Corner sharpness: Four Thirds vs 35mm (bandwidth warning)

Started Oct 27, 2007 | Discussions
joe mama Forum Pro • Posts: 12,623
Questions:

The reason people say FF corners suck is because someone using an
f/1.4 lens on FF expects to actually use 1.4. And when they do, the
corners often do suck.

Question1: Can you produce a link to a single image at f / 1.4, where sharp corners are a consideration. Even one. I can produce literally hundreds of pics taken at f / 1.4, and even f / 1.2, on 35mm FF and corners don't matter on a single one of them. But that's just me. I'd be pleased to see someone else who takes pics at f / 1.4 where the corners matter. If the corners ever matter, and sometimes they do for me, I stop down. That's all there is to it.

True, there is often not even an option to do a comparison between the
equivalent FOV and DOF with a 4/3 lens, but it doesn't change the fact
that at that f-stop, the FF lens has rubbish corners.

Question 2: Can you produce a link to a single image where a sharp corner matters at a more shallow DOF? Because, f-stop for f-stop, 35mm FF has two stops less DOF than 4/3. In other words, can you link even one image where the FF shooter needed more DOF but could not stop down to achieve it?

The way I see it, FF and 4/3 have opposite problems. FF has the
ability to give very shallow DOF, but on many lenses the F-stops that
provide the shallow DOF also provide compromised IQ.

Question 3: Can you tell me a single reason that the FF camera cannot stop down to get the same DOF and thuse the same corners? I'll give you some help at this one: let's say you're at f / 5.6, ISO 1600 on 4/3. You'd need to be at f / 11, ISO 6400 on FF to get the same DOF and shutter speed, which requires shooting in RAW, underexposing by a stop, and overexposing by a stop in the conversion -- a royal pain.

So, I will ask you, can you produce a link to a single image shot under those conditions that would look better than the FF image using a more shallow DOF?

4/3 can give very deep DOF, but diffraction starts very early on in closing
down the aperture, so to get that super-deep DOF that you sometimes
need, you once again get compromised IQ.

No. Diffraction does not start early. Diffraction "starts" on all systems at the same DOF, even for smaller pixels. 'Tis true. Let me explain how that works: the smaller the pixels are, the more of them you have, so the more detail you capture. However, you will have more diffraction softening since the pixels are smaller. But that softening will exactly cancel the effect of more detail from having more pixels.

All systems, regardless of pixel size, suffer the same diffraction effects at the same DOF. It is another principle of equivalence.

fldspringer Senior Member • Posts: 1,427
Re: Questions:

joe mama wrote:

Question1: Can you produce a link to a single image at f / 1.4,
where sharp corners are a consideration. Even one.

You know, Joe, there are many situations that show up in photography that sometimes break with the expected. Sometimes the subject (or one of the subjects) don't end up in the center of the frame. I don't have one of the f1.4 lenses at my disposal, but here is one at f2.8 (wide open) with a subject in the corner. In this instance, I should have cropped a bit anyway, but I think you'll get the idea.

The problem with your previous threads is you came with an agenda. You played the role of a fan boy. You still are. That brings all sorts of others wanting to defend their system. The whole thing end in ... Well, jets just say the whole subject has filled numerous threads. Fans of their system are entrenched and we get nowhere.

Joe, your system isn't perfect. It isn't the perfect system for everyone. People can choose a different system for another reason than being a complete idiot.

Same thing applies to 4/3. There is no ideal system because all have tradeoffs. My issue with your previous threads is exactly that. Time to chill.

 fldspringer's gear list:fldspringer's gear list
Olympus E-3 Nikon D3S Nikon D500 Olympus Zuiko Digital ED 50mm 1:2.0 Macro Nikon AF-S Nikkor 300mm f/4D ED-IF +11 more
joe mama Forum Pro • Posts: 12,623
Nothing's changed.

Question1: Can you produce a link to a single image at f / 1.4,
where sharp corners are a consideration. Even one.

