other 95% of the frame. However, I've yet to take a single pic, or see someone else's pic, where the 5% extreme corners mattered to me. Others may feel differently.

Anyway, an outstanding comparison in the regard is linked in the essay's "Evidence" section:

http://forums.dpreview.com/forums/read.asp?forum=1022&message=25380951&changemode=1

the reason I mention the 24-105 is because that lens interests me and it
shows poor MTF results near the long end of its zoom. Maybe I should
say a D300 with a 70-200mm F2.8 to make the comparison at 105mm.

It's "improper" and just plain odd to compare lenses on different systems that have radically different FOVs. In other words, what is the point of comparing a a 24-105 / 4L IS on FF to a 70-200 / 2.8 on 1.5x even at the same FOV?

I was just reading an article on luminous landscape about FF and it
said something about people that go for FF get it and those that
don't, don't. I don't get it.

I don't know the argument. However, I can summarize the advantages of FF thusly:

1) FF usually has more pixels, thus giving greater detail

2) FF is sharper in the middle 95% of the frame for the same FOV and DOF

3) FF allows a more shallow DOF, if desired (it can go just as deep as crop, however)

4) FF will have less noise than crop if you are willing/able to sacrifice DOF/shutter speed to get it (there is no noise advantage for FF for the same DOF and shutter speed)

So for those that have FF, have you noticed a new level of quality in
your images?

Big time. However, let's put "big time" into perspective. At 300 PPI, an 8x12 print is only 8 MP. Thus, more MPs only help with prints larger than 8x12 or allow more lattitude to crop. My office is "littered" with tons of framed prints from 8x12 to 20x30, from compact digicams to the 5D. Can I tell which are from which? Yes -- the 20x30s from the compacts are obvious. : )

However, the 8x12s and 12x18s from the 300D, 20D, and 5D are indistinguishable in terms of IQ. Then again, I don't have pics of the same scene from each camera, so maybe it would have made a difference. For example, I have this one 12x18 ISO 3200 pic (in poor light) from the 5D that is noiseless. Would it have been as good from the 20D? I don't know.

People who come to my office praise my pics all the time: "They're all so beautiful", "That's amazing", "Are these digital?" Yet not one person, ever , has pointed to a 5D pic and said how much better it is than any of the others. In fact, two of the pics that get the most praise are crops (to fit the 3:2 aspect ratio) from a Canon G1 (!) at 20x30in, that are obviously lower in IQ than my other pics. But only when you get close enough, and even then, no one seems to care but me.

The simple fact of the matter is that it is the scene that matters far more than anything else for the vast majority of people. And while there is a significant difference between compacts and DSLRs (but there is relatively no difference between 1.5x and FF), these differences manifest themselves mainly in DOF and high ISO situations.

Furthermore, the manner in which an image is processed often matters significantly more than the equipment that made the image.

It seems to me that the AF system on a FF camera wouldn't have to be
as accurate/precise compared to that of a smaller sensor camera. Is
this true or not?

AF accuracy has nothing to do with the sensor size. It is related to the pixel size in that the more the focus is off, the less useful the additional pixels are to the image. Furthermore, the deeper the DOF, the less AF matters. In addition, the less the subject moves, the less AF matters. For example, landscape and macro photographers do just fine with manual AF for exactly those reasons, whereas AF speed and shutter lag are very important to those that shoot shallow DOF candids (hello!).

Bottom line: if you are printing no larger than 12x18 (and quite possibly 20x30), the differences between modern DSLRs matter more to the photographer than they do to the non-photographer, if they matter to the non-photographer at all. That said, I'm a photographer, and the differences between the 20D and 5D are often striking to me for the type of photography I do.

