OMD with more MP?

[1943Mike]

Does anyone think that Olympus will offer an OMD camera with a 24 mp or greater sensor in the near future?

 

[MatsP]

Does anyone think that Olympus will offer an OMD camera with a 24 mp or greater sensor in the near future?

 

[photohounds]

[snip] when Olympus will go FF. Never going to happen

Well, my contention is that it's better go FF35, if one needs more than 16 Mpx. Sony demonstrated it. There is no reason why Oympus of Fuji not to do the same, instead of pushing to extremes a cropped sensor and fail, like it happened to Sony before their A model.

Sony has a FF Zeiss 35/2.8 which actually costs like the Oly 17/1.7 but can meet a resolution of 36Mpx.

The advantage seems clear enough.

Am.

--
'Photo & Poetry'
http://amalric2014.blogspot.it/
http://www.flickr.com/photos/amalric

Weight, size, cost, point beyond bragging?
.
Just checking, you mean THIS lens?
.
If so, almost TWO stops Vignetting wide open and still ONE stop at f5.6.
Surely that's a joke, and it can't even outresolve the Zuiko 17mm/f1.8 .
.
It DOES cost nearly double though.
There ARE some good lenses in the lineup, that isn't one of them, but then, value was never a SONY lens strong point.
.
OH ... this is FULL frame , 35mm is full MARKETING frame.
.
Another misconception" four thirds does not use "CROP sensors" they are designed that size. FMF was adapted from 35mm miniature format.

--
Well designed gear performs better for longer than well marketed gear.
Odd that people complain a lens is not sharp enough,
and then proceed to make pics where 95% is OOF ..
General Pics:
http://photohounds.smugmug.com/
Oly and other .. Gear test samples - even RB-67!:
http://photohounds.smugmug.com/Gear-tests
How DO OMDs cope with dim-light action and smoke?
http://photohounds.smugmug.com/Performing-arts

 

[Alex Notpro]

FYI all 16MP cameras use a Bayer CFA sensor, which means they're really 4MP interpolated to fake 16MP.

Actually, they're 16 MP monochrome pixels interpolated into 16 MP full color pixels.

OK but they're not even "full monochrome". Two thirds of the light and color information is discarded in each pixel. You need 4 of these partially blind monochrome pixels to make 1 real full color pixel.

I'd like a real 16MP camera please.

It would be matched by a 32 MP sensor covered with a Bayer CFA and outperformed by a 64 MP sensor covered with a Bayer CFA.

You are probably right (based on your track record), but I find it counter-intuitive that the equivalence ratio would be 2x rather than 4x. Are you fully considering "information content" of each pixel? Is there an additional layer of information interpolation lurking in there?

--
https://www.flickr.com/photos/atramos/collections/72157628525671669/

 

[amalric]

Yawn, different sources, perhaps sample variation.

Try to learn some politeness first, some fair play.

I understand those a rare commodiies, but try.

Then perhaps

 

[tjdean01]

Even if it's BETTER at High ISO I'd rather it be 12MP and gain the stop or two in noise performance. I'd almost always take a stop or two in noise performance over an extra 3-4MP, down to about 8 MP, that is. Look at the Sony A7s. 12MP low ISO beast!

 

[Digitall #107280]

Existing pixels are sufficient, at least until mastered the best noise in low lights, rather, is what is more important, much more important than the race to the pixels, which is already out of fashion.

The race to the pixels does not make sense when you do not dominate the low noise level, and this rule is for all manufacturers, no exclusions!

 

[Great Bustard]

FYI all 16MP cameras use a Bayer CFA sensor, which means they're really 4MP interpolated to fake 16MP.

Actually, they're 16 MP monochrome pixels interpolated into 16 MP full color pixels.

OK but they're not even "full monochrome". Two thirds of the light and color information is discarded in each pixel. You need 4 of these partially blind monochrome pixels to make 1 real full color pixel.

This is true.

I'd like a real 16MP camera please.