You know, Joe, there are many situations that show up in photography
that sometimes break with the expected. Sometimes the subject (or
one of the subjects) don't end up in the center of the frame. I
don't have one of the f1.4 lenses at my disposal, but here is one at
f2.8 (wide open) with a subject in the corner. In this instance, I
should have cropped a bit anyway, but I think you'll get the idea.

http://farm2.static.flickr.com/1134/755273795_4b58fe0874_b.jpg

I more than get the idea. For that pic, f / 2.8 on 4/3 -- the same DOF as f / 5.6 on 35mm FF -- is far from f / 1.4, don't you think? And, even then, are you going to say that the corners matter in that pic? The thumbnail is that important to the image?

The problem with your previous threads is you came with an agenda.
You played the role of a fan boy. You still are. That brings all
sorts of others wanting to defend their system. The whole thing end
in ... Well, jets just say the whole subject has filled numerous
threads. Fans of their system are entrenched and we get nowhere.

No, my problem is that people say I came in with an agenda. Link to one thread that supports that assertion. Link to a thread where I bashed 4/3. I have said, and said many a time, that FF has higher IQ. Amin's test pics prove that point to me. Did 4/3 beat the 5D in the extreme corners? You bet. Did the 5D win everywhere else? You bet.

Now, if you're arguing that those corners mean more to you than the rest of the image, then please specify that. I stand by what I've said, the FF has higher IQ, and I also stand by what I said, that the differences in those two images is so small that it doesn't mean jack for the vast majority of people.

Joe, your system isn't perfect. It isn't the perfect system for
everyone. People can choose a different system for another reason
than being a complete idiot.

Find one post where I've ever said different. Dude, do me just this one favor: at least read the conclusion of my equivalence essay:

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

People keep arguing with me saying I said things I never said, or implied, and saying things that I did say. For example:

Same thing applies to 4/3. There is no ideal system because all have
tradeoffs.

See?

My issue with your previous threads is exactly that.

My issue is that I never said what I've been accused of saying. Find it, link it, post it, and I'll apologize for it if I'm wrong. But no one has ever apologized for accusing me of having said, or even implied, things I didn't say or imply.

I have an agenda: the facts. I have a theory: FF has higher IQ. The facts I've seen bear out my theory. I have a philosophy (last paragraph of the conclusion of the essay):

"The bottom line is that we use a camera to create images. It is important to understand the advantages of any particular system as a whole, both in terms of IQ and operation. By understanding the principles of equivalence, we can understand the different capabilities of different systems, and then make an informed choice as to which system best meets our personal needs."

Time to chill.

Wouldn't you get tired of people not reading what you wrote, and saying you said and implied things you never said or implied? It was time to chill long ago.

fldspringer Senior Member • Posts: 1,427
Re: Nothing's changed.

Yes Joe, Nothings changed. Your still in the Oly forum selling 5Ds. Still have the same arrogant edge. Nothing's changed.

You certainly are not a man of you word. You said you'd book. Made it for how long? Till you were done with dinner?

I'm not supplying you with a link. I'm not supplying you with a pic again. You can, with all due respect, KMA

I know it won't make a difference, but you lens cap says Canon. Its a fine system. They have a forum too.

 fldspringer's gear list:fldspringer's gear list
Olympus E-3 Nikon D3S Nikon D500 Olympus Zuiko Digital ED 50mm 1:2.0 Macro Nikon AF-S Nikkor 300mm f/4D ED-IF +11 more
pidera Contributing Member • Posts: 925
OT : 14-42 versus 7-14 corner samples on cameralabs (link)

Rgds, Pieter

 pidera's gear list:pidera's gear list
Nikon D800E Fujifilm X-E1 Fujifilm X-Pro2 Fujifilm XF 16mm F2.8
joe mama Forum Pro • Posts: 12,623
OK.

Yes Joe, Nothings changed. Your still in the Oly forum selling 5Ds.
Still have the same arrogant edge. Nothing's changed.

OK.

You certainly are not a man of you word. You said you'd book. Made
it for how long? Till you were done with dinner?

'Till people stopped responding to my previous posts.

I'm not supplying you with a link. I'm not supplying you with a pic
again. You can, with all due respect, KMA

I asked three questions, you answered one of them wrong, you didn't answer the other two, and you didn't give me any links where I said or implied anything you said I said or implied. And then you tell me to KMA after you say it's time to chill. I'll leave it at that. Unless, of course, you want to drag it out further.

