Last updated:- 26th August 2023
Introduction.
What are the benefits of using ISO Low, L100, and L64, and does the camera's Dynamic Range (DR) change at anything other than its Native ISO? These and other questions initiated my journey to better understand digital cameras. I hope this article inspires you on your journey.
How much should we know about the camera's Exposure Mix, SNR to ISO ratio, and Image Sensor? Do they improve our ability to manage the performance of the sensor? For example, does the sensor receive enough light at ISO3200, and how do we measure the light exposing the image sensor?
We know sports photographers need higher shutter speeds. For example, they will up their ISOs for higher shutter speeds in sports arenas. It's normal to have higher ISOs in good ambient light. Please take a moment to think about the following questions:
- Do sports photographers risk having more noise in sports arenas?
- How do we know the image sensor received sufficient light?
- When will you typically consider using a flash?
I am discussing the following:-
- Introduction
- A brief description of the digital camera
- A brief overview of the image sensor
- A brief description of the ISO function
- The solution to the challenge in Part 1
- ISO-Low, L100, L64, & Dual ISO
- Why do we have ETTR?
- Conclusion
2. A brief description of the digital camera
Digital cameras have analog and digital parts. The sensor's electronic (analog) components increase the sensor's noise floor. Some electronic components add more noise than others. This article focuses on the analog part of the camera and image signal.
The camera's digital section isn't our focus in this article because the digital image signal differs from the analog image signal. Digital noise is also treated differently by the camera...
Which analog components have more noise? The 3 components with the highest noise levels are the Image Sensor, the ISO Amplifier, and the A/D Converter. We do not control the A/D converter, but we do control the ISO and the reflected light exposing the image sensor.
Figure 3 explains the image signal pathway from the subject to the SD card. The aperture and shutter speed control the reflected light from the lens to the image sensor, and the ISO amplifies the recorded image signal from the image sensor. The following illustration explains the role of each of the main components in the image signal path.
Our focus is on the Sensor and ISO Amplifier because our exposure strategy impacts the SNR of the image signal. How does this happen?
- Shutter Speeds and Aperture determine how much reflected light the sensor receives,
- The ISO function amplifies the Image Signal plus the image sensor's noise floor.
3. A short overview of the image sensor
Theoretically inclined photographers will value the articles below.
Please study the first 3 for this discussion...
- An article I wrote about DxOMark and IQ - link
- Evaluating different sensor applications - link
- Studying temporal noise and sensitivity - link
- Sensor discussion for mono cameras (PDF) - link
- Sensor discussion for color cameras (PDF) - link
- Interesting read on contrast limitation (PDF) - link
The following illustration helps us to master the digital camera and the image sensor. Each sensor has a unique photons-to-electron graph (Fig 1). Variables like AST, temporal noise, sensor gain, and DR are specific to each image sensor. Each sensor's photon-to-electrons graph is plotted with a fully saturated sensor. The photons to electrons graph are designed to assist us in planning, analyzing, examining our highlights and shadows, and determining local or global exposures.
How do we apply local or global exposure adjustments? The Gradation function is one option, gradient ND filters are another, and a flash or LED light source is another.
- Plan and select the right camera with an appropriate sensor
- Plan and execute technically challenging photo sessions
- Evaluate and analyze camera performance & images
- Improve their exposure and problem-solving skills
The information in this article is not intended to radically change your photography. The data was carefully selected to help you improve your knowledge of digital cameras. This article is more advanced but supports better exposure skills and higher image quality.
What about the Fuji X-T4 or the Olympus EM1 III? The same information we discuss in this article applies to cameras like the Fuji X-T4 and the EM1 III. Studying the characteristics of these two cameras and using theoretical models like the photon-to-electrons conversion graph enables us to determine that both the X-T4 and the EM1 MKIII have specific strengths and are excellent tools.
One of the most exciting differences is their image sensors, sensor sensitivity, and the noise floors of these sensors. It's new for crop sensor photographers to consider theoretical models because they've been taught to only think of sensor size, IBIS, ISO values below f5.6, and lens diffraction...
How do we analyze the above image? I overlayed an example from this article onto the photons-to-electrons graph (Fig 2). The sensor was not fully saturated, the SNR was lower, and the noise increased in the shadows. This process of analyzing images helps us to develop a good feel for image sensors. What can we do to improve our image data in the shadows?