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

 

[joe mama]

A mix of 8x12s (the middle is a 12x18) from compacts, 300D, 20D, and 5D:

http://www.pbase.com/joemama/image/84366444



8x12s from the 300D, 20D, and 5D:

http://www.pbase.com/joemama/image/84366445



All 300D:

http://www.pbase.com/joemama/image/84366446



3:2 crops at 20x30 from a Canon G1:

http://www.pbase.com/joemama/image/84366447



Flowers are from the 20D, the center pic from a Canon G1:

http://www.pbase.com/joemama/image/84366448



And there are more that I didn't take pics of, for whatever reason. : )

Anyway, they're all good, or I wouldn't proudly display them in the office. That said, the 5D is significantly better, even if I'm the only one that cares.

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

 

[deeyay]

...and not the other way around. With full frame you get a much narrower depth of field so focusing accuracy is much more important (the biggest change for me when switchingb to FF). Slight misfocus on an image is way more noticable on FF than on APS-C.

--
Picture agency

• http://danieljansson.se

Web portfolio

• http://danieljansson.com

 

[joe mama]

...and not the other way around. With full frame you get a much
narrower depth of field so focusing accuracy is much more important

You do not get a more narrow DOF on FF unless you want a more narrow DOF. For the same perspective and FOV, set the f-ratio 1 1/3 stops higher on FF than 1.6x. If the same shutter speed is necessary, up the ISO 1 1/3 stops. While that increases the noise, it only increases the noise to 1.6x levels for images of the same output size and detail. That is, a 12.7 MP 5D image at ISO 1000 will be more noisy than an 8.2 MP 30D image at ISO 400, but if you resample the 5D image to 8.2 MP, then the noise will be basically the same, as you are basically trading detail for noise. On the other hand, if you upsample the 30D image to 12.7 MP, then the upsampled 30D image will have less noise, but also less detail. In this case, you would apply NR to the 5D image until it matched the level of detail of the 30D image, and, once again, the noise/detail parity is maintained. Actually, in both instances, the 5D image will still be slightly sharper since the lens is stopped down to achieve the same DOF.

(the biggest change for me when switchingb to FF). Slight misfocus on
an image is way more noticable on FF than on APS-C.

Yes. The more shallow the DOF, the more important precise focus becomes. However, it's important to understand that shallow DOF is an option of FF, not a requirement.

More on the differences between systems with different formats:

http://www.josephjamesphotography.com/equivalence/

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

 

[Dan62326]

Firstly, all of your questions are discussed in quite some depth here:

http://www.josephjamesphotography.com/equivalence/

Thanks so much for that link! Quite a bit of info that'll take a bit of time to digest. Did you write that?

So back to the essence of my original question. If I slap-on a 24-105
on a 5D, will it outperform a D300 with say a 24-70 F2.8?

Most likely, yes. But there are many variables. First of all,
assuming the glass is sharp enough to resolve each individual pixel,
then the sensor with more pixels will resolve more detail, regardless
of the size of the sensor.

How come the blur plot of the 24-105 on the 5D is worse than the one on the 20D? I assume it has something to do with equivalence?

http://www.slrgear.com/reviews/showproduct.php/product/145/cat/11

the reason I mention the 24-105 is because that lens interests me and it
shows poor MTF results near the long end of its zoom. Maybe I should
say a D300 with a 70-200mm F2.8 to make the comparison at 105mm.

It's "improper" and just plain odd to compare lenses on different
systems that have radically different FOVs. In other words, what is
the point of comparing a a 24-105 / 4L IS on FF to a 70-200 / 2.8 on
1.5x even at the same FOV?

I think what I meant to say was to compare the FF at 105mm to the 1.5x at 70mm. This would be the 24-105 at its worst and the 70-200 at its best.

2) FF is sharper in the middle 95% of the frame for the same FOV and
DOF

What if we use the same F-stop number instead but keep the same FOV?

(but there is relatively no difference between 1.5x and FF), these
differences manifest themselves mainly in DOF and high ISO situations.

I'm all for better high ISO performance. Is this just because of the larger pixels? DOF I care less about.

