Last Update: 27th June 2025
Introduction
This article reviews the "size and capture" theory, ISO invariance, and insights from my series on the extended ISO range of Olympus cameras. This is not an academic paper exploring ISO invariance or the "size and capture" theory, as these are commercial programs designed to promote FF cameras. As you go through this article, consider why manufacturers deem it necessary to apply sales arguments supported by the "size and capture" theory or promotional campaigns relying on astroturfing.
Also, study these articles:
The above video compares the Fuji GFX100S to the Sony A6000. While the Fuji has an MF sensor and the Sony a crop sensor, Fuji's pixel area is only 5% larger. The same Sony sensor is larger than the EM1 III sensor, and its pixel area is only 15% larger. Acknowledging optical differences between sensor sizes, such as blurring the background, we know image sensors also have technical limitations. For instance, image sensors only saturate when fully exposed, and each sensor produces noise (SNR) as a result of the sensor's so-called noise floor. I thought the video was entertaining, informative, and a welcome change to the almost daily "size and capture" or "bigger is better" promotions.
Why would marketers use undisclosed promotions?
How does the engineering community view the world?
Design engineers review each aspect of a camera, whereas marketers deliberately focus on sensor size and pixel pitch. Engineers consider effective pixel area, lens image circle, image processor, general energy needs, the camera's physical size (temperature & cooling), IBIS (size and weight efficiency ratio), and the size and weight of the camera/lens combination. In other words, the engineering community carefully considers the technical and optical limitations of each camera solution.
What distinguishes effective pixel area from conventional pixel area? The pixel wiring of standard CMOS sensors is placed above the pixel's light-sensitive area. This diminishes the effective pixel area, thereby affecting the sensor's efficiency in capturing reflected light. In contrast, BSI (backside illumination) and Live MOS sensors represent wiring enhancements over standard CMOS technology. BSI sensors achieve an effective pixel area exceeding 98%, while Live MOS sensors hover around 87%. In comparison, standard CMOS sensors have an effective pixel area of less than 70%.
See the video below for more about sensor types.
- Olympus improved the ISO performance of the EM1 II with a 3rd stop (4%). How?
- Why do we never see discussions on the Live MOS sensor, and those who ask are...
- Pixel pitch defines the resolution/quality of TV displays. Why apply that to sensors..?
- How could it be that 2 cameras with the same sensor size and resolution are different?
- Sony users can reconfigure the Sony A7 series to APC mode. What happens to the IQ?
- Photographers typically see only small changes when "upgrading" to FF cameras. Why?
- Up to 98% of all camera reviews repeat the same "size and capture" phrases/keywords.
- Why did reviewers start to include ISO Invariance in more recent camera reviews?
- Have you seen any manufacturer actively promoting ISO Invariance? Why NOT?
- Which information is repeated most in reviews? Sensor size or Sensor Sensitivity?
Did 12 years of undisclosed promotions add any value to photography?
Start by counting the times you find sensor-size "phrases or statements" on forums and in reviews. You will find that paid promoters link almost anything to sensor size. For example, how often do forum posters (astroturfers) review the benefits of having a saturated sensor with a higher SNR?
Why keywords? A well-known speaker said people accept anything if the story is big and repeated enough. This is why the mainstream media repeat "big" stories (keywords) for days. Like "size and capture," people learn to trust anything that is presented as fact and repeated enough.
"Size and capture" keywords are literally repeated like a BIG news story:-
- The BIGGER one captures more light
- The BIGGER one has more image quality
- The BIGGER one produces less image noise
- The BIGGER one improves your dynamic range
- The BIGGER one offers better subject separation
- The BIGGER one has more... really, is there NO end..?
How often do we see discussions about the optical and technical differences of digital cameras? Why do marketers lump everything into one basket while promoting the bigger is better story?
Undisclosed promotions offer nothing new. The biggest concern with undisclosed marketing is the high level of dishonesty. The dishonesty is obvious in the "size and capture" theory and ISO invariance. I am always stunned at people accepting almost anything from the same marketing techniques.
Why is this important? Oversimplification limits the photographer's capacity to fully grasp and apply their cameras. For example, did you know that we can effectively control the camera's visible image noise (SNR)? Simply purchasing a new camera with a larger sensor is not necessarily a solution.
Let's consider some basic technical aspects.
It's interesting to study manufacturers. For instance, the trend is BSI sensors with more megapixels, and one of the more intriguing areas of R&D is the physical design of the pixel (XT-5). Another critical aspect of sensor design is firmware replacing control wiring. The result is sensors with less control wiring and a smaller noise floor. For example, sensor manufacturers use "lens elements" over the pixels to improve the pixel's optical efficiency or firmware to reduce the image noise associated with more pixels. Olympus used firmware to boost the EM1 II's noise performance by +1/3 EV.
Why didn't Olympus cameras go beyond 20MP? The reasons are clear when we consider efficiencies and pixel area. More pixels mean more control wiring and smaller pixel areas. This typically adds more noise to the sensor's noise floor and lowers optical efficiency. Instead of simply adding more pixels, Olympus opted for sensors with higher efficiencies, enhanced TruePic processors, and the quality improvements that come from advancing their 16MP to 20MP sensor technologies. This choice accounts for the enhanced image quality seen in the E-M1 III, OM-1, and OM-3 models. Other factors to consider include file sizes, computer configurations, image editing, and the ideal resolution for various kinds of photographers, whether they are casual, hobbyist, or professional.
How much of a difference does any of the above make? Study this article.
One clearly recognizes the main message in the above description, especially if you value accurate information. The GH5S achieved the highest-ever M43 sensitivity with 50% fewer pixels, and Dual ISO made it possible to more effectively manage/control the visibility of the sensor's noise floor.