A less saturated sensor (lower SNR) and higher ISOs are the main reasons for noise. This reminds us of sensor sensitivity (image detail) and fully exposing the sensor (higher SNR). The Sony A7S III delivers good shadow detail because BSI sensors are more sensitive. Sony A7S III cameras are known for having a smaller noise floor or less noise to amplify at higher ISOs.
4. A short overview of the ISO function
The ISO function was not designed to change the gain of the sensor. The sensor's native gain is set at the factory. The ISO value is one of 5 variables calculating the camera's exposure. As seen in Fig. 3 & 4, the physical role of the ISO function is to amplify the image signal from the sensor. The image sensor is usually underexposed at higher ISOs. This ability to quickly review your camera settings and the relationship between the ISO and SNR is vital.
What is Maximum Sensor Saturation, DR, or SNR? The sensor reaches its full potential or has a fully saturated sensor, DR, and SNR when the reflected light is enough to fully expose the sensor. This is also the point at which the Photon to Electrons graph is plotted for each sensor.
"Size and capture" promoters claim "one sensor receives more light than another." This is repeated in almost every camera review. Universities teach that pixels are designed to saturate when they receive the right level of reflected light. The amount of reflected light reaching the sensor is linked to the ambient or available light and your camera settings. Photographers determine how the sensor performs and NOT the size of the image sensor...
Source: DPReview.com
Photographers decide how much reflected light exposes the sensor. How do we control the light? With the aperture and shutter speed. The key is to select an ISO for the scene and to manage the reflected light with the aperture and shutter or an external flash. Each photographer should know this. Why should we fix the ISO? We like to prevent the ISO function from changing our Exposure Mix. See this article for more about the histogram and ISO.
How did I create the following illustration? I did not change the ambient light..?
One can safely say the Exposure Mix of the camera determines the "working" Saturation and SNR of each pixel. For example, the displayed exposure on the camera will stay the same when we increase the ISO with 2 stops and decrease the reflected light by 2 stops. This means the image brightness stays the same, but the sensor receives 2EV less luminance. We basically managed the SNR to ISO ratio. The result is a less Saturated sensor with a smaller SNR. Try and plot this on the photons to electrons graph in Fig 4. This is an example of managing your Exposure Mix. What will happen to our image quality in this example? See the summary below:-
- Less reflected light means the sensor is less exposed (Aperture & shutter speed)
- That means the sensor is less saturated, and the image signal has a lower SNR
- A lower SNR with higher ISO amplification means we have more image noise
- It means less light with fewer details in the shadows and more shadow noise
You need to carefully plan your Exposure Mix. You are responsible for your image quality. A better way of determining the exposure is to start with the old exposure triangle. The next step is to know what you want and to manage your final exposure mix (ISO to SNR ratio).
When do we use higher ISOs? Photographers use more ISO amplification for higher shutter speeds and too little luminance to expose the sensor. See the summary below for why we need higher ISOs:-
- With low-intensity illumination or available light and no available tripod
- Indoor sports events, artificial lighting & high shutter speeds are needed
- Outdoor or normal lighting conditions and higher shutter speeds (wildlife)
Think carefully before using any of the Auto Modes. We cannot manage our sensor performance with the ISO in Auto Mode. The reason is the ISO will change randomly in Auto ISO mode. It is only possible to control the exposure mix (SNR to ISO ratio) with the ISO in Manual Mode.
5. The solution to the challenge in Part 1?
I asked readers to create 3 illustrations. Nobody did the challenge, which made me wonder, was it too advanced, or did I give too little information? The 3 illustrations are educational because they show the positive and negative image effects of applying the 3 practical ways of managing the image sensor's performance or the relationship between the ISO and SNR.
Below are 3 ways of managing the performance of the sensor:-
- Increasing or decreasing the sensor SNR with luminance (aperture and shutter Speed)
- Amplifying or lowering your image noise (the effects of increasing ISO amplification)
- Controlling the image sensor Saturation and SNR with an external light (using a flash)
How do these 3 methods impact the camera's image quality? A less saturated sensor means a lower SNR, less tonal data, and more visible image noise. A fully saturated sensor means good image quality.
How did I create these illustrations?
The 3 illustrations came to life when I searched for a simple answer to illustrate the benefits of managing the image sensor performance. I placed an A4-sized white paper against the wall and my older Olympus E-30 on a tripod. The illumination in the room stayed the same and was the primary light source. I created the 3 illustrations with this setup.
Why did I use my older Olympus E-30? Older cameras have more noise (noise floor) with more noise elements than newer cameras.
5.1 Controlling the SNR of the image signal (Aperture and Shutter Speed)
I created the above illustration from a constant image signal amplification (ISO3200) and only controlled the luminance exposing the image sensor. The 0EV sample is 18% gray, and the rest of the 5 images vary +/- 1EV steps. This changed the reflected light intensity on the image sensor. Each of the 5 exposures changed the Saturation and SNR of the sensor. We can plot these 5 images on the photon-to-electrons graph. The final brightness of each sample was adjusted in PS.
The -2EV sample simulates an underexposed image sensor. That means sensor saturation is low at point A in Fig 1. The SNR is also smaller at -2EV and, therefore, the increased noise.
The manufacturing segment likes more information about the types of noise in the sensor. We should learn from them? For example, it's nice to have new AI noise reduction, but why not manage the visibility of your sensor's noise floor? The -2EV sample gives us a view of the different noise elements in the sensor noise floor. We know shot noise looks different from temporal noise. The +2EV sample shows the benefits of saturating the sensor.
What did we learn from this example? Apart from what I discussed, "managing" your image sensor's performance is the same as controlling the visible image noise. This means we can now make informed decisions about image quality or when buying new cameras.
What are the two most important things about the image sensor?
- Sensor Sensitivity - This determines how image detail the sensor records
- Sensor Noise Floor - Every image sensor comes with a unique noise floor?
5.2 The effects of ONLY using the ISO (image signal amplification)
I tested a different way of managing the image sensor. In preparation, please study Figures 3 & 4. It is critical to understand that even though the ISO is just another variable in the exposure formula, it's physically positioned between the image sensor and the A/D converter. In other words, the image sensor records the image and sends it to the ISO amplifier. We also see the effects of ISO amplification on the image sensor's noise floor.
How do we manage the image sensor with the ISO setting? Go back and study the first illustration. You will recall I kept the ISO constant for the 5 samples. That means some were brighter and others darker. I adjusted the brightness of each image (like ISO amplification) in Photoshop. Why do we increase the ISO? It is only necessary when we have too little illumination (underexposed image sensor). Like Photoshop, the ISO helps us to "correct" our final image brightness. It's critical to practice with the SNR to ISO ratio and this concept of managing your Image Sensor...
How did I create the illustration in Fig 8? I kept the luminance (aperture and shutter) constant to see if the noise would increase at higher ISOs (amplification). That means the saturation and SNR of the sensor stayed the same. This example shows the effect of amplifying the image signal and the sensor's noise floor. The ISO400 sample is correctly exposed, and the 18% gray version,
What do we learn from this illustration? This is something all photographers should know. The noise increases when we increase the ISO, right? We saw that the SNR of the sensor changes with a different exposure mix. The ISO amplifies both the image signal and the noise floor.
Folks, we basically learned 3 things:-
- The noise floor and the noise elements are linked to the design of the image sensor
- Thinking and informed photographers will use a suitable sensor for the application
- There's a tight balance between managing the SNR (Aperture & shutter) and the ISO
5.3 Using an external light source to manage the SNR
The above illustration is similar to the first. The difference is I used an LED light source to increase the illumination of the A4 paper. I then used my shutter speed to manage the sensor's saturation. I used the same ISO and controlled the luminance reaching the sensor with the shutter speed.
What do we learn from this example? Light and the ability of photographers to work with light are critical. The right level of illumination is vital for managing image sensor performance. Light determines if the image is flat or multi-dimensional. For example, anyone who says a tripod or an external light source (flash) is old-school should re-think.
6. ISO Low, L100, L64, ETTR, and Dual ISO
Most electronic components have a Linear Operating Range. The linear operating range for an image sensor is between points A and B in Fig 1. This means the photon-to-electron graph is linear and is predictable between points A and B. For example, a hundred photons will result in a fixed amount of electrons between points A and B.