It seems to me that the AF system on a FF camera wouldn't have to be
as accurate/precise compared to that of a smaller sensor camera. Is
this true or not?

AF accuracy has nothing to do with the sensor size. It is related to
the pixel size in that the more the focus is off, the less useful the
additional pixels are to the image. Furthermore, the deeper the DOF,
the less AF matters.

What about the range of what "in-focus" means? Would it be easier to make the AF system faster on an FF camera compared to a 1.5x camera if the AF systems were the same?

the differences between modern DSLRs matter more to
the photographer than they do to the non-photographer, if they matter
to the non-photographer at all. That said, I'm a photographer, and
the differences between the 20D and 5D are often striking to me for
the type of photography I do.

I guess I'm coming at this from an operational standpoint. I like the zoom range of the 24-105 but its resolution ratings aren't that great at the long end. But I'm trying to see if this matters in real life.

Thanks for your response.

 

[joe mama]

Firstly, all of your questions are discussed in quite some depth here:

http://www.josephjamesphotography.com/equivalence/

Thanks so much for that link! Quite a bit of info that'll take a bit
of time to digest. Did you write that?

Yes, but I got a huge amount of help from people in these fora in creating it.

How come the blur plot of the 24-105 on the 5D is worse than the one
on the 20D? I assume it has something to do with equivalence?

http://www.slrgear.com/reviews/showproduct.php/product/145/cat/11

Honestly, I'm not 100% sure how their "blur index" works. I do know, however, that you would have to compare 24mm f/4 on 1.6x to 35mm, f/6.3 on FF. But, what's odd in their test is that the center at 35mm f/5.6 on FF is softer than 35mm f/4 on FF -- a very odd result to be sure.

I think what I meant to say was to compare the FF at 105mm to the
1.5x at 70mm. This would be the 24-105 at its worst and the 70-200 at
its best.

It's possible, of course, that one lens on one format may outperform another lens on another format. But I still don't know why you'd compare a 24-105 / 4L IS on FF to a 70-200 / 2.8L on 1.6x even though 70mm on 1.6x is close to 105mm on FF. You either want the 24-105mm range on FF (15-65mm on 1.6x) or the 100-300mm range (70-200mm on 1.6x). You should compare the equivalent lenses, e.g. the 24-105 / 4L IS on FF to the 17-55 / 2.8 IS on 1.6x, or the 100-300 / 4 on FF to the 70-200 / 2.8L on 1.6x.

2) FF is sharper in the middle 95% of the frame for the same FOV and
DOF

What if we use the same F-stop number instead but keep the same FOV?

FF will be sharper in the center (but not significantly so) and much softer on the edges with more vignetting. However, the softer edges are usually due as much, if not more, to the edges of the frame being out of the DOF as they are to the lesser performace of the lens in the outer range of its image circle.

I'm all for better high ISO performance. Is this just because of the
larger pixels? DOF I care less about.

It's odd that so many say they "care less" about DOF. The more shallow the DOF, the more blurry the parts of the image that are outside the DOF are going to be, so much less of an image with a more shallow DOF will be sharp than an image with a deeper DOF. And you seem like someone who cares about sharpness. : )

As for the higher ISO performance, it's primarily due to the fact that a FF sensor collects 2.56x as much light for the same perspective, FOV, and f-ratio. But, of course, if we use the same DOF and shutter speed, then all sensors, regardless of size, collect the same light, and then the only differences in noise will be due to the different efficiencies of sensors.

How different are sensors in terms of efficiency? Modern DSLRs all seem to be within a stop of each other, and likely even less.

What about the range of what "in-focus" means? Would it be easier to
make the AF system faster on an FF camera compared to a 1.5x camera
if the AF systems were the same?

That's beyond my expertise. You can start here and follow the rest of the thread for more info on that:

http://forums.dpreview.com/forums/read.asp?forum=1032&message=25148117

the differences between modern DSLRs matter more to
the photographer than they do to the non-photographer, if they matter
to the non-photographer at all. That said, I'm a photographer, and
the differences between the 20D and 5D are often striking to me for
the type of photography I do.

