The new Panasonic-Fuji organic sensor has a theoretical dynamic range of 88dB (29.2 EV!). The current best DSLR the Nikon D800E has a measured dynamic range of 46dB (15.3 EV) (Source: Adorama via PetaPixel). If the next Panasonic and Fuji cameras can really achieve such an impressive Dynamic than these will be a huge step forward! It will not only close the gap to FF sensors but on paper even “beat” them. Of course, it has to be seen that how long it will take until the new sensor can find his way in real mass production cameras. And of course the competition is certainly working on similar sensors too!
UPDATE: There seems to be a big confusion on the conversion between “dB” and “stops (EV)” A 43rumors reader shared that table:
46 dB = 4.6 B –> 10^4.6 = 2^15.3 –> 15.3 EV (Nikon D800E)
88 dB = 8.8 B –> 10^8.8 = 2^29.2 –> 29.2 EV (New Organic sensor from Panasonic-Fuji)
10 dB = 3.32 EV = log 10 / log 2
(see also http://en.wikipedia.org/wiki/Decibel)
P.S.: Short focus on real existing cameras: ….Adorama (Click here) published the full list of 115 newly discounted Olympus cameras and lenses. The same list but separated can be seen at Amazon (here at the cameras and here are the lenses).
65 Responses
No doubt Dynamic Range is extremely important in photographing nowadays regardless portraits or landscapes .
Of course it is delightful if Pana can acheive such high DR as mentioned & hopefully it won cost as D800 when come to production .
Hopefully they ( Fuji & Pana ) would take too long to bring it to this planet .
Organic Sensor , I intend to have it .
If this information really true,.it will be a huge leap on DR improvement,.its really insane and hard to believe the sensor technology nowadays even from such a small sensor,..
Many people cannot envision what it means to double the dynamic range of “existing sensors” so they remain skeptical. Forget the numbers… seeing is believing. We will see…
I’m pretty excited because this is not incremental engineering but a fundamental shift in technology from silicon photodiode to organic photoelectric layer. This is enabling technology that will hasten Global Shutter(extra circuitry without severe sensitvity penalty) and higher resolution(smaller pixel size without severe sensitivity penalties).
Yes the first commercial product will be out soon and this tech will still be refined over the years still with better and better materials and incremental engineering.
Panasonic’s micro colour splitter is another fundamental shift in technology and I’m sure panny is already putting these two tech together into a future sensor.
So the 25 megapixel(a quarter of FF’s 100M) organic colour splitter sensor with global shutter for m43 is coming!
A game changer for real!
Doesn’t the D800e have 14.3 stops DR?
According to DxO unless I’m missing something here.
I believe it is ,Nikon D800/E its measured on ISO100 its more than 14 -stop
Admin, I am greatly appreciative of your blog and the rumors, resources, and sales links you make available. I have used the links before and plan to do so more in the future to help you to keep this great blog free for all of us!
I just wanted to let you know that DxO Mark lists the Nikon D800e as having 14.3 Evs of dynamic range. I think you may have been confusing something. The Panasonic GH3 has a 12.4 Evs and the Olympus OM-D E-M5 has a 12.3 Evs of dynamic range. So even with your confusion (unless I’m confused), the new theoretical limit of 14.6 Evs for this organic sensor is excellent news!
Happy parenting with that new wee one at home!
I’m quite sure you are comparing apple and oranges (or, in denglish, pears) here, since there are different ways to measure dynamic range – for example, DxOmark measurements are up to 14 stops.
See here for example:
http://www.dxomark.com/index.php/Cameras/Camera-Sensor-Database/Pentax/K-5-II
D800E has 14.33 EV
They are probably confusing different definitions of decibel, which differ by factor 2 (amplitude or power). Those 46 dB of D800 then actually mean about log2(10^(46/10)) = 15.3 eV and not 7.7 eV.
You are right:
46 dB = 4.6 B –> 10^4.6 = 2^15.3 –> 15.3 EV
88 dB = 8.8 B –> 10^8.8 = 2^29.2 –> 29.2 EV
10 dB = 3.32 EV = log 10 / log 2
(see also http://en.wikipedia.org/wiki/Decibel)
Would be really amazing!