It would be matched by a 32 MP sensor covered with a Bayer CFA and outperformed by a 64 MP sensor covered with a Bayer CFA.

You are probably right (based on your track record)...

Might wanna rethink that. ;-)

...but I find it counter-intuitive that the equivalence ratio would be 2x rather than 4x.

I'm not sure of the details, but I'm pretty sure it's because detail information comes primarily from the green channel, and there are two green filters in an RGGB CFA. Thus, the 2x factor between Bayer and Foveon is simply an average. If, for example, you were taking a pic of a scene that was dominated by a color far from green (for example, a tightly framed photo of a red rose), then it would be closed to 4x -- say 3x.

Are you fully considering "information content" of each pixel? Is there an additional layer of information interpolation lurking in there?

There's spatial information and color information. From what I've seen, the 2x approximation for Bayer vs Foveon seems about right. For example, that means that the current 15 MP Foveon sensors should resolve as well, on average, as current 36 MP Bayer sensors, lens permitting, of course.

 

[Anders W]

FYI all 16MP cameras use a Bayer CFA sensor, which means they're really 4MP interpolated to fake 16MP.

Actually, they're 16 MP monochrome pixels interpolated into 16 MP full color pixels.

OK but they're not even "full monochrome". Two thirds of the light and color information is discarded in each pixel. You need 4 of these partially blind monochrome pixels to make 1 real full color pixel.

This is true.

I'd like a real 16MP camera please.

It would be matched by a 32 MP sensor covered with a Bayer CFA and outperformed by a 64 MP sensor covered with a Bayer CFA.

You are probably right (based on your track record)...

Might wanna rethink that. ;-)

...but I find it counter-intuitive that the equivalence ratio would be 2x rather than 4x.

I'm not sure of the details, but I'm pretty sure it's because detail information comes primarily from the green channel, and there are two green filters in an RGGB CFA. Thus, the 2x factor between Bayer and Foveon is simply an average. If, for example, you were taking a pic of a scene that was dominated by a color far from green (for example, a tightly framed photo of a red rose), then it would be closed to 4x -- say 3x.

We are better at perceiving luminance than color detail, and luminance is primarily based on green, yes, for reasons explained here:

http://en.wikipedia.org/wiki/Luminance_(relative)

I would think that holds even if the object is strongly colored like your rose, when distinguishing bright from dark red.

Are you fully considering "information content" of each pixel? Is there an additional layer of information interpolation lurking in there?

There's spatial information and color information. From what I've seen, the 2x approximation for Bayer vs Foveon seems about right. For example, that means that the current 15 MP Foveon sensors should resolve as well, on average, as current 36 MP Bayer sensors, lens permitting, of course.

Should perhaps, but do they? Here are a set of crops from optyczne.pl (aka Lenstip when translated to English) showing the 36 MP Bayer D800 versus all three 15 MP Foveon Sigma DP Merrills. The images were processed (by optyczne.pl) from RAW by means of DCRAW (no sharpening, no NR). The crops from the Merrills were upsampled (by me) in PSE 2.0 to roughly the same size as that from the D800. Does it really look like even-steven resolution-wise to you? ;-)

Note: Click on the image to view it properly.

 

[Bassam Guy]

Does anyone think that Olympus will offer an OMD camera with a 24 mp or greater sensor in the near future?

 

[Steve_]

If the pixel pitch and read noise per pixel are both halved, there will be no change in noise at all. So, if a 32 MP mFT sensor had pixels with 71% the read noise of a 16 MP sensor, then there would be no difference in noise.

Halving the pixel pitch would not halve the read noise. Were that the case, all modern digital cameras would have much finer pixel pitches than they currently do.

 

[Steve_]

Smaller pixels result in "higher IQ" if the smaller pixels do not result in greater read noise per area on the photo ...

That's a pretty big if.

Not really. For example, the 16 MP EM5 has less read noise per area than the 12 MP EP3.