I guess I'm coming at this from an operational standpoint. I like the
zoom range of the 24-105 but its resolution ratings aren't that great
at the long end. But I'm trying to see if this matters in real life.

In truth, there will be little difference between the 40D + 17-55 / 2.8 IS and 5D + 24-105 / 4L IS. If DOF is unimportant to you, and that zoom range is your primary interest, then unless the extra range of the 24-105 / 4L IS is important to you (24-105mm on FF has the same FOV as 15-66mm on 1.6x), I'd recommend getting a 40D + 17-55 / 2.8 IS.

Thanks for your response.

Not a problem. I've already mailed out my bill. : )

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

 

[deeyay]

Thank you for the clarification!
--
Picture agency

• http://danieljansson.se

Web portfolio

• http://danieljansson.com

 

[robrone9]

There are lots of reasons why i like FF better than crop... and you have probably read them all. Less noise, big viewfinder, blah blah. But there are definite subjective qualities as well for many people. I can't tell you why exactly, I just have more fun on the wide end, and the 5D is awesome for that. I used to own and 12-24 lens and a crop camera, and it was never as fun as my 24-105 on the 5D is.

If I shot mostly birds, I would own a 40D, and use the tele more.

The differences in build quality don't wash with me, as I used a crop body for years in horrible conditions and never had a problem... current bodies are up to the same standards if not leagues ahead (even the pro-sumer level).

Try both, see what you like most, and use it. That simple.

ps. you could always split the difference, and get a 1.3 crop, or maybe (if you own next to no canon lenses) try out the D3 (it will do both).

--

 

[Rick Knepper]

Joe Mama oand others use the equivalency method to prove the point, and is quite valid. Requires one to think in 1 1/3 stop conversions.

For me, I found it quite useful to compare both formats on the same FL. Think of this as the unequivalent method. Anyway, at the following link are RAWs, a single scene, shot with various apertures on a 300D and 5D with the 35L.

http://www.box.net/shared/ct24xpam2x

 

[adethefoot]

I went from 300d to 350d to 30d to 5d

I used my 24-105L on all but the 300d

My opinion is the 5D stomps all over the crop cameras for image quality especially on A3 prints and upwards. They are more silky smooth and are definately sharper. And as for low light... thats where the 5D really shines.

If you can live with cleaning the image sensor on a regular basis and the below average LCD then its a bargain...

 

[landmarks4711]

In general (though not every time) you can trust what you read on
Luminous Landscape a lot more than what you read from Ken Rockwell...

what did Luminous Landscape say about full frame advantage?
(link?)

Just sayin'

Dan

--
---
G Dan Mitchell
SF Bay Area
http://www.gdanmitchell.com/

 

[landmarks4711]

basically yes bigger size of individual pixel from the same type of sensor (cmos) will have better IQ.

but some idiot stock website out there only required 12MP sensor and up for the photos, which indeed sounds idiot

 

[Jason Stoller]

If you search this forum, you will find that the Original 1D with only 4MP, equaled or beat out many cameras with more Mega Pixels. The Photo Sites on the sensor were Huge. This is exactly why I hung on to my 1DMKI even though I bought is successor, the 1DMKII as well. At this time I am also considering a 5D. My thinking is that the Current Model with its larger Photo Sites spaced well on a Full Frame sensor, will most likely rival the quality of its replacement. It does rival the quality of the 1DsMKII while the MKII has only a slight edge. I am sure I might give up something in focus speed and FPS in the new model, but the current 5D is already depreciated in price enough where it makes purchasing it a good decision. Once the New Model is announced and released, it will probably fall even more in price but I do not think it will be that significant. I can tell you from experience, that I paid around $5000 for my original 1D because I was an early adopter. I have made money with it and I do not regret buying it. It is now worth much less but I can still get over a grand for it.

Jason

It seems all very confusing.
However it is true that bigger pixel gives better image quality if
there are enough that bigger pixels to form a high resolution image.
Higher pixel count does not auotmatically resulting better image
quality.
The combination of the pixel number, sensor size, lens quality and
image processing determines the final outcome.
That is the reason a 12MP P&S camera could not compete its image
quality to an old EOS 1Ds with less pixels (11MP).

--
http://www.pbase.com/ltjiang

--
Jason Stoller [email protected]

We are just Beta Testers who pay the Camera Companies to test their new products!

 

[joe mama]

Joe Mama oand others use the equivalency method to prove the point,
and is quite valid. Requires one to think in 1 1/3 stop conversions.

Technically, it's 1.6x that "requires" one to think in 1 1/3 stop conversions, as lenses are labelled in terms of 35mm FF terms. : )

For me, I found it quite useful to compare both formats on the same
FL. Think of this as the unequivalent method. Anyway, at the
following link are RAWs, a single scene, shot with various apertures
on a 300D and 5D with the 35L.

http://www.box.net/shared/ct24xpam2x

Get this -- there are times such a comparison is indeed "equivalent". The max f-ratio of many Canon lenses is f/22. However, at f/22, a 1.6x DSLR can achieve the same DOF as f/32 on 35mm FF, but only a few lenses stop down that deep, so some argue that crop has an advantage for ultra deep DOF, just as FF has an advantage for ultra shallow DOF.

However, if we use the same lens on FF and crop out the 1.6x portion from the center, then we'll have the same image. "But," you say, "then the FF image uses only 40% of its pixels." Quite right! But here's the beauty of it: the effects of diffraction softening cancel out the advantage of the extra pixels, so you get an image with the same amount of detail regardless (assuming the cameras began with the same pixel count). Just how cool is that? : )

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

 

[peripheralfocus]

The simple fact of the matter is that it is the scene that matters
far more than anything else for the vast majority of people.

Yes. And understanding this -- really getting it rather than just paying lip service to it -- is a crucial part of learning how to be a better photographer. It can, for example, cause one to spend one's time looking at great pictures to try to get a feel for how light, color, composition, and human emotional content (facial expressions etc.) work in 2-dimensional representations, rather than spending one's time studying whether there is a 1/3rd or 2/3rds stop difference in noise performance between two cameras at ISO 3200.

This is not to dismiss equipment performance -- in a small minority of pictures it can be crucial -- but in the vast majority of pictures it's better to approach equipment performance as a necessary condition -- i.e. a certain minimum level of quality is required to prevent the picture from being ruined (at the presentation size you desire) but after that 97% of the emotional and artistic power of the photo inheres in the qualities of the scene (light, color, composition, light, and, oh don't forget: light).

Equipment questions then become issues of whether a given camera (or camera system) is designed in a way that works well for your particular needs, helping you to efficiently and pleasurably get those scene qualities into the kinds of pictures you take.

 

[gdanmitchell]

Iif you shoot the original image with a high quality lens that properly frames the subject, the FF image has the potential to hold up to larger print sizes before resolution becomes an issue.

I won't go into the details about why this is the case here, but it is really almost impossible to believe otherwise if you think about it.

There are other factors to consider, though, when purchasing equipment. For example, if you don't make very large prints any resolution improvement will probably be wasted. If you are unable to fill the frame with your subject on FF and end up having to crop, some or all of the advantage could be lost.

Dan

--
---
G Dan Mitchell
SF Bay Area
http://www.gdanmitchell.com/

 

[john5519]

Joe, beautiful shots in your galleries, I was also curious as to how you got the 3d raised effect in your framed photos, are they right up to the glass with some sort of clear archival tape?

A mix of 8x12s (the middle is a 12x18) from compacts, 300D, 20D, and 5D:

http://www.pbase.com/joemama/image/84366444