Exactly. There is nothing confusing about decibell, it is just 1/10 of a Bell, which is a power of 10. ” SLRlounge writes that 6db roughly equals 1 stop, so 88db = 14.6 stops.” is just a BS.
Thanks for clearing that up! I had a hard time figuring out how to convert the dB calculation to EV dynamic range stops and saw that 6dB = 1 stop somewhere.
I updated the calculation on slrlounge.com
– Joe fotosiamo
Actually, you got it right and admin got it wrong.
It is indeed roughly 6 db = 1 EV.
Yes, by math definition it is about 3 db = 1 EV, however, it is tradition like calculate sensor size, you can’t see because 1 inch = 25.4mm, so, 1 inch nikon 1 sensor has 25.4 in diameter.
i.e. aptina’s 16M aps-c cmos claimed to having 72 db
A more accurate conversion for voltage-based dB numbers is 6.02 to a stop. But everyone rounds to 6 for the most part.
Yes, each step of the EV scale represents a factor of 2 (or 1/2) in the light captured.
http://en.wikipedia.org/wiki/Exposure_value
As described by others, a factor of 2 on the Decibel scale is around 3 dB (3.0103 more precisely) for power or 6 dB (6.0206 more precisely) for amplitude.
http://en.wikipedia.org/wiki/Decibel
So, the only issue is whether to equate the EV scale with dB as a power or amplitude quantity. I think because film is sensitive to the amount of light hitting the surface, it should be amplitude, which is why 1 EV equates to about 6 dB. This is also the common usage in control theory applications, with which I am more familiar.
My qualifications are Chief Engineer for the following project:
http://en.wikipedia.org/wiki/Boeing_X-48
Normally I’d say that this kind of news is totally boring because many patents and prototypes turn out to be dead ends never making it to actual products.
However, the official press release on the Fuji homepage makes it quite explicit that they are determined to use this technology. Very exciting.
WOW!!! ….I own fourthirds, micro fourthirds and fuji xf equipment, so whatever comes to fruition from the Sony & Olympus camp or Fuji & Panasonic camp I stand to benifet.
Full frame 135 format may continue to keep it’s lead in the future sensor wise but all things considered it’ll probably get to the point were most people won’t want more than what a m4/3 or APSC sensor offer and would much rather enjoy the size and weight benefits of the smaller formats.
Thinks leaves Canon in a very awkward position sensor wise unless they have some secret weapon they’re preparing.
Canon’s secret weapon is that they have lots of money, so when people want a “good camera” they automatically think Canon or Nikon 🙁
Also, if this all this really pans out what does this mean for sigma’s foveon sensor tech?
Seriously good work Admin, thanks and keep it up. Anything new on the OMD Pro-e7 hybrid and those new m4/3 f2.8 zooms?
Since this new sensors won’t find place into this year new cameras, pleas Admin, can you tell us some more rumors concerning Pana RF styled camera, new OMD, new OMD firmware? 😉
yep coming soon. Working full time on it!!!
Thaaaaaanks! 🙂
Thank too!!! 🙂
Thanks
double dynamic range? can be it right?
When, oh when.
soap bubble, pipe dream, wunschträume, quatschkram
@Whenguy: 2018
Of course, even if we assume this new tech is as good as it sounds and there are no hidden downsides, there is no guarantee it will appear in a mirrorless camera first. My guess would be that it will debut in phone cameras, which represent a higher-volume market and which would really benefit from more dynamic range. Even if it appears in a camera that most of us would want to use, I have to guess that the sensors for next year’s cameras are in production now, and the cameras a year after that are already designed and planned. So maybe in three years we might see this improvement, if it turns out that it IS an improvement…
I would hold my breath though. When Sony announced and patented reversed CMOS technology, everyone was dreaming about an Alpha, when to date they only have compact camera’s using the technology…
In the anouncment they say: “The two companies will promote the application of this organic CMOS image sensor technology to a wide range of products including security cameras, in-vehicle cameras, mobile device and digital cameras.”