The native ISO of the sensor is fixed with the sensor at full saturation. This means the DR will not increase below the native ISO of the sensor. The user's manual says the DR will decrease below the native ISO. While this is technically correct, ask yourselves, is this the only interesting information, or do these lower ISOs also help us to improve our image quality? Yes, they should because a lower ISO means less image signal amplification.
Why should we use ISO Low, L100, and L64? They offer better image quality because they use Less Image Signal Amplification. The aim of managing your ISO versus SNR ratio is to improve your sensor saturation and the SNR of the image signal. What about dynamic range? Dynamic range is a common weakness for all digital cameras. The solution is to learn how to work with the camera's dynamic range and the experience to make the right decisions for each scene. Typical options are the time of day, bracketing or HDR, Polarizer or ND filters, Log profiles, fill-in flashlights, and more...
Maria Schwartzmann used a unique example of why to use ISO64 on the EM5 MKIII and the Olympus LIVE COMP function - link (start at the 1:06 minute point).
How much less amplification do we have at L100 or L64:-
- ISO L100 is equal to ISO100 which is 1EV (Full Stop) less amplification
- ISO L64 is equal to ISO64, which is less than a full-stop ISO amplification
7. Why would we use ETTR?
What happens when we use ETTR? It simply means we will saturate the sensor. ETTR improves the SNR and DR while "shifting" the histogram to the right. The next step is to reset the 18% gray exposure values in WS. The benefit of using this technique is the sensor is at maximum saturation, DR, and SNR. The noise floor is less visible at both the native and higher ISOs. ETTR also improves the tonal data because the histogram shifts to the right. (Study this article at Adobe)
What if the DR quoted in camera reviews differs from the manufacturer's measurements? How do reviewers determine the DR? Do they measure the dynamic range, or do they calculate it? In other words, do they consider the DR headroom of each camera? Could we trust any DR information from camera reviewers pushing the "size and capture" theory?
What are the advantages of using older cameras? Apart from 2nd hand pricing advantage, the main benefit is more creative and artistic freedom. Older sensors are a joy when doing creative photography. If you like a camera with more creative freedom and sensor control, get an older Olympus camera. The IQ of M43 cameras like the EM1, EM-5, and E-P3 benefit most from ETTR.
Should we use ETTR at higher ISOs? We have seen that higher ISOs are used in low-light situations. In other words, the sensor is underexposed (between points A and B in Fig 1). Does that mean the DR of the sensor changed? The solution is to focus on the best exposure mix or use ETTR to get more light to the sensor. Everything we discussed in this article applies to higher ISOs.
8. Conclusion
The "size and capture" theory cannot explain shadow noise or the differences between two 24MP full-frame sensors. Paid "size and capture" promoters will not discuss the evolution of image sensors because they cannot explain any of the improvements with the "size and capture" theory. They end up repeating the same lines they receive from their marketing managers. The info in this article is a theoretical explanation for the sensor. It helps us to answer almost any question about image sensors.
Many photographers prefer a more creative photography style. They will use their exposure settings to control the look and feel of the image. A personalized style is good because it's the artist's prerogative. This explains the popularity of cameras like the Pen F and E-P7. The info in this article supports creative artists plus those photographers asking for the best possible image quality.
Many support the idea that image editing should be a part of the image-taking experience. Olympus cameras make it possible with the Enhanced Raw Format and Workspace.
Also, see the more complex scene I discuss in this article.
Take care and God's Bless
Siegfried
You will find more information in the articles below:-
Various References:-
- Equivalence theory for cross-format... - link
- Great video with Olympus photographers - link
- Interesting video discussing the OM-1 and exposure - link
- Understanding luminance for more control - link
- Great article and video on how to work with light - see the link
- Understanding camera performance Specs - link
- Fascinating presentation explaining sensors - link
- How to evaluate camera sensitivity - link
- EMVA Overview - Imaging performance - link
- Temporal noise reduction in CMOS sensors - link
- White Paper on the optical efficiency of the sensors - link
- The basics of the Low Pass or Anti Aliasing filter - link
- More about the image sensor and the EMVA1288 - link
- More about Low Pass filters and why they are used -see the link
- A general page on consumer camera sensors from Sony - link
5 comments:
I can tell you the GH6 has been a different beast as it's not ISO invariant. It's been a bit of a learning curve coming from Olympus.