Why did marketers thought it's necessary to invent ISO Invariance?
- Each image sensor comes with a native noise floor
- The noise floor has shot, read, and Temporal noise
- AST is the sensor's Absolute Sensitivity Threshold
- The image sensor will saturate when fully exposed
- Sensor saturation and SNR are critical IQ elements
- Guess what? Promoters reject this information...
Figure 3. The above illustration gives more accurate reasons why image sensors are different.
What factors cause noise floor variations between sensors? The primary elements determining image sensor sensitivity is quantum efficiency, optical efficiency, type of sensor, and pixel design. Additionally, factors such as sensor architecture, image processing algorithms, firmware, and wiring can affect temporal noise. It is important to note that researchers do not overlook other sources of noise, including elevated sensor temperatures, shot noise, read noise, and quantization. It is a common misconception that larger sensors inherently produce lower noise levels.
Why do reviewers and forum experts ignore the sensor's noise floor when claiming large sensors are better? ISO Invariance is the promoter's version of addressing the sensor's Noise Floor.
Promoters like to suggest ISO-invariant sensors have more dynamic range. The reason for having more dynamic range is linked to the design of the sensor and the absolute sensitivity threshold (AST). See the above illustration. Marketers will alter any information to push a commercial narrative.
It's impossible to claim that large sensors receive more light, have no read noise, and agree that all sensors have a noise floor. How does one explain the phenomena of sensor saturation and signal-to-noise ratio (SNR) if photographers are led to believe that sensor size is the one and only image quality variable? ISO invariance replaced the theoretical understanding that all image sensors have a noise floor and/or technical limitations. In essence, ISO-invariant sensors supposedly have no noise floor, because ISO-invariant sensors are only affected by shot noise (dust particles). Why is this an unlikely scenario? Misinformation is confusing and invariably leads to further inaccuracies.
Why would they say larger sensors receive more light?
One of the more frequently used forum statements suggests that it's physics. That means we should simply accept the "science or physics." You don't need more information to accept that larger sensors receive more light. Let's have a quick look at how they visualize this "more light" concept.
What is the "Size and Capture" theory?
A quick search shows the EM5 II and the D5500 (above illustration) were launched in 2015. The EM5 II has a Live MOS CMOS sensor, and the D5500 has a standard CMOS sensor. Live MOS sensors have larger effective pixel areas than standard CMOS sensors. The D5500 has a 24MP sensor, and the EM5 II has a 16MP sensor. The total pixel area of the D5500 is 8% larger than the EM5 II, and the D5500 has 50% more pixels. This suggests that the EM5 II has a larger effective pixel area and possibly a lower noise floor. DxO test results show the D5500 has 13% more dynamic range, and the D5500 has a tiny ISO advantage in sports photography. No matter how hard you look, there is no equivalence between the EM5 II and D5500 to warrant an objective comparison of these two cameras.
Reviewers praised the sensor of the A7RII but later proposed that BSI (backside-illuminated) sensors do not significantly enhance sensor performance. The dynamic range of the Sony A7S MK3 has been enhanced by employing fewer and larger pixel areas, reducing pixel wiring, and achieving a lower noise floor (see Figure 3). This raises the question: what is responsible for the conversion of photons into electrons? Is it the dimensions of the sensor or individual pixels? Ultimately, the interpretation of this information is contingent upon its presentation to the audience.
Reviewers should provide a more detailed explanation of the "size and capture" theory. What's the basis for the claim that larger sensors capture more light? Does this suggest that larger sensors inherently have a greater capacity to attract light? It is important to clarify these marketing claims regarding "size and capture." Additionally, don't theoretical models suggest that the lens is engineered to cover the entire image sensor (sensor image circle) and that the sensor will saturate when properly exposed? I was unable to find any theoretical explanation of "size and capture" in the documentation provided by reputable camera or sensor manufacturers.
At what point are pixels too small to "capture" photons, or what is the minimum pixel area to make a noticeable image quality difference? The above DPR reviewer raved about an 8% difference between two sensors. Would a 224% difference cause a visible difference or excite this same reviewer?
The cameras in the above example had a uniform luminance perspective. Analyzing the results one sees both cameras captured enough "reflected light" (image data) to saturate their sensors.
The actual differences between these two cameras are:
- The pixel area of each pixel in the EM1 III is 224% larger than the Pentax QS-1.
- The EM1 III has a Live MOS sensor, and the Pentax QS-1 has a BSI CMOS sensor.
- That means the Pentax sensor has a 15% sensitivity advantage over the M43 sensor.
- What if the Pentax sensor has a higher native gain and requires smaller ISO steps?
- Fewer pixels on the Pentax Q means less pixel wiring and a smaller noise floor.
It's incorrect to say a large sensor captures more light without reviewing its technical attributes. It is indeed reckless to hastily conclude that larger sensors capture more light or that size is the only consideration when evaluating digital cameras. It's critical to consider lenses, sensor attributes, supplier software, and image processors. Once again, it looks like marketers often highlight promotional priorities over a nuanced understanding of image sensors and digital cameras.
2 comments:
I had a Nikon d3200 at the beginning and later an Olympus epl9. And the fact is that I used the nikon only up to ISO 400 and the Olympus quietly up to ISO 2000 and the photos are much sharper. Probably ( come to think of it rarely now) the reflection in the mirror also does its job.
You made an interesting observation. The mirror would cause shot noise. The mirror movement were also known for amplifying shutter shock.
I think many had a similar experience when trying different cameras.Those shouting sensor size are also known for bullying anyone who disagree.
The best thing people can do is gaining knowledge and trust your senses and eyes. The majority experience exactly what you experienced...
Thanks for the feedback.
Best
Siegfried
Post a Comment