Which means you will see the organic CMOS in security cams and Cell Phones first, to maybe later see some compact camera’s use the tech. I doubt we will ever see hat happening in a 4/3 camera, much less in an APS-C or FF…
Ermm…. you simply divide the dynamic range in dB by 3 to get EV’s 🙂 The result will be the same (46/3=15.3, 88/3=29.3), but no need for confusing formulas 🙂
PS this is because both dB and EV are based on logarithmic scale, except that EV=3 db, with 3 dB being a conventional generic measure for 1 stop of dynamic range (which is applicable to other media such as audio as well).
I have no doubt that Pana-Fuji WILL create such a sensor for mass production which will beat the best of Nikon and Canon hands-down. Why not? Please post some pre-order links.
Quote from the Fuji press release:
_”The new organic CMOS image sensor offers the industry’s highest dynamic range of 88dB, advanced sensitivity 1.2 times more sensitive than conventional sensors”_ http://www.fujifilm.com/news/n130611.html Bottom of paragraph two.
Read carefully, it says _”sensitivity 1.2 times more sensitive than conventional sensors”_ An increase of 1.2 times is the same as saying a 20% increase. Twenty percent is a significant increas, but not as earth-shattering as everyone seems to thinks.
sensitivity != dynamic range
You are wrong. It also states that when they refer to sensitivity, it’s in reagrds to light sensitivity and is for High ISO noise.
” The sensitivity of image sensors refers to the conversion ratio of light into electric signals. The greater the sensitivity is, the clearer you can capture low-light scenes.”
sensitivity (ISO) != dynamic range….
“Read carefully, it says _”sensitivity 1.2 times more sensitive than conventional sensors”_ An increase of 1.2 times is the same as saying a 20% increase.”
This is incorrect, read carefully, it says _”sensitivity 1.2 times **more**”, not “1.2 times the”, the keyword here being more, implying 1.2 times additional. That makes it 120% more, not 20% more.
30 stops of dynamic range would be amazing. Even the best b&w films are what, about 19 stops?
I was under the impression that each stop meant twice as much as the previous stop. So then wouldn’t double the dynamic range mean that it has one full stop more? Example: 15 stops/EV vs 14 stops/EV is twice as much.
To have 29+EV would then be a ridiculous amount more, and I don’t think it could possibly be true that this new sensor is literally light years ahead (more than 16,000 times more dynamic range…) anything we’ve ever seen…
To clarify:
29EV – 15EV = 14EV, so if each EV is twice as much as the previous, then 2^14=16384 times more dynamic range than Nikon’s sensor.
Maybe it’s just me that finds this to be a ridiculous jump right now?
Apples and oranges. You’re comparing an unspecified as to how it was measured theoretical dB with a measured after the fact captured dB. The D800 number that DxO publishes is a dB defined as saturation down to 1:1 signal to noise ratio. This is post shot noise, post dark current noise, and post read noise. Not theoretical at all.
We don’t know the efficiency of the Fujifilm/Panasonic sensor, or how the noise number was calculated, so we can’t actually make a logical inference as to its actual performance yet. However, Aptina lists the dB number for its 1″ sensor (used in the Nikon 1 cameras) at 84dB. (source: http://www.aptina.com/products/image_sensors/ar1011hs/)
Given the fact the they released absolute numbers (i.e. 88dB DR, 1.2x sensitivity) rather than a general claim (e.g. ? time more), I’m inclined to believe that this a indeed a real world measurement. Remember, we aren’t talking about a patent. This is a press release preceding actual conference presentations this month, so it’s a safe bet they have a prototype. (Much like Panasonic’s color splitter tech a few months ago where they clearly had manufactured test versions.)
That said, I will agree to your latter point about calculations… In particular, dB has an extraordinarily annoying property of having two different calculation methods: power/intensity and field/amplitude. While fundamentally the same, the latter gives a value of exactly twice the former. Many signals have conventions, but for dynamic range of light, it’s unknown.
