Dynamic range of mft cameras

[Norwegian wood]

I know too little about the mft system, but Ive snappet up that a number of users have issues with burned out images when photographing in conditions with high contrast areas.

As this might be a deal braker to go to mft systems, I started wondering if this is an issue with full bodied four third systems too? I mean, they share the same sensor size doesnt they? Is this a problem related to poor quality mft or 4/3s sensors compared to APS C sensors?

Are there differences between mft and 4/3s regarding burned out images? Why is that, if they use the same sensors? Or do the latest mft sensors have less quality than the earlier 4/3s ?

I would happy read any posts that can enlighten me in this matter.

 

[liviutza]

Hi,

first of all, there is no "quality issue" regarding m43 sensors vs. APS sized ones. There is a matter of pixel size, with APS sensors having larger ones. For the same sensor type, pixel size is an essential factor in DR, so this is one area where theoretically m43 (or 4/3) sensors have lower results.

Actually, the size factor aside, latest developments in 4/3 sensors make them very close, or even superior to APS sensors, for their size. Check out reviews on this site, and you will find that while in absolute terms the DR is still not exactly equal to APS sensors, it is very close - closer than the size difference would suggest. And I have seen magnificent shots taken in difficult situation with such "smaller" sensors - there is also the matter of getting used to the camera, and what we have now in stores is some outstanding hardware.

Indeed, at this point you can get slightly better results for the same money with APS sensors cameras, and in DR terms this should remain the same. However, if Fuji enters the m43 game, we will probably have results to equal "FF" conventional sensors in DR, albeit at half resolution.

 

[Bilgy_no1]

Hi there,

I think the differences between cameras are more to do with processing options than with inherent sensor issues. The E-620 was reviewed to have better DR than Canon 500D and Nikon D5000 (in jpeg), while the E-P1 has similar DR to the 500D according to this website. On the contrary, the E-P1 has better noise performance and higher resolution results. Yet, they use the same sensor.

Damien

I know too little about the mft system, but Ive snappet up that a number of users have issues with burned out images when photographing in conditions with high contrast areas.

As this might be a deal braker to go to mft systems, I started wondering if this is an issue with full bodied four third systems too? I mean, they share the same sensor size doesnt they? Is this a problem related to poor quality mft or 4/3s sensors compared to APS C sensors?

Are there differences between mft and 4/3s regarding burned out images? Why is that, if they use the same sensors? Or do the latest mft sensors have less quality than the earlier 4/3s ?

I would happy read any posts that can enlighten me in this matter.

--
http://www.flickr.com/photos/bilgy_no1

 

[sean000]

I know too little about the mft system, but Ive snappet up that a number of users have issues with burned out images when photographing in conditions with high contrast areas.

I would have the same issues with my Nikon DSLR APS-C Sensor cameras if I didn't adjust exposure (or exposure compensation) under contrasty conditions. If you browse DSLR forums you will see plenty of images with blown highlights and/or dark shadows. Even full frame DSLR shooters sometimes have to use exposure blending/HDR, graduated neutral density filters, or just a little dodging and burning in Photoshop. I try not to worry about it too much, considering I have taken successful photographs in the past with compact P&S digital cameras that had far less DR than a m4/3 camera.

As this might be a deal braker to go to mft systems, I started wondering if this is an issue with full bodied four third systems too? I mean, they share the same sensor size doesnt they? Is this a problem related to poor quality mft or 4/3s sensors compared to APS C sensors?

Are there differences between mft and 4/3s regarding burned out images? Why is that, if they use the same sensors? Or do the latest mft sensors have less quality than the earlier 4/3s ?

So much can affect the apparent dynamic range. You can see more scientific comparisons of DR by reading the reviews on this site for cameras like the GF1. But the photographer bears some responsibility for avoiding blown highlights or shadows that are too dark. Camera meters can be fooled for one. The exposure compensation feature, as well as different metering modes, exist to help the photographer get more accurate exposure out of the camera. Sometimes this requires reviewing your shots and making adjustments when the histograms or reviews tell you that the highlights were blown out or the shadows blocked up.

The tonal curve used by the camera to output a JPEG can also affect the apparent DR. Shooting RAW or using a conservative (less punchy) tonal curve will help, but sometimes a scene is just too contrasty for any camera to capture in a single take or without help from something like a graduated neutral density filter or fill flash. Even with these options a scene can just been too much without blending multiple exposures using HDR techniques or layer masks.

