Another Olympus E-M1 and SONY a7 walkabout test, high ISO

Started 8 months ago | Discussions thread
blue_skies
Senior MemberPosts: 7,044Gear list
Like?
Re: Thanks for the truth!
In reply to Anders W, 8 months ago

Anders W wrote:

blue_skies wrote:

Anders W wrote:

blue_skies wrote:

Anders W wrote:

blue_skies wrote:

Lab D wrote:

blue_skies wrote:

Lab D wrote:

blue_skies wrote:

Lab D wrote:

skyglider wrote:

blue_skies wrote:

skyglider wrote:

blue_skies wrote:

How so: in equivalence terms, the Oly 12-40/f2.8 becomes 24-80/f5.6 versus the Sony 24-70/f4.0?

Hi Henry,

I understand how the Oly 12-40mm is equivalent to 24-80mm in 35mm terms because of the 2x crop factor of the micro 4/3 sensor. But AFAIK, the aperture remains the same.

IOW, if an Oly 12mm lens could be used on a full frame (35mm) sensor, only the center 1/4 of the sensor area would be illuminated which is equivalent to 1/2 the horizontal and vertical pixels. Thus the crop factor of 2x. But the aperture remains the same so the same amount of light is hitting that portion of the sensor by the F2.8 aperture. So the aperture does "not" become equivalent to a F5.6 aperture.

Is my understanding incorrect?

Thanks,
Sky

Hi Sky, no your understanding is correct.

In terms of ISO, aperture and shutter speed, the exposure is the SAME on the small and the larger sensor.

If you are after (fast) shutter speed, you only care about the lens speed, and your thinking is correct.

But in terms of noise, the larger sensor allows to operate at a much higher ISO level and produce the same amount of noise: in m43/FF this is a 4x ratio, or 2 stops. This means that eg. m43 at ISO 1,000 is as noise as FF at ISO 4,000.

So, if you want to produce comparable images, you would shoot the FF image with two stops slower aperture and at two stops higher ISO. This is equivalent exposure.

Ironically, the same metric applies to DOF - you roughly see about the same DOF in both images as well.

Thus, shooting the Oly at 12mm f/2.8, ISO 1,000 as 1/60th is equivalent to shooting the Sony at f/5.6, ISO 4,000 and 1/60th. Aside from the resolution, the two images will appear similar.

If you use the same exposure on both, you are comparing the FF at ISO 1,000 versus the Oly at ISO 1,000. The FF image will have a lot less noise, ie. be much cleaner, but the DOF will also be reduced. The images are not equivalent at that point.

On FF you can trade noise for DOF, assuming you have a fast lens. There is a 2 stop delta to play with.

In the example, using the Sony lens at f/4.0 implies that the ISO goes to 2,000. This is one stop below the equivalent values of f/5.6 and ISO 4,000, so, with more shallow DOF, you will have a less noisy image on the FF.

-- hide signature --

Cheers,
Henry

Hi Henry,

I think I understand your point now. I wondered how many stops difference there was between a micro 4/3 sensor verses a full frame sensor. Now I know it's a 4x ratio or 2 stops difference.

That only applies for Depth of field. If you check the results from DxO tests, the difference is much less. A good example is dynamic range where the E-M1 is only about 1/2 stop different than the A7 for most of the ISO settings DxO tested.

Remember if you need F/2.8, 1/60th sec shutter speed, and ISO800 on the E-M1, you will need the SAME settings on an A7, a Nikon 1 and a Sony RX100. It would even be the same on your smartphone!

Again, exposure = exposure and equivalence = equivalence.

And you know in this case equivalance does not tell us the facts. Admit it, there is NOT a 2 stop difference for noise or dynamic range at almsot all the ISOs DxO tested.

Just admit the truth for once.

Henry showing the difference between sensors

-- hide signature --

Cheers,
Henry

You are right with the DxO link. The difference is really negligible. There is not a 2 stop difference almost all of time here.

?????

Please go take ISO 6400 pictures with your EM1 then

You conveniently deleted that I stated that it is immaterial whether it is 2x, or 1.65x, or 1.5x stops.

Where did you state that? And of course it isn't immaterial. If it's less than two stops (as it certainly is), MFT has an advantage with regard to signal-noise performance for equivalent images (same DoF, same shutter speed, different exposure, different ISOs).

http://www.dpreview.com/forums/post/53010416

"If DxO points out that it is 1.65x stops or 1.5x stops, it does NOT CHANGE the general thoughts"

It certainly does. The facts are as follows:

1. With respect to exposure, that is, the amount of light falling on the sensor per area unit (e.g., per square millimeter), f/2.8 on MFT is equivalent to f/2.8 on FF.

2. With respect to the total amount of light falling on the sensor, DoF, and diffraction, f/2.8 on MFT is equivalent to f/5.6 on FF (assuming, with regard to DoF, that the AoV and the focus distance is the same).

3. The total amount of light falling on the sensor is not the only determinant of signal-noise performance. What additionally matters is the quantum efficiency and the read-noise performance. In these regards, smaller sensors tend to be more efficient than larger sensors, as illustrated, for example by the graph posted by Lab D. Consequently, the difference between the A7/A7R and the E-M1 with regard to signal-noise performance at the same exposure is less than two stops.

You consistently overlook point 3 in spite of the fact that it has repeatedly been pointed out to you for months.

Not at all Anders, you have elaborated on this in the past,

Nevertheless, you overlook it time and again.

and the difference, even per DxO, they rate the ISO performance for the cameras as EM1/A7/A7r as 757/2248/2746. Two stops would have been 3028. So we all agree?