I would say that at the moment, it’s most reasonable to assume that the 88dB is calculated as a field (which is actually reasonable, as volts are a field, and volts are what is being read out of the sensor). Thus the reported value in EV is only:
88dB(volts) -> 44dB(light-power) -> 14.6 EV
Now, it’s worth noting what the article does rather clearly claim is an “improved the signal saturation value by four folds”, which would imply +2EV on the bright side. Given that they’ve apparently also reduced noise (makes sense given more room for circuitry, I suppose), their claims seem to indicate somewhere around 2-4 EV improvement over existing sensors.
That would be a quite solid improvement in line with reality (unlike 88dB(power)!). Why a sensor that is claimed better than existing ones is coming in somewhat less than the D800E’s is anybody’s guess. It could be different measurement methods, or it could be that it’s only vs Panasonic’s existing ones. Dunno.
At the very least, 20% better sensitivity and less angular vignetting don’t hurt. Combined with Panasonic’s color splitters, the future for their sensors is looking *cough* bright :).
Note that there is a difference between sensor dB in spec sheets and the dB that a system using such a sensor actually records. The first is theoretical maximum signal retention, the latter is practical signal actually retained in the resulting data (and the DxO numbers are down to a signal-to-noise ratio of 1:1, which is an unusable value, IMHO).
That’s one reason why I used the Aptina sensor as an example. Aptina’s theoretical dB is 84. DxOMark’s measured dynamic range is 11 stops.
We don’t yet know the efficiency, read noise, or well saturation numbers of the Fujifilm/Panasonic sensor, so we can’t state with any certainty what it can do in a practical sense, but I’ll bet it measures less than 12 stops on DxOMark when it eventually gets there.
Moreover, I’d like to point out that if you do take advantage of the lower angle of incidence capability of a back side sensor with 100% fill factor like this, you’ll have the same problem that Kodak had on their 14n DSLR: potential side-to-side color issues. The Bayer layer is immediately above the sensing area. What happens is that light coming in at low angles goes through the edge of the adjacent Bayer area. You’d need to be very careful with lens design with such a sensor, I think.
Here’s what I don’t understand
We don’t yet know the efficiency, read noise, or well saturation numbers of the Fujifilm/Panasonic sensor, so we can’t state with any certainty what it can do in a practical sense, but I’ll bet it measures less than 12 stops on DxOMark when it eventually gets there.
Dear Thom,
You are correct. The 88dB (by electron number) shoud convert to EV by 6, then 88dB = 14.67 EV.
The new sensor’s well saturation is 8500e per um2. The stated “88db” is for the 3x3um pitch or 76500 electron at saturation. Panasonic stated that they have archived 2.9 electron readout noise so the DR is 76500/2.9=26379 (times) = 2^14.67 or 20log(26379)=88dB.
A/D bit number also affects the real camera DR. If a camera use 12bit A/D, then the DR cannot get more than the 12EV (if not lesser as normal). If you want get 14.66EV our of the sensor, then you have to at least using 15bit or 16bit A/D.
We shouldn’t get too euphoric too early. Many of us expected so much of the Foveon sensor and although it delivers in fantastic style at base ISO, it fails to deliver at higher ISOs and also has got it’s problems in other areas.
So let’s wait and see if this promising concept can be realised without compromises in other areas such as high-ISO performance, read-out speed etc.
I wonder if this is why Panasonic has suddenly developed a need for IBIS. Together with this sensor it enables them to offer a big step forward to Olympus users, an offer that may be taken up. (also other users who want IBIS)
The issue is, will Pany/Fuji, kill their own market by greed of selling to all an sundry. Or, will they try to dominate and lead the market, while they have the possibility of doing so? If they sell to all and sundry the larger format benefit ratio stays and there will be little shift in the status quo of camera sellers.
Also what will be the effect on Oly? Will they survive?
well a panasonic with this new sensor OMD level IBIS and panasonics full on video modes would be a hard to resist setup!
If they kept the sensor ‘in house’ it would be hard to resist indeed.
If the article heading of double the dynamic range comes from Pany/Fuji, the argument over which calculation method is used is not relevant. The publicity would be using the same calculating mythology. Or, would they?