I know it won't make a difference, but you lens cap says Canon. Its
a fine system. They have a forum too.

My lens caps, all of them, say "Tamron" -- I prefer their center-pinch operation.

joejoecamlive New Member • Posts: 24
Re: Questions:

fldspringer wrote:

joe mama wrote:

Question1: Can you produce a link to a single image at f / 1.4,
where sharp corners are a consideration. Even one.

You know, Joe, there are many situations that show up in photography
that sometimes break with the expected. Sometimes the subject (or
one of the subjects) don't end up in the center of the frame. I
don't have one of the f1.4 lenses at my disposal, but here is one at
f2.8 (wide open) with a subject in the corner. In this instance, I
should have cropped a bit anyway, but I think you'll get the idea.

A Canon 5D shooter could have gotten that image, same DOF, at f4. I would probably shoot that at 50mm or longer, and it would certainly be sharp across the whole frame, if, all subjects are at the same distance. How about posting a full size so we can see if that bird is sharp?

pidera Contributing Member • Posts: 925
Re: OT : 14-42 versus 7-14 corner samples on cameralabs (link)

Retry :
http://www.camera
labs.com/reviews/OlympusE714mm/page4c.shtml

 pidera's gear list:pidera's gear list
Nikon D800E Fujifilm X-E1 Fujifilm X-Pro2 Fujifilm XF 16mm F2.8
SirSeth
SirSeth Veteran Member • Posts: 9,962
So what?

You would have to crop out the corners for printing anyway. I don't find it too distracting in the Canon shot. I would like to see a comparison between the 7-14mm Zuiko and any Canon lens with the same field of view (at 14mm on FF). I've heard that Olympus really has an advantage there.

But whatever, you buy these two cameras for reasons that are vastly different. I would prefer the E-3 to the 5D any day of the week, but that's my mileage, but corner sharpness as posted in this thread is no biggie to me.

Cheers, Seth

-- hide signature --

What if the hokey pokey really is what it's all about?

--
wallygoots.smugmug.com
wallygoots.blogspot.com

 SirSeth's gear list:SirSeth's gear list
Olympus E-1 Olympus E-M1 Fujifilm X-T1 Olympus Zuiko Digital ED 50-200mm 1:2.8-3.5 SWD Olympus Zuiko Digital 11-22mm 1:2.8-3.5 +5 more
dgrogers Veteran Member • Posts: 7,079
Re: FF corner test using my primes

Ominous wrote:

But why the fascination with corner sharpness?

For landscapes...often 50% of the corners is nothing but sky.

It's the other 50% that catches your eye (if you're the photographer, anyway).

If you are shooting people....I would hope you are not putting
someones face in the corner....talk about distortions!

There are cases where you may want to do that, and 4/3 lenses are controlled enough at the edges that you can without noticeable distortions.

If I'm shooting wide open...the corners are going to be out of focus
in nearly any shot....if I'm shooting stopped down...I don't seem to
have a problem keeping the corners sharp enough for anything I have
printed.

Then it seems like you got the camera that suits your shooting style.

-- hide signature --
 dgrogers's gear list:dgrogers's gear list
Olympus E-1 Samsung NX20 Olympus OM-D E-M10
slowsync Regular Member • Posts: 245
Re: more comparo

First of all, let me preface this thing by saying that I have no interest in slamming either system (specially since like amit, I own both). I use each for slightly different purposes and they both complement each other to fit my needs, however, my needs are probably different than yours so you will need to come up with your own conclusions. All I did with the following was to set both on the same tripod, click, crop the relevant portions and present them here.

So here goes another half-ass*d comparison. Unfortunately, in my rush to get out, only 1 camera had a CF card that was already mostly filled so I had a limited number of shots.

Tripod, EV metering on both, focused on whatever the AF decided to hit, ISO100, 2 second timer before the shot. I tried not to move the tripod but didn't bother trying to get both images exactly aligned. I have no control over, wind, changing lights and clouds, insufficient ingestion of warm caffeinated drinks so for all I know these tests are completely invalid--take them with a grain of salt.