This is not a relevant statement. If the improved noise characteristics of the EM5 sensor were a result of its finer pixel pitch they would just have made those 12MP sensor 16 instead.

Perhaps that is one reason the D800 doesn't sport anything like a .0026mm/pixel pitch.

For any given tech and sensor size, there is a "sweet spot" for pixel count, and this "sweet spot" is continuously moving towards larger and larger pixel counts with successive generations of sensors.

I have stated that higher resolutions and finer pixel pitches are possible due to improved sensor technology. Strangely enough you at that point flail away at the pixel-size dead horse, implying that nothing else has or will change and that read noise will somehow hold constant. Now at this juncture you admit that other facets of sensor technology enable a particular pixel count (and thus pitch at the same sensor size). This is what I have been saying all along... Sensor technology is not currently sufficiently advanced as to permit the .0026mm/pixel pitch that results from a 32MP m43 sensor to produce adequate results.

And I'd be surprised if its successor does, either. If I do the math correctly that would produce a 128MP solution.

The thing is, more pixels result in more resolution, regardless of the lens (of course, the sharper the lens, the more resolution you'll get). More resolution means more latitude for NR (noise reduction). So, even if more pixels are intrinsically more noisy, depending on how much more noisy, the greater pixel count will still result in more detail with the same or less noise with appropriate processing.

It's the 'depending on how much more noisy' part that has been understated in your comments. By the time you get to the pixel pitches proposed in this thread, the gains from the increase in sensor resolution will be swamped by the increase in noise, perhaps even at base ISO.

 

[Steve_]

Most users don't realize how noisy m43 sensor are in DSLR terms because there's a lot of NR being applied to the JPEGs. But once you compare to good APS-C output in LR the rather dramatic increase in noise at base ISO is apparent. Fortunately Olympus and Panasonic has filtered what is possible through a recognition that IQ must compete with DSLRs before the platform is tractable.

You are right that there is still a difference of about one EV in base-ISO shadow noise (effectively indexed by DR) between the very best (rather than average) APS-C sensors and the current generation of MFT sensors. But if you expose optimally (ETTR), current MFT sensors are already good enough that this difference is hardly noticeable except for high-DR scenes where you have to push the shadows in PP. And for such scenes, you can often solve the problem by bracketing exposure and then merging/aligning the images in PP.

What is your factual basis for this? What is 'hardly noticeable'? What is 'high-DR'? What is 'often'?

 

[Steve_]

That's not the way I see it ...

Everyone gets an opinion.

Sure. Just questioning the factual basis of yours. Take this RAW from IR for example:

http://www.imaging-resource.com/PRODS/olympus-e-m1/EM1FAR2I0100.ORF.HTM

Do you find the noise troublesome if you let LR develop it at default settings?

Anders, I've been polite with you. I don't care what you make of IR's test shots. I shoot the camera in the conditions that matter to me and use the required RAW adjustments. I also do the same with a D7000 and K3 Pentax. The noise in the EM5 shots is much greater, to the point that I could not stand more noise. Feel differently? I don't give two shits.

Lowering their resolution would not fix what ails them, making the entire point irrelevant to the discussion at hand.

What is your factual basis for claiming that lowering the resolution of Canon FF sensors would not fix what ails them?

That's just priceless. To quote you from earlier in this thread:

"Note that current MFT sensors are better in terms of DR (about one EV better in the case of the E-M1) than current Canon FF sensors with close to 24 MP. Nevertheless, I haven't seen a whole lot of suggestions that Canon should drop the pixel count down to 12 or so MP."

First you ridicule the idea of lowering their resolution, and than protest the notion that lowering the resolution would not help. It's pretty obvious you are just arguing for the sake of argument, a point that you will no doubt argue as well.

And since you think lowering the resolution would not help in Canon's case, why do you think increasing the resolution would be detrimental in the case of MFT?

Because of the consequences of reducing pixel pitch 31% from a level that produces noise performance that is just barely good.