Sometimes you have to make the decision to either wait for better light with less contrast (harsh noontime sunlight that creates bright highlights and dark shadows is some of the worst light to shoot in anyway) or decide what parts of the scene are most important to expose correctly.

I don't feel like my GF1 is severely limited compared to my APS-C DSLR. It's way better than the Panasonic LX2 I used to carry as my portable camera. Here is a shot taken on an overcast day (conditions that often make it difficult to get a good exposure without blowing the sky) with the GF1 and 20mm pancake on Saturday:



Not a great photo, but I shot from the hip while walking my dog in the park. deliberately had the exposure comp dialed back since I didn't want the sky blown out. Of course the grass was underexposed by at least a stop or two. With only -1/3 exposure comp dialed in the exposure was as bright as it could be without blowing the sky. I used the graduated neutral density tool in Adobe Camera RAW (no replacement for a real graduated ND filter, but fine for small adjustments) to keep the sky and water in check while I quickly brightened up the rest of the photo. This is a crop by the way. I would have had to do the same adjustment, perhaps to a slightly lesser degree, in post-processing if I had taken this photo with my DSLR.

Sean

 

[TorsteinH]

Actually the E-620 and E-30 has better dynamic range than many modern APS-C cameras. The same goes for E-P1 and 2. Just check the DR on the reviews here at dpreview.

Hi,

first of all, there is no "quality issue" regarding m43 sensors vs. APS sized ones. There is a matter of pixel size, with APS sensors having larger ones. For the same sensor type, pixel size is an essential factor in DR, so this is one area where theoretically m43 (or 4/3) sensors have lower results.

Actually, the size factor aside, latest developments in 4/3 sensors make them very close, or even superior to APS sensors, for their size. Check out reviews on this site, and you will find that while in absolute terms the DR is still not exactly equal to APS sensors, it is very close - closer than the size difference would suggest. And I have seen magnificent shots taken in difficult situation with such "smaller" sensors - there is also the matter of getting used to the camera, and what we have now in stores is some outstanding hardware.

Indeed, at this point you can get slightly better results for the same money with APS sensors cameras, and in DR terms this should remain the same. However, if Fuji enters the m43 game, we will probably have results to equal "FF" conventional sensors in DR, albeit at half resolution.

--
http://www.ohb.no/foto
************
Torstein

 

[sean000]

Actually the E-620 and E-30 has better dynamic range than many modern APS-C cameras. The same goes for E-P1 and 2. Just check the DR on the reviews here at dpreview.

This is certainly true at higher ISOs, but at ISO 100 I would say they are almost comparable. I say almost because most APS-C sensor cameras have more DR at the highlight end vs less at the shadow end. The larger sensor cameras also tend to have a little more headroom for recovering detail at the highlight end. I think the highlight end is noticed more and seems to be the area of concern for most people. Of course your mileage may vary depending on whether you shoot JPEG (and which settings you use) or RAW (and which converter and post processing techniques you use).

Regardless of what camera you use you must watch your histograms and adjust your composition and/or shooting technique to compensate.

 

[steveblack]

Somebody else posted this here before. This link compares Dynamic range of GH1, D7 and 5D MKII. GH1 is compares very nicely and infact at lower ISO, is better than 7D.

http://www.dxomark.com/index.php/eng/Image-Quality-Database/Compare-cameras/%28appareil1%29/337 |0/%28appareil2%29/329|0/%28appareil3%29/305|0/%28onglet%29/0/%28brand%29/Panasonic/%28brand2%29/Canon/%28brand3%29/Canon

 

[Jay Turberville]

For a given level of detail, a 4/3s sensor - all other things being equal - gives up a half a stop or slightly more of dynamic range to an APS sensor and it gives up two stops - for the same level of detail - compared to a 36mm x 24mm sensor. Actually, that's a bit of an oversimplification. The guys that know the physics better than I do can wade in and be more precise if they want. I'm probably giving a bit too much of a theoretical advantage to the larger sensor. But that's a decent "broad strokes" view.

The whole bit about comparing at the same level of detail is important. Many people will get fixated on things like pixel well size and end up comparing dynamic range on a "per pixel" basis. This leads to wrong conclusion. You reduce noise by giving up detail. It works. I've tested and measured it.

You have to be very careful about things like "headroom" and so forth. Digital sensors are very linear for most of their useful response. The whole headroom thing often has more to do with the tone curves chosen by a manufacturer than with anything else. Raw headroom has to do with default ISO assignments and what particular RGB filters are used on the sensor.

It has been a while since I looked at DPReviews DR tests, but the last time I did, I thought they were pretty poor. You really have to understand the issues to get a good picture of real sensor performance from them.

But getting back my original point that the smaller sensor has an inherent disadvantage given the desire to have the same level of detail, this is true if "all other things are equal." But if things aren't otherwise equal, that inherent limitation might widen or narrow.

As a practical matter, my initial experiences with my GH1 suggest that it is an overall better performer than the sensors in any of my previous 4/3s cameras. I've very happy with it. If I really needed great low light performance or extra wide DR, I'd look for a sensor larger than an APS-C. I'd especially look for a very new/modern design. It seems like the smaller sensors often get the best new technology first and that the tendency is to narrow the inherent gap rather than widen it. This is because they typically sell a lot more smaller and mid-sized sensors than those big 24mm x 36mm ones.

--
Jay Turberville
http://www.jayandwanda.com

 

[Robbie_EP1]

DXO also say that a LX3 has better dynamic range than an e-p1 and G1. Are you going to believe that? DXO just does not have any (or very little) real world accuracy. You will find I'm not alone in my thoughts of DXO.
--
http://www.flickr.com/photos/Robbie_e-p1/

 

[steveblack]

DXO also say that a LX3 has better dynamic range than an e-p1 and G1. Are you going to believe that? DXO just does not have any (or very little) real world accuracy. You will find I'm not alone in my thoughts of DXO.
--
http://www.flickr.com/photos/Robbie_e-p1/

I am a newbie and have very little to add on my own. I just pointed to a link that I found interesting. I would leave it to DXO to prove their credibility.

Thanks the warning though, I will be careful with them.

 

[Norwegian wood]

Thanks all! A lot of great and explaining answers here.

I believe from the answers that 4/3ds and mft have about the same issues to DR, but its not really too poor compared to APS sized cameras.

As a currently owner of a D70 (APS sized sensor) Im quite happy to know that some of you guys put the DR of the GF1 on par with APS sized cameras. Still I really would like my new camera to give me a bit improved "headroom" over the D70, afterall the D70 have a some years behind it. I would think that more than 5 years of sensor/camera development should have resulted in improved DR friendly sensors/cameras.

Taken the size and usability of a nice GF-1, it is still tempting to go for one. Maybe I will, or maybe I will wait until the 2nd edition is here. But sure thing is that the size of that camera paired with a reasonable sized sensor and interchangeable optics is a winner. Just crazy they didnt arrive earlier!

 

[jim stirling]

Actually the E-620 and E-30 has better dynamic range than many modern APS-C cameras. The same goes for E-P1 and 2. Just check the DR on the reviews here at dpreview.

I think that is only in relation to jpeg output
Jim

 

[Jogger]

• the best smaller than FF sensor is the 12mp CMOS in the D300s
• DR is very close... ONLY at the lowest ISOs... bump up the ISO and m43 loses big time in DR

I know too little about the mft system, but Ive snappet up that a number of users have issues with burned out images when photographing in conditions with high contrast areas.

As this might be a deal braker to go to mft systems, I started wondering if this is an issue with full bodied four third systems too? I mean, they share the same sensor size doesnt they? Is this a problem related to poor quality mft or 4/3s sensors compared to APS C sensors?

Are there differences between mft and 4/3s regarding burned out images? Why is that, if they use the same sensors? Or do the latest mft sensors have less quality than the earlier 4/3s ?

I would happy read any posts that can enlighten me in this matter.

 

[Knight Palm]

Being old technology, hence you'll find the D70 at the bottom of the DxOMark data. Any Panasonic sensor put into 4/3 cameras measures far better than the D70.

Sensor technology is constantly improving, and the smaller the sensor, the higher the volume it sells in. Compare the sensor size in the Haselblad with an iPhone.

So don't worry that much over sensor sizes, since there's no need for a photo-viagra.

The latest generation sensors are all pretty good, and the camera performance between the different sensor sizes are constantly shrinking.