Here is the general scenery, first E330.

http://members.dslextreme.com/users/slowsync/images/Oly_1122_F8 (GMB@12mm).jpg

Next, the 5D. Both the small photos have been sharpened in PS after downsizing.

http://members.dslextreme.com/users/slowsync/images/Canon_24105_f8 (GMB@24mm).jpg

Both were developed in silkypix, using the same default parameters for both. I have not bothered trying to get WB or anything else to match. Both are at f8. No PP was done for the crops.

Full crops, first one is E330 with 11-22 at 12mm and 5D with 24-105 at 24mm. The 11-22 is 1mm away from it's widest end and the 24-105 is at its widest setting.

http://members.dslextreme.com/users/slowsync/images/12_24mm (GMB).jpg

Second one is E330 with 14-54 at 14mm and 5D with 24-105 at 28mm. This time, the 14-54 is at its widest settings and the 24-105, at 28mm, is not.

(I don't know why the detail portion of the 5D at F4 is blurry, probably the AF picked up the trees).

So while there is more smearing at open apertures with the Canon, I don't think it's as bad as everyone says it is and certainly stopping down some will reduce it to virtually non-existent levels.

What about primes you say? Good question, however, (a) I have strong preference for zooms ( my needs), (b) I only have one prime lens, (c) I did not take any test shots with. Maybe next time or maybe never.

And since we are comparing, we might as well compare noise too (oh noes). The following is a shot of GMB color checker (at f8 to avoid vignetting issues). I zoomed in to fill the frame and focused at infinity (the image is defocused which is how noise tests should normally be done to avoid picking up detail). Manual exposure, spot metered on the 18% gray chip. ISO was increased for every shot and shutter speed was adjusted the corresponding amount. IIRC, both cameras metered the same so they are using the same exact f-stop, shutter speed and ISO's for each comparable frame. NR was completely turned off in silkypix. I should have probably turned off sharpening too but did not, however since they are both sharpened the same amount, it should not make any difference.

My personal observation regarding noise is that both are comparable up to (or close to) ISO400. Above that, well, that's one of the reasons I have the Canon, however, unlike some of the posts I've seen, I don't think it's a two stop difference, maybe 1.5 or so at the high end.

My conclusion from all this--nothing much really, in the end, they are both fine systems and at my usual working ISO's and apertures, both of them should do well.

One last thing, all shots were taken in raw. I don't have enough space to host them but if anyone is crazy enough to volunteer, I can send you all 250GB raw files for linking (or selected ones).

flame away (and fill up this thread)

paulbod2 Senior Member • Posts: 1,004
Please test the 3G

Till you try to use a lensbaby or a viewcamera, you do not have accumulated the technical expertise to discusss in depht the lenses properties. I beg you to try the Lensbaby 3G, it is easier to use than its ancestors.

http://www.lensbabies.com/index.php?r=GSLensbaby&gclid=CKzrxdi2so8CFRoGEgod9QIXOg

Thisjustin Senior Member • Posts: 1,084
TROLL!

No text.

Thisjustin Senior Member • Posts: 1,084
Wow, imagine that...

The noise issue is not all its hyped up to be? who knew!?

The corner detail issue is more than certain fan boys what you to believe? who knew!?

All sarcasm aside, thank you. I think ill store away those tests for a...rainy day...

Thisjustin Senior Member • Posts: 1,084
Thats funny, i thought you were on a crusade against 4/3s? n/t

No text

Lee Jay Forum Pro • Posts: 54,518
Re: more comparo

slowsync wrote:

And since we are comparing, we might as well compare noise too (oh
noes). The following is a shot of GMB color checker (at f8 to avoid
vignetting issues). I zoomed in to fill the frame and focused at
infinity (the image is defocused which is how noise tests should
normally be done to avoid picking up detail).

When you do noise testing, it should be done at constant final size and constant detail preserved. The difference between 7.4MP and 12.7MP will give you an additional ~ 1.5 stops of noise performance at the same size and detail on top of the per-pixel difference.