Hi John, it's nice to see a comment from you again. Congrats with the GH6. Seems like the 25MP Live MOS sensor is a really interesting sensor. The image processor is also 2X faster. I saw the size capture sanctuary (dpreview) like people to think it could be a BSI sensor... interesting.
It seems the GH6 is a little complex, especially the video part of the camera. I studied the info on the Panasonic website. A truly interesting camera. Are you doing video or mainly photography?
Best, Siegfried
The GH6 has been a completely new adventure for me. I do shoot some video and I like learning more about the capabilities.
The main features of this sensor are:
1) 5.7K resolution at 120fps (the 120fps video is downsampled from that, unlike most FF sensors that pixel bin and line skip to get that speed)
2) Handheld 100 megapixel shots with voodoo-like magic processing.
The files are ginormous and in 16 bit containers. I believe they're probably 15-16 bit files. Before the GH6, only the GH5s had 14 bit photos.
Hi Siegfried,
Today is a holiday in Switzerland and I hope you have enjoyed it (my colleagues work in Geneva).
Thank you so much for this amazing blog! I have a similar curiosity, beliefs and philosophy, but you are much more advanced! I am learning a lot and also confirming my own suspicions and theories.
With regards to low ISO, most formal testing of sensors shows a small drop in DR, with the exception of the PEN-F, it seems. I have yet to use low ISO on mine but will be now running some tests to see if the noise floor actually improves (it should, and probably to a greater degree...I am coming to believe this camera is indeed very unique).
Even so I am less concerned about DR, and I laugh at how most photographers who brag about their high ISO/low noise and high DR cameras don't realize they are trading off DR when bragging about the quality at higher ISO. And you have taught me that not just noise but the noise FLOOR increases. What I have learned here is that it doesn't matter if you have enough light!
In terms of exposure, I also discovered through experience that Olympus tends to underexpose but also that the exposure curve while looking flat does boost the midrange, and this is a good compromise. At the same time I find there really is not that much headroom or floor which confirms this theory.
I was always impressed at Olympus auto exposure accuracy and I rarely go beyond + 1 EV in low contrast scenes and - 1/3 EV in high DR and give a slight correction in processing.
I compare with my GX85 and G9 where in the same scene the auto metering almost always shows blown highlights with the zebra pattern for the same exposure that the Olympus camera is giving me. And interestingly on my G9 I can set the zebra pattern limit to 105%, which is very telling about how much headroom there actually is in the raw file. There is a lot more shadow detail to recover in my GX85 raw files when I underexpose, but that comes with considerable noise and color shift.
My assumption is that on Lumix cameras (at least mine) metering and the exposure curve is tuned more for video than still photography.
My conclusion is that the ISO part of the exposure triangle is not a constant and varies quite a bit from camera to camera, but not the whole story as the A-D conversion processing plays a role. This makes using an external exposure meter not very reliable beyond measuring 18% gray, and I will always tend to expose for the highlights anyway knowing more or less the limit where my highlights will be unrecoverable in the raw file.
I also do not depend on the histogram in the camera as it is really tuned for the JPEG. I would actually really love one adjusted more precisely for the actual DR of the sensor instead.
My wife and I just returned from a lovely walk with the dog in nature. Thank you. I had fun filming with my EM5 II. Your feedback is exciting, and I enjoyed reading it. It reminds me of my own journey as I read between the lines. I like how you use common sense to test and experiment with your camera. I can leave you one tip. Study the photons to electrons graph I always show in my articles. For example, the sensor's sensitivity is linked to the quantum and optical efficiency of the pixels. Less sensitive sensors have a lower DR than high-sensitivity sensors. Why? Because high-sensitivity sensors convert more photons to electrons. Plot this, and you will see how the one has a higher DR than the other. It's all about sensor design and NOT size.
I will create a video to discuss the photons to electrons graph.
We can spend hours discussing this. Again, thanks for your feedback.
Best Siegfried
Post a Comment