Thanks all! A lot of great and explaining answers here.

I believe from the answers that 4/3ds and mft have about the same issues to DR, but its not really too poor compared to APS sized cameras.

As a currently owner of a D70 (APS sized sensor) Im quite happy to know that some of you guys put the DR of the GF1 on par with APS sized cameras. Still I really would like my new camera to give me a bit improved "headroom" over the D70, afterall the D70 have a some years behind it. I would think that more than 5 years of sensor/camera development should have resulted in improved DR friendly sensors/cameras.

Taken the size and usability of a nice GF-1, it is still tempting to go for one. Maybe I will, or maybe I will wait until the 2nd edition is here. But sure thing is that the size of that camera paired with a reasonable sized sensor and interchangeable optics is a winner. Just crazy they didnt arrive earlier!

--

Ludwig Wittgenstein; British philosopher born in Austria; a major influence on logic and logical positivism (1889-1951):

“What can be said at all can be said clearly; and whereof one cannot speak thereof one must be silent”

 

[jim stirling]

Being old technology, hence you'll find the D70 at the bottom of the DxOMark data. Any Panasonic sensor put into 4/3 cameras measures far better than the D70.

I think the D70 holds up pretty well for a camera that came out 6 years ago, here is the dxo comparison and far better seems a bit generous
Jim

 

[sean000]

• the best smaller than FF sensor is the 12mp CMOS in the D300s
• DR is very close... ONLY at the lowest ISOs... bump up the ISO and m43 loses big time in DR

According to the tests on this website (in-depth reviews) the EP-1 can capture 9.1 EV at ISO 200, while the D300s captures 8.4 EV. They are closer at ISO 1600 and 100... so there isn't really a high ISO advantage to the larger sensor other than controlling noise better.

The difference between the m4/3 cameras and the APS-C cameras seems to be that the m4/3 cameras offer more range at the shadow end, and APS-C cameras offer more range at the highlight end. I guess this is why people (including myself) think of APS-C DSLR cameras as having more dynamic range... because people tend to notice blown highlights more than dark shadows. The larger sensor cameras appear to retain more detail in the highlights (like the sky) when the exposure histogram is pushed all the way to the right, and through RAW processing you might recover even more highlight details.

From http://www.dpreview.com/reviews/olympusep1/page19.asp

"These days there's very little difference between the tone curves or usable dynamic range of competing cameras, and the E-P1 matches similarly specified SLRs, though the Nikon D5000 and Olympus E-620 give you around half a stop more at the top end, thanks to their more gentle roll-off. We did find the E-P1 had a tendency to clip highlights in bright scenes, partly due to the tone curve, partly due to metering."

The GF1:
From http://www.dpreview.com/reviews/PanasonicGF1/page18.asp

"Thanks to the rather harsh tone curve the GF1's highlight dynamic range can't match the best mid-range DLSRs, and at 3.1 EV it can't quite compete with the Olympus E-P1 (at ISO 200). The fact that the GF1 tends to meter quite conservatively (i.e. slightly under exposes) - or can be forced to with a -0.3 EV compensation - means that the highlight retention in 'real world' images is as good as - often better than - the E-P1."

 

[Knight Palm]

It will be accentuated, if you substitute with the Olympus E-3 and Panasonic DMC-GH1.

However, the 6Mpix CCD sensor in D50/D70 could deliver stunning photo prints.

CCD at base ISO has some specific attraction, which also can be heard from those experienced with the Kodak sensors.

Being old technology, hence you'll find the D70 at the bottom of the DxOMark data. Any Panasonic sensor put into 4/3 cameras measures far better than the D70.

I think the D70 holds up pretty well for a camera that came out 6 years ago, here is the dxo comparison and far better seems a bit generous
Jim

http://farm3.static.flickr.com/2719/4326257406_9db22400bd.jpg

--

Ludwig Wittgenstein; British philosopher born in Austria; a major influence on logic and logical positivism (1889-1951):

“What can be said at all can be said clearly; and whereof one cannot speak thereof one must be silent”

 

[iano]

DSLR (use of a mirror) vs live view affects framing the picture...not the picture itself. So I assume the reference to DSLR is meant to be a reference to APS-C cameras? Or are you suggesting it is the reflex system that makes the difference?