 

[1943Mike]

Develop along two lines! Different photographers have different needs, which is clearly obvious in this thread. If i were Olympus I would let the next EM-1 keep 16 Mp but improve AF and high ISO performance for those who want to shoot fast moving subjects and/or in bad light. And put a 24 Mp or greater MP count sensor in the next E-M5 for those who shoot landscape and architecture in good light and to whom resolution is important.

I suppose this would be my wish. I do both kinds of photography. For the moment, shooting BIF, I use my Canon 7D with the Canon 100-400. For macro photography I use one of my Canon DSLR's (either the 7D or the 5D) with either the Sigma 180 macro or Canon 100 macro lens. However, when hiking (almost always in good weather and in bright conditions) I carry my OM-D E-M5 with the Panasonic 12-35 f/2.8 attached. I like the Olympus 5 axis IBIS and rarely carry a tripod anymore. I believe that the engineers/scientists who develop µ43 sensors will be able to develop a 24mp sensor that should be acceptable to me and other landscape photographers who don't usually have to worry about lack of light. There are, of course, very good landscape photographers who DO worry about the lack of light (evening/twilight/early morning images not to mention night shots) but that's not me.

So, having started this thread I would say the discussion regarding amount of mp's on a sensor has been, on the whole, productive and given myself and others some food for thought.

Thanks you all.

 

[Bhima78]

Even if it's BETTER at High ISO I'd rather it be 12MP and gain the stop or two in noise performance. I'd almost always take a stop or two in noise performance over an extra 3-4MP, down to about 8 MP, that is. Look at the Sony A7s. 12MP low ISO beast!

Let's see how the A7s plays out first. It might be a case of the ISO performance/DR being completely negated once you take the A7R image and shrink it down to 12MP. If this is the case, there is literally no benefit to the 12MP sensor, and in fact you are actually given less flexibility. Not saying this will happen, but I'm interested in seeing how the tests play out.

 

[bluelemmy]

Bassam Guy wrot

Please explain why some of you really want 24 mp? I don't honestly need it. I'm content with the 16mp my E-PL5 offers. Outside of drastic crops or printing on huge paper, what practical benefit is there?

I, too, like other posters, would prefer lower noise than more pixels. I'd much rather get a decent 90 second ISO 200 shot.

Exactly my own feeelings.

 

[Anders W]

That's not the way I see it ...

Everyone gets an opinion.

Sure. Just questioning the factual basis of yours. Take this RAW from IR for example:

http://www.imaging-resource.com/PRODS/olympus-e-m1/EM1FAR2I0100.ORF.HTM

Do you find the noise troublesome if you let LR develop it at default settings?

Anders, I've been polite with you.

Polite by your standards possibly, hardly by mine.

I don't care what you make of IR's test shots.

I didn't ask you to care. I asked a question. You dodged it. I rest my case.

I shoot the camera in the conditions that matter to me and use the required RAW adjustments. I also do the same with a D7000 and K3 Pentax. The noise in the EM5 shots is much greater, to the point that I could not stand more noise.

What you personally can and cannot stand is of no interest to anyone but you. What is of interest is what you can substantiate. And judging by your response to my question above, you are not willing or able to enter a discussion about that.

Feel differently? I don't give two shits.

I didn't expect you to.

Lowering their resolution would not fix what ails them, making the entire point irrelevant to the discussion at hand.

What is your factual basis for claiming that lowering the resolution of Canon FF sensors would not fix what ails them?

That's just priceless. To quote you from earlier in this thread:

"Note that current MFT sensors are better in terms of DR (about one EV better in the case of the E-M1) than current Canon FF sensors with close to 24 MP. Nevertheless, I haven't seen a whole lot of suggestions that Canon should drop the pixel count down to 12 or so MP."

First you ridicule the idea of lowering their resolution,

I didn't ridicule that. I just pointed out that I haven't seen a whole lot of suggestions in that direction.

and than protest the notion that lowering the resolution would not help.

I didn't protest anything. I just asked you a question. You dodged this one too.