5D versus 20D at the pixel level, 1/3 stop improvement, but the 5D shot has way more detail:

http://photos.imageevent.com/sipphoto/samplepictures/20D%20versus%205D%20detail.jpg

5D versus 20D at constant detail and size, 1 1/3 stop improvement:

http://photos.imageevent.com/sipphoto/samplepictures/20D%20400%20versus%205D%201000.jpg

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
Amin Sabet
OP Amin Sabet Veteran Member • Posts: 6,763
Re: more comparo

Very nice work. Thanks for posting this. Some specific comments below.

Full crops, first one is E330 with 11-22 at 12mm and 5D with 24-105
at 24mm. The 11-22 is 1mm away from it's widest end and the 24-105
is at its widest setting.

I'd love to see the Canon shot at f/16 if you have it.

Second one is E330 with 14-54 at 14mm and 5D with 24-105 at 28mm.
This time, the 14-54 is at its widest settings and the 24-105, at
28mm, is not.

Here if we compare the Canon at f/16 to the Olympus at f/8 (same DOF), the Oly has comparable corners to the Canon, but the Canon is sharper elsewhere. This could be taken two ways. One way to take it would be that the Canon is equal or better everywhere. The other way to take it is that the Oly has more consistency across the frame. One thing I'd point out is how smeary the 24-105L extreme corners are at 28mm and f/8. I went back to some of my old photos with the 24-105L, and I have to retract my comments about the 28/2.8 outperforming that zoom at the wide end. Still, the 28/2.8 holds its own pretty well IMO, getting only modestly outdone.

So while there is more smearing at open apertures with the Canon, I
don't think it's as bad as everyone says it is and certainly stopping
down some will reduce it to virtually non-existent levels.

Good enough for you and me.

My personal observation regarding noise is that both are comparable
up to (or close to) ISO400. Above that, well, that's one of the
reasons I have the Canon

Interesting findings.

My conclusion from all this--nothing much really, in the end, they
are both fine systems and at my usual working ISO's and apertures,
both of them should do well.

Same sentiments here.

 Amin Sabet's gear list:Amin Sabet's gear list
Nikon D810 Nikon D500 Panasonic Lumix DMC-GX85 Sony a9 Panasonic Lumix G 14mm F2.5 ASPH +16 more
Lee Jay Forum Pro • Posts: 54,518
Re: more comparo

Amin Sabet wrote:

One thing I'd point out is how smeary the 24-105L extreme corners are
at 28mm and f/8.

That also surprises me. I did a test of the 24-105L on the 5D using every 1/3 stop and the very worst corner of the very worst image (upper right, 24mm, f4) is posted here at 100% crop from the 5D:

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
thw Veteran Member • Posts: 8,089
Some comments

i) 24-105 f/4 has a lot of sample variation. Adam-T and Lee Jay have good working copies which are sharp from edge to edge. If you find yours lacking in this regard, you should get it fixed.

ii) In general though, FF lenses have more corner softness

iii) Olympus DSLRs have poor per pixel sharpness, no doubt about that. Confirmed in photozone quite a while back.

iv) Older Olympus 4/3rd sensors have poor high ISO performance.

v) Glad Nikon users finally get their FX cameras or they'll be here to criticize the weaknesses of FF too.

-- hide signature --

See the colors of my world in:
thw.smugmug.com

Xargo Contributing Member • Posts: 850
Re: more comparo

Nice test. I found the noise comparison informative. Thanks for that!

slowsync wrote:

(I don't know why the detail portion of the 5D at F4 is blurry,
probably the AF picked up the trees).

I think this makes the test more like an extreme corner bokeh test than anything. Clearly if the DOF is not even deep enough to bring the trees and the wall in focus at the same time, the frontal pavement must be severely OOF as well.

Inconsistent focusing makes it impossible to do sharpness comparison between the photos.

-- hide signature --

DMC-L1, E-330, 11-22mm, 14-50mm Vario-Elmarit-D, 25mm Summilux-D, 180mm Apo-Telyt-R, 250mm Telyt-R, 60mm Macro Elmarit-R, 135mm Elmarit-R, 90mm Elmarit-R, 35mm Summicron-R, FL36, EC-14, 2x Leica-R extender

Keyboard shortcuts:
FForum MMy threads