Not quite. While these figures too show the difference to be less than two stops, they underestimate the actual advantage of the E-M1 for equivalent images. The DR curve posted by Lab D provides a better indication with regard to shadow noise, which in turn is the most apparent problem in high ISO shooting. What the curves indicate is that the E-M1 is about 1.5 stops ahead of the A7/A7R for equivalent images (DoF held constant) and about 0.5 EV behind at the same exposure (and different DoF).

Sure, MFT has a 'technical advantage' but the image is worse - my eyes don't deceive me - download Daniel's newly uploaded full-image side-by-side and see for yourself.

Your eyes may not deceive you but your brain certainly does. Daniel's images are not equivalent images (see definition above). Rather they are images shot at the same exposure, in which case the A7 can be expected to do better with regard to signal-noise performance (although less than two stops better) but at the expense of two stops less worth of DoF.

And this is my argument: users shoot alike Daniel, and this produces images that may or may not be equivalent

An argument for what? It is certainly not a relevant objection to the point Lab D and I were making.

- even in Daniel's case he tried to match FL, aperture and framing as best as he could. Yet the resulting images speak for themselves.

As a rule (his practice was not entirely consistent) he matched f-stop but as we know that does not give equal DoF. If he had chosen to shoot equivalent images, as he might well have done, the outcome with regard to signal-noise performance would have been different.

It is immaterial, imho, as I would not use a smaller sensor camera at higher ISO. Why not? Because the image quality degrades too much. I would rather use longer exposure and lower ISO, if possible.

When you are in a position to freely increase the length of the exposure (because the scene is static), you are also free to bracket exposure and merge/align multiple images. This makes sensor performance a moot point.

When you are not in a position to do so, MFT has the advantage with regard to signal-noise performance for equivalent images.

See the image above, sure, it can be improved upon, but as Daniel said "the way a user would do it". Clearly the images are not similar in IQ, neither at ISO 100 nor at ISO 3200.

Again, the images above are not equivalent.

They are not, and I expected the high ISO image to perform better on the larger sensor. Again, read noise and pixel noise are not the same, and the larger sensor has larger pixels.

I don't think anyone has different expections than you in that regard. If you give the larger sensor the same exposure, it will have an advantage with regard to signal-noise performance at the expense of DoF.

But the low ISO image surprised me, I would have expected the ISO 100 images to be much closer to each other in IQ than they in fact are.

This is a scene with extreme dynamic range. Here's how to do such a scene better from a signal-noise point of view than either an E-M1 or an A7/A7R can manage in a single shot (nine shots with the E-M5 and the 12/2 at f/4, with shutter speeds varying from 1/8 s to 30 s, base ISO, merged and aligned in PP).

And we have had this discussion so many times already. Why the nitpicking?

Because you are not admitting the truth.

What truth?

The truth contained in my point 3 above.

Which shows that the ratio is not proportional to sensor size. It still confirms that larger sensors perform better.

No it doesn't. Rather it confirms that the smaller sensors do better from a signal-noise point of view when DoF is held constant. In order to do better from a signal-noise point of view, the larger sensor has to sacrifice DoF.

Let me explain: per your point, an 'equivalent' exposure on the smaller sensor would be of higher IQ. But the larger sensor allows more noise/DOF trade-off which results in a higher IQ image (but non-equivalent), at the cost of shallow DOF (which may or may not matter).

As far as I am concerned, DoF is just as much a part of IQ as is signal-noise performance.

I thank Daniel for uploading the images - they speak for themselves.

Only if you know how to interpret them correctly and you just showed you didn't.

Which image would you hang on the wall then? Perhaps not equivalent, but one camera is cleanly cleaner.

And the other clearly has better DoF. In most of the sample images, the DoF provided by the A7 is too shallow based on my preferences.

Technical slides do not tell this story, and in the analysis we get lost in sensor efficiency, not image IQ.

"Technical slides" (I'd call them diagrams) do tell the story as long as they reflect relevant and accurate measurements (as they do in this case). And sensor efficiency has implications for image quality as indicated by my point 3 above.

Yes, and you repeat this as often as you can. But to my point - show side-by-side images, let them speak. We have side-by-side images here to point at, and I would definitely pick the larger sensor camera.

I have no objections to looking at images side by side. But those we got here do not show all the relevant comparisons and my choice between those we can see would be different than yours. As I said, I find the DoF of the A7 images too shallow in most cases (1st, 2nd, 4th, and 5th pair). Pair 3 is a failure (wrong focus on the MFT image) and for pair 6, DoF is irrelevant.

Change the venue and shoot people walking at low ISO, I would expect the smaller sensor camera to pull ahead.

On what grounds do you think so?

Like I said, both cameras have pros and cons, but low light & high ISO (and shallow DOF) are not the smaller sensor's strong points.

The point here is rather that the smaller sensor is ahead in low light/high ISO whenever DoF is in short supply.

Lower the ISO, and both cameras are great. They both have their pros and cons.

They both have their pros and cons, but it is important to be factually correct about what those pros and cons are.

See my comments here: http://www.dpreview.com/forums/post/53009228

Pros and cons are subject to an individual bias.

The facts are objective, the evaluation of those facts is not.

Then show us how to shoot the M1? Obviously, users don't know how to.

What specifically makes you think they don't?

I like the image that you included. But I am not that fast in rejecting an image because of shallow DOF - there are (lots of) times when it works, but not always.

As to this debate - 32 posts and counting merely because I (dared) compared the 4x sensor difference to roughly 2 stops? And perhaps also because I dared looking at actual images, rather than the theoretical perfect ones?

Ok, I get it, you guys are relentless, and will never stop.

Can we call this a dead horse now?

-- hide signature --

Cheers,
Henry

 blue_skies's gear list:blue_skies's gear list
Canon PowerShot S95 Sony Alpha NEX-7 Sony Alpha NEX-6 Sony Alpha 7 Sony a6000 +30 more
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
NiceNew
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow