Last updated:- 29th February 2024
Introduction.
I support traditional science and marketing values. That means accurate and mathematically correct data has priority over commercial preferences. I studied electrical engineering, worked as a project engineer, and held several marketing positions in my working career. My marketing training started at a global manufacturer of industrial automation solutions and electrical equipment, and my interest in consumer behavior developed with product launches, the first digital calculator, the original XT PC, automation solutions, large project sales, and the photography segment.
Please study my ISO Low, ISO100, and ISO64 series. My focus in this discussion is luminance (reflected light). We will review the image signal path from the subject to the sensor because our ability to optimally capture reflected light (image signal) depends on the sensitivity (efficiency) of the image sensor and our ability to manage the digital image-taking process. This article illustrates why many photographers question those promoting the oversimplified "size and capture" theory...
Study this article discussing the 7 points each photographer should know...
What are the main technical differences between sensors? We know sensor sensitivity is the sum of the optical and quantum efficiencies of the image sensor. Pixel area (size) influences these optical and quantum efficiencies. The visible impact that pixel area has on quantum and optical efficiency is a good question? For example, the Canon 6D and Olympus Pen-F are 20MP cameras. The pixel area on the 6D is 248% larger than the Pen F. How much does this benefit the Canon's image quality, and what should one look for? One option is the DxOMark IQ database, and the practical option is shadow details. We also know each image sensor has a native noise floor that influences our IQ. The pixel's effective photon-sensitive area also changes for BSI, Live MOS, and Standard CMOS sensors...
The technical characteristics of image sensors are, therefore, unique. The design specs of each image sensor determine its technical characteristics and NOT its physical size. These characteristics include the saturation level, dynamic range, noise floor, and sensitivity of each sensor.
What is the "size and capture" theory? The best place to learn more about this theory is the well-known "size and capture" authority DPReview. Their camera reviews repeatedly explain the benefits of large sensors capturing more light than crop sensors. The "size and capture" theory predominantly applies to sensors smaller than full-frame sensors. It does not equally apply to FF and MF cameras. These are the benefits you should expect from your new FF (large sensor) camera:
- They capture more light...
- Have better image quality...
- Almost no image noise...
- Much better low-light IQ...
- DR with No highlight clipping...
- Better Auto-focus & video...
- The magical FF look...
- Better background blur...
- More and bigger bokeh...
- The joy of perfect IQ...
- A better way of doing photography
- Testing the Pen F and the A7S III
- A quick review of the test results?
- A few additional thoughts
- Conclusion
1. A better way of doing digital photography
- Your ISO function adjusts the sensor's sensitivity
- Never use ETTR at higher ISOs because the DR is less
- You don't need a flash because FF cameras have no noise
- They never use a tripod because new cameras have IBIS
- They need high-resolution cameras because they CROP
- They always argue while using the analog exposure triangle
- They depend 110% on FF sensors, AI, and the perfect AF
- Crop sensor lenses suffer from high levels of diffraction...
- They always hope for something new to have more IQ
- N - The aperture or f-stop
- t - The Shutter speed
- S - ISO setting (image brightness)
- L - Avg. scene luminance (illumination or a flash)
- Sensor pixel diameter influences sensitivity - fewer pixels are more sensitive
- Higher pixel sensitivity improves the sensor's ability to capture shadow details
- Each image sensor has a unique noise floor (noise floor size and types - Fig 1)
- More megapixels means adding noise to the Noise Floor. (pixel control circuits)
- There are two forms of noise. Shot noise and the sensor's Noise floor (Fig 1)
- When calibrating the sensor, the sensor's sensitivity is fixed/set at the factory
- High-sensitivity sensors mean less high ISO noise (low calibration multiplier)
- High-sensitivity sensors typically have a higher saturation point plus DR
- The old analog exposure triangle is not the best choice for digital cameras
See this article discussing the 7 points each digital photographer should know...
2. Why test extremes like the Sony A7 III and Olympus Pen F?
Because the difference between these 2 sensors is BIG? What happens when we underexpose the shadows while correctly exposing the mid-tones to highlights? Will "size and capture" fanboys claim it's all about DR, sensor size, and smaller sensors capturing less light, or are there technical reasons why scientists investing their time and energy to design more sensitive sensors? Are image sensors as basic as "size," or is there a technical explanation for sensor performance?
A quick reminder:- Your ISO setting does not create noise. The ISO setting amplifies the image signal and the existing noise floor of the image sensor. The sensitivity of your image sensor and the SNR at each exposure will determine how much visible noise you see in your final image.
That said, I wanted to test if my thinking process is correct, or should I repent and forever accept the "Size and Capture" theory and focus on that ONE variable, SIZE..?
3. Can we explain these results?
I have no doubt that the Sony A7S III is a fantastic camera. My son uses the Sony A7S III, his Sony A1, and RED video cameras professionally. His customers are happy with his work. My own experience with the A7S III is only positive. The Sony A7S III is a unique camera aimed at videographers.
The same is true for the Olympus Pen F. Against all odds, it has a loyal following, and many new creative enthusiasts are discovering this unique camera in 2023/24. Does it mean we should compete with the newest and most popular cameras? I really do not see any value in that..?
The reason for this test is NOT which is better, or my M43 sensor is super awesome. Each image was taken in a semi-controlled space. I upped the brightness so you can study the shadows. Olympus said the differences between M43 and FF cameras are tiny. Will we see that in this test?
The change in shadow detail between 0EV and +1EV demonstrates the changing sensor performance (saturation) between the two exposures. This level of control is only possible if you know your digital camera and how the performance of the camera/sensor works. (It is not only ETTR.)
This exercise was exciting. The SNR response is different for each camera, and the saturated and unsaturated parts of the sensor determine the final image look. "Size and capture" fanboys cannot explain these performance differences between the shadows and well-exposed areas.
4. A few more thoughts on the above test images
The two images below are the fully edited raw versions of the above +1EV images. The Sony A7S III has more shadow details, and it took more effort to recover the Pen F shadow details. The reason for this is the sensitivity differences between the A7S III BSI sensor and the Live MOS sensor of the Pen F. Another reason is my Pen F recorded less tonal data in the shadows (See the histogram). I purposely left these final images slightly "flat" so you can study the "recovered" shadow details.
Here are a few final thoughts about these images:-
- Sensor technologies - LiveMOS versus BSI (Both CMOS but different architectures)
- Technical differences - 2016 to 2020 (Much development happened in these 4 years)
- Sensor Sensitivity - Sensor evolution focuses on Quantum and Optical Efficiencies...
- Sensor sensitivity - Sony selected a super high-sensitivity BSI sensor for the A7S III.
- Pixel Size - It makes a difference + the delta pixel area is the highest for this example
- Sensor Noise Floor - The A7S III sensor benefits from having a smaller noise floor
- Sensor Noise Floor - The BSI sensor + four years of R&D improved the noise & eff.
12 comments:
I totally agree with your comments about today's marketing. I think it was Canon that started most of the "bigger sensor = better" marketing. Mind you, others probably did the same in the film days.
What would be interesting in this comparison would be to use continuous lighting and shoot at the native ISO without clipping anything. After, raise the shadows in post and see what happens. My bet would be that the A7S III would do better in terms of noise due to a bigger pixel pitch, nothing to do with the actual size of the sensor. Also, it shoots 14 bit and the Pen F only 12. Wouldn't that also make a difference? My feeling is "yes", but not as much as one would think.
Hello John, again thanks for your support.
I had the A7S III only for one day. My time was very limited and I had to be clear with my test sequence. I purposely desaturated the sensor. It is so important that each photographer learn how to manage the image sensor. I did nothing to improve the images because I wanted to study the "raw" results...
12 Bit versus 14 Bit is not something I spend enough time on to date... will do it in the future. I read the comments and have my own opinion but its better for me to really prep before I say too much... :-))
The A7S III was ready for a job and came in its SmallRig cage. I had the option to use the f2.8 35mm or the f1.4 35mm. In both configurations, this was a huge and heavy package.
I have no doubt in my mind that I will be able to get awesome results from both the +1EV images if I do the effort. Will I advise my son to replace his A7S III with my Pen F because I like it? Absolutely NOT, the Sony is perfect for what he needs it. Does it mean the Pen F is really bad in shadows? Again absolutely NOT, the Pen F is surprisingly capable when "managed" optimally...
Will the A7S III do better in the shadows? YES absolutely, its sensor is ultra-sensitive (not only because of popular marketing reasons) and it was designed to deliver good shadows for video. This is a specialized camera and one can see that from only going through the menu...
Does it have the color control of the Pen F in the camera or Live View and Enhanced Raw Files in WorkSpace? Again absolutely NO, the Pen F is a specialized camera for the creative photographer...
Both are awesome cameras...
Best
You have bought into "the lie" a bit too. The folks (mustn't name names) who work out sensitivity and low light response divide the numbers by the ratio of the area size to that of a 24x36 full frame. Because they don't know any better or they favour full frame. There is a code of practice that says do not do that. For Fujifilm it is 2.35, for MFT 4/3 it is 3.56, or multiply the MFT by 1.35 to compare to Fujifilm. The photosite (pixel) size is important as is newer/better technology. It's a tool - compare apples to apples not pears.
The smaller the sensor, the less heat is produced and that more evenly. Easier to cool too. That means less electronic noise. Poor cooling produces a magenta tinge cf Sony (oops). Olympus are more prone to magenta than Panasonic, identical sensor, better heat sink. Background physicist-engineer/radiometrologist.
Hi Jef
Thanks for your feedback. I have read it a few times and it's still not clear to me. Please give me more info about the lie I bought into. I hate to think I am giving incorrect info to my readers. That would go against what I spend so much time on to prevent in terms of an accurate and basic theory we can use to help photographers improve their camera skills.
Then please explain the ratios or numbers you quoted. I get what you write that we shouldn't use the size of the backplate (sensor), but what do you suggest we should use instead?
Then who will the code protect? Those who are tweaking technical info for marketing purposes or those supporting accurate info? Shouldn't one point out when info is used inaccurately?
Noise and heat are clear, thanks.
Best
Siegfried
Good morning!
Highly informative and detailed post and an admirable effort! Your conclusions are in line with the fact that under certain and/ or controlled conditions - sensor size and corresponding photo-site quantum efficiency make a little, if any, difference to image quality. Nonetheless, for certain applications (in my case need for high iso at relatively high speed where, e.g. IBIS is irrelevant) - sensor size does make a difference.
See:
https://bitphotospace.blogspot.com/2021/11/
I've used the OMD E-M1 which for all intent and purposes has similar DR/ SNR etc. to the PEN F (if you were to trust DXO). Compare noise and DR output (generated in rawtherapee with no profile applied) to that of a 2008 D700. See how the E-M1 "underexposes" by roughly half a stop at same conditions as means, presumably, to protect highlights together with compressed comparable DR.
For certain applications - sensor size does matter. YMMV.
Hi MB
Thanks for your input, I really appreciate your effort.
Here a view comments:
- I agree that Olympus cameras generally underexpose, does it impact 18% gray?
- I trust DxO because they do the effort and do not only talk like some forums?
- We also see how photographers benefit from the RAW capabilities of DxO...
- I guess you tested the older EM1 MKI and not any of the later models?
DxO gives us very helpful info - try your tests again by using these values below:
- D700 - use ISO 2400 with no exp comp and in Aperture Mode
- EM1 - use ISO850 and +1EV exposure compensation in Aperture Mode
- Set the WB for the scene for both - which final image has the right neutral gray?
Why the different ISO's? Simply because these are the max ISO values at which these cameras are still performing well as per DxO Sport rating. This supports your theory that the D700 is probably the better of the two for sports photography. Have you considered the original Sony RX10? Why would this be potentially better than both the older EM1 and the D700?
What do you think was more important in your test?
- The pixel or photocell area (size of the pixel)
- The size of the backplate housing the pixels (sensor size)
- Optical efficiency - (Sensor tech (BSI, MOS, CCD, only CMOS), funneling, microlenses ++)
- Quantum efficiency - electron multiplication, better materials, better capacity to read photons
- The quality of the lens you use, quality of light, and more
- Any software the camera or sensor uses to cancel noise
- The rest of the components the camera use like the analog A/D or analog signal amplification
- A good knowledge and understanding of the camera and image sensor
- Don't you think it is a little naive to single out the size of the backplate that does not record photons?
For example, do you think it's OK if forum promoters say to a person, get yourself a FF camera, and all your problems will be solved? What if that person gets a Sony A7RIV and discovers this camera's extreme noise at higher ISO's
My articles are not intended to prove one size is better than the other but to show photographers that there are much more than this oversimplified commercial "Size and Capture more Light" theory...
MB my reply was not intended to be too direct, I really appreciate your input and I believe you did the right thing to think and test for yourself plus the discussion is important.
See my follow-up article or Part 2. It should be ready in 7 days and will give more on how to manage the performance of the image sensor.
Siegfried
Thanks for the comprehensive response��
Yes full frame shouldn't be pushed as the ultimate imaging medium - and it's fortunate that excellent blogs like yours educate wannabe photographers.
Full frame is just a system like any other. Whilst it will not make you a better photographer, it fulfills specific needs including, all being equal, better SNR (the massively overkill D850 still has 4.34 µm pitch) which is a function of physics. FF is also still built into DSLRs which I prefer practically and ergonomically and gives you the ability to use reasonably priced primes in case your needs centre around DOF control (the Olympus 25 mm 1.2 pro will cost you a kidney for a 2.4 mm FF equivalent FOV, weighs 410g and 9 cm long). I am not familiar with Sony sensors as the A7 system is not my cup of tea but I presume being full frame, as you've highlighted, does not equate to better image quality at certain settings (all else being equal / some photographers may equate the ability to heavily crop at low ISO as quality).
As for the E-M1 MK1 - I used it for several years as my only travel kit (pre COVID ��) and always been extremely happy with the outcome especially the deeper relative DOF (at the equivalent FOV) at smaller apertures coupled with IBIS.
Looking forward to part 2!
Hi MB
Again thank you for taking the time. Your input inspired a few changes to Part 2 of this series. I like to stress the fact that my articles are not intended to prove one size is equal or better than the other. The only reason I am addressing inaccuracies is to give folks the chance to step away from false "facts" quoted on a daily basis on social media.
This is the only way folks can improve their skills and master their digital cameras. Yet a small part of photography it's so rewarding to get the best from your camera because you are skilled to do so.
Here is an excellent example of an inaccurate statement. You wrote:-
"Whilst it will not make you a better photographer, it fulfills specific needs including, all being equal, better SNR"
The first and foremost mistake. The backplate housing the pixels does not record any photons. The only component that is capturing photons is photocells (pixels). Marketers want us to believe "size" is the key and again that is so inaccurate. It's really the ability of the photocell to convert the captured light (luminance or photons) into electrons. This is huge. Then SNR has nothing to do with size/pixel/backplate. It's a function of design and completely independent of size. We saw that with the example between the Pen F and the Sony A7SIII. The shadows or underexposed parts of the sensor reduced the SNR. The result is more visible noise in the shadows for both sensors. The Sony did better because it has a more sensitive sensor than the one used in the Pen F. That said, pixel size is one of many variables contributing to the sensitivity of the sensor. Which one of these variables is more important? It's not size!!
Yes, it's specifics but so important if you like to manage your image sensor. Photographers really need to break free from all the marketing inaccuracies if they want to master the digital camera...
Best
Siegfried
Thanks for taking the time too!
Re. my quoted statement being exemplary of inaccuracy, not sure you've noticed the "all being equal" part - I somehow dropped the "else" :) but I presume the intentions were clear. Now this should redeem my statement out of the inaccuracy remit should it not? :)
All else being equal, the laws of physics prevail:
https://www.semanticscholar.org/paper/Roadmap-for-CMOS-image-sensors%3A-Moore-meets-Planck-Catrysse-Wandell/bae1e3b15cee0c8dc3e5f4864656173e7f0df48d
And I dont want to even go through admittedly trusted DxO sensor comparison between say the Pen-F and an older gen A7S II - the difference is just astronomical. Check it out.
One note re. your test, the Sony somehow front focused (likely missed since this is CDAF) nailing the ribbon whilst the Pen-F blurred the ribbon but I guess this has no impact on your conclusions - am I correct?
As mentioned, looking forward to seeing part 2!
MB you are all over the place, if not "for certain conditions/applications" or "controlled conditions" or "trusting/not trusting DxO" then it's "all things equal", and now it's which one was focussing correctly. In fact, I was waiting for the equivalence line to come next...
Through all that you didn't try the settings, I suggested for your test images and you just cannot get over the fact that the dead backplate housing the pixels (sensor size) cannot even read one photon. Yes, probably the most misleading theory ever in photography. (Now don't list new conditions pls) Yes, the "size and capture more light" theory was deceptive and somehow OK when this promotion was introduced 15 years back, but photographers are smarter today and they are looking for real information.
I can see you are a smart guy, but I do not want to discuss a senseless theory. I will only continue with this discussion if you are ready to constructively contribute to developing something better than the commercial "size and capture" theory or the old exposure triangle. Something that will help photographers to get more from their existing cameras (not getting them to run to the shops).
Thanks again,
I guess we got lost in translation somewhere:) thus you are right - this thread is no longer constructive!
I will continue to follow your excellent blog since am still an avid MFT (Olympus) user despite having drifted to full frame use in recent years.
I appreciate your support. As said, our discussion was helpful and I really thank you for that. The content for part 2 is clear in my mind, but as you pointed out, English is a challenge and takes up a lot of my time. For example, I spend good time on research, study, and experimenting, but instead of the writing being a small percentage, I spend most of my time writing and making corrections...
Post a Comment