Anyway...a point to note is that the GH1 is probably the best performing micro 4/3 camera for dynamic range so data from this camera should be considered.

 

[Ehrik]

For a given level of detail, a 4/3s sensor - all other things being equal - gives up a half a stop or slightly more of dynamic range to an APS sensor and it gives up two stops - for the same level of detail - compared to a 36mm x 24mm sensor. Actually, that's a bit of an oversimplification. The guys that know the physics better than I do can wade in and be more precise if they want. I'm probably giving a bit too much of a theoretical advantage to the larger sensor. But that's a decent "broad strokes" view.

Well, not to pretend I'm a physics guru, but I have to say I can't agree here - or that I agree that you give too much advantage to the larger sensor.

I would say the theoretical advantage is 1/3 EV for APS-C and 1 EV for FF.

If we have the same size pixels with the same DR per pixel, we have about 4x as many on an FF sensor. To think about DR per image it helps to bin the FF pixels so we have just as many output pixels. The sum of the max signal is then 4x as much but the sum of the noise (which limits the dark end) is only 2x as much, because of the way random noise adds. So the DR is just twice, or one stop more.

If we allow 4x bigger FF pixels (for the same pixel count) it doesn't change the conclusion much - with current technology at least. Small-pixel cameras like the LX3 and G11 have more DR per area than many DSLRs!

Where there is a two-stop difference is in ISO. The FF camera can be expected to have the same DR (and noise) at an ISO 4x higher.

In reality, the implementation matters a lot, everything isn't equal, so the theories are only rough guides.

I agree on everything else you wrote, good points!
. . .

For the OP, I would say it is true that DR is one of the weaker points of µ43 (with current technology), but clipping highlights is not evidence of poor DR but rather an effect of how the default metering behaves (and the headroom the manufacturer has chosen with the tone curve they've applied).

I'd also advise against letting yourself be influenced by DPR's "DR test". It's the weakest section of their reviews, IMO.

 

[sean000]

DSLR (use of a mirror) vs live view affects framing the picture...not the picture itself. So I assume the reference to DSLR is meant to be a reference to APS-C cameras? Or are you suggesting it is the reflex system that makes the difference?

Sorry... I've been a DSLR user for so long I forget that there are other APS-C cameras out there ;-)

No I am not suggesting in any way that the presence of a mirror makes a difference, but there is far more to DR than the sensor itself... otherwise all APS-C cameras would have the same DR characteristics.

Anyway...a point to note is that the GH1 is probably the best performing micro 4/3 camera for dynamic range so data from this camera should be considered.

That depends on whose tests you want to believe. According to DPReview's tests the EP1 has the most DR of the m4/3 cameras to date and the GF1 has more dynamic range than the GH1. At ISO 100 the GF1 captures 8.5 EV and the GH1 7.9 EV. That may sound like a significant difference, but in terms of the DR for highlights they are almost the same at 3.1 vs 3.0. The GF1's advantage of the GH1 is in the shadows with -5.4 EV vs -4.8 EV. The GH1 also appears to have more RAW headroom which kind of makes up for the lesser dynamic range, but that could come down to a change in testing methodology and/or ACR versions.

Which all brings me back to my point: The differences in DR between m4/3 cameras are so small that I would never base a camera buying decision on them, and most APS-C cameras have a slight advantage in the highlights while m4/3 have a slight advantage in the shadows. All that says to me is that in really contrasty scenes I might be able to expose to the right a little more with my APS-C DSLR. I will probably still need to brighten up the shadows, but not as much. On my GF1 I will need to watch those highlights a little more carefully, but underexposing a bit won't hurt me much since I will have more room to brighten up those shadows compared to my APS-C camera. That said brightening shadows too much reveals noise, and that could limit you when trying to lift the shadows on a single m4/3 exposure.

I guess I kind of see this thread as splitting hairs because I know that when we really run into the limited dynamic range of our cameras, a half stop or so doesn't make a huge difference when you are trying to capture a scene that needs even more range. That's why we often use things like graduated ND filters that are 2 to 3 stops, and why we often blend multiple exposures that are several stops apart. To get that kind of gain out of a sensor you would need to move up to full frame. The RAW headroom in full frame sensors is enormous and you can really brighten those shadows without revealing as much noise as you will brightening shadows in images from smaller sensors.

Sean