It's pretty obvious you are just arguing for the sake of argument, a point that you will no doubt argue as well.

Based on ordinary logic, the only thing that's obvious this far is your unwillingness and/or inability to answer questions and substantiate your position.

And since you think lowering the resolution would not help in Canon's case, why do you think increasing the resolution would be detrimental in the case of MFT?

Because of the consequences of reducing pixel pitch 31% from a level that produces noise performance that is just barely good.

As everyone can see for themselves, you effectively dodged this question too. First, you failed to address the inconsistency in your position that my question spells out. Second, you failed to spell out what specific (presumably negative) consequences of reducing the pixel pitch you have in mind.

 

[Anders W]

Most users don't realize how noisy m43 sensor are in DSLR terms because there's a lot of NR being applied to the JPEGs. But once you compare to good APS-C output in LR the rather dramatic increase in noise at base ISO is apparent. Fortunately Olympus and Panasonic has filtered what is possible through a recognition that IQ must compete with DSLRs before the platform is tractable.

You are right that there is still a difference of about one EV in base-ISO shadow noise (effectively indexed by DR) between the very best (rather than average) APS-C sensors and the current generation of MFT sensors. But if you expose optimally (ETTR), current MFT sensors are already good enough that this difference is hardly noticeable except for high-DR scenes where you have to push the shadows in PP. And for such scenes, you can often solve the problem by bracketing exposure and then merging/aligning the images in PP.

What is your factual basis for this? What is 'hardly noticeable'?

See examples below.

What is 'high-DR'?

What I talked about was "high-DR scenes where you have to push shadows in PP". See exemplification below.

What is 'often'?

In this case, I take "often" to mean at least 50 percent of the time.

Below you find examples of three scenes. The first is an example of a scene with low DR. The second is an example of a scene with high DR (requiring a bit of shadow pushing if you expose so as not to clip the highlights) but not so demanding that it is beyond what the E-M5 sensor can manage in a single shot. The third is an example of a scene whose DR is far beyond what the E-M5 sensor can manage (unless you accept that much of the scene goes impenetrably black). It is the result of blending nine different shots with one EV difference in exposure between each shot.

Now my question to you: Do you find any of these excessively noisy?

 

[Alex Notpro]

Should perhaps, but do they? Here are a set of crops from optyczne.pl (aka Lenstip when translated to English) showing the 36 MP Bayer D800 versus all three 15 MP Foveon Sigma DP Merrills. The images were processed (by optyczne.pl) from RAW by means of DCRAW (no sharpening, no NR). The crops from the Merrills were upsampled (by me) in PSE 2.0 to roughly the same size as that from the D800. Does it really look like even-steven resolution-wise to you? ;-)

Note: Click on the image to view it properly.

A B&W image seems to me like a really strange way to compare sensors where the key difference between them is how they capture color. It's like using a scale to compare two cars' speeds. Or trying to use a speed radar to figure out which one is heavier. What am I missing???

--

Look Here First

The Dopiazza Set Generator is no longer updating these sets. For my latest uploads, see my photo stream.

www.flickr.com

 

[Great Bustard]

If the pixel pitch and read noise per pixel are both halved, there will be no change in noise at all. So, if a 32 MP mFT sensor had pixels with 71% the read noise of a 16 MP sensor, then there would be no difference in noise.

Halving the pixel pitch would not halve the read noise.

I made no such claim. I simply said that if that were the case, then there would be no noise penalty for smaller pixels.

Were that the case, all modern digital cameras would have much finer pixel pitches than they currently do.

Funny thing about that. We all know that they can make really small pixels that are very good quality -- just look at the Sony RX100. An mFT sensor with RX100 pixels would have 40 MP, and a FF sensor with the same pixels would have 150 MP. So, why don't they? The tech is already here. Why is 20 MP good for the RX100, but 40 MP of the same pixels is not good for mFT and 150 MP of the same pixels is not good for FF?

That's a discussion worth having, methinks.