The heading is implying that there is a 2 to 1 benefit with the new sensor, irrespective of how or what calculations used. However it does not qualify at what settings this 2 to 1 advantage was seen. As another camera (known entity) is us it would be reasonable that there is a relative comparison.
The original Fuji article makes no mention of “double”, that’s a 43rumors thing based on the press release’s “88dB” figure. The original does make mention of a “four fold” increase in saturation value, but that equates to a mere 2EV, all else being equal.
Only here can someone say “a mere 2ev” and get away with it. It took 3 years from the 5d mk2 to 5d mk3 and the improvement is somewhere along the lines of .5ev. 2 ev is a ridiculous jump forward
read it again!!,this prototype sensor use organic photoelectric layer from FUJI and Panasonic’s semiconductor substate
although i don’t know format of sensor may M43 or compact format , but it still uplevel of performance.
Read what I said again.
the good thing about m43 is the sensor is small enough to keep a faster tech turn around than FF… and if m43 gets say 3 -4 stops better than the OMD then any other improvement becomes academic!
Once you can use an F3.5 lens in a candel lit room with clean ISO 48,000 then we are done!
So does this mean that I shouldn’t upgrade my G3 and wait until a new camera with this organic stuff comes out, and should I invest in 35-100mm lens instead?
88dB … theoretically, this reads very well! But can such dynamics are practically only used when the signal processing it also can handle and the RAW-files are present in at least 14-bits, better 15- or 16-bits! Which mft-cam offers 14-bit RAW?
Here’s a fun fact: +1 EV means the dynamic range is doubled.
2^(29.2EV – 15.3EV) = 2^13.9 = 15,286.81
That’s fifteen thousand times the dynamic range.
Nobody’s bullshit meter went up? No? Keep believing in fairies, then.
If you bothered to read the article you would see that all that they are promising is “signal saturation value*7 by four folds compared to that of conventional image sensors” which means a 2EV gain.
I’ll welcome that anyday…
Improvement of DR is great value! Apart from that I suggest we wait until there is a camera with the new technology that can be evaluated to HOW MUCH better it is in real life.
Someone above claims “it would outperform even FF”. Well if the new technology is applied on a larger sensor than m43 (i.e. Fuji FF) there will still be a difference based on a 4 times larger sensor.
Keep also i mind that other sensor manufacturers also develops new technologies that might be more or less significant,
One thing for sure: Sensor technology (sensor, surrounding electronics, coding) is far from matured!
Dynamic range is not complicated. Take a properly exposed image. Then under/over expose by 1,2,4,8 stops etc. If you can push or pull the exposure say in Lightroom, and have an acceptable, low noise image equal to the properly exposed image. That is your dynamic range. I can take a 2 stop over exposed image from a GH-2 Raw file and get a good image. Imagine if you were off by 8 stops? Pretty amazing.
so much speculation for and against. Bring out the sensors and we will find out. For now I would like to point out that the new technology must be a measurable improvement on the XTrans sensors or why bother?
> or why bother?
Because this is a sensor destined for thin smartphones, not m4/3 or larger cameras. It’s essentially an organic BSI that’s thinner because it doesn’t have the photon conversion layer in silicon. Note that one of the versions being talked about has .9 micron photosites. Light collection is partly dependent upon the square area being collected, so .81 square microns is a long way from 16 square microns.
The release of this information predates a speech Fujifilm will be presenting at an upcoming sensor conference. The “why bother” part is “we want some attention.” If I’m reading the original Japanese information correctly, the sensor as described can not yet be manufactured in quantity, either. But that’s part of the point: most camera “design wins” are more than a year in the future. Decisions are being made now for 2015 cameras (and in some cases).
Unfortunately for Fujifilm, everything is a moving target right now. With Samsung’s S4 Zoom announcement and Nokia’s upcoming announcement, it’s unclear whether you even want one of the small smartphone sized sensors in your next smartphone. Rumors abound about Google, Nikon, and Sony all working on larger sensor phones, too.