Last update:- 16th January 2023
While working on Part 2 of this article on ISO and Image Quality, I thought it was a good idea to set the stage with a few random thoughts and a basic challenge. Thinking about it, every photographer should develop the ability to analyze digital images. A good understanding of the digital camera and the ability to apply this knowledge benefits all digital photographers...
Taken at a constant luminance perspective and a variable image signal amplification
Taken at a constant image signal amplification (ISO3200)
You are welcome to try the following challenge. Place an A4-sized white paper against the wall and your camera on a tripod. The challenge is to recreate the above 2 illustrations. The info needed to create a basic plan, take the images, and build the final illustrations, is all in this article.
Olympus Pen F with 25mm f1.4 Leica, ISO80(Low), f3.5, 1/1600 - Edited in DxO PL-4 (See more info further down...)
Here are a few general questions for you:-
- Prep a short explanation of what happens inside the camera for each illustration
- Think of a few examples and list the benefits of knowing your digital camera...
- Why do you think it's safe, or not safe to use the ISO Low, L100, or L64 options?
- Most social media experts tell us it's not OK to use ISO Low, L100, or L64, why?
- Which of the 5 images in each of the above illustrations are 18% gray samples?
- What is the link between the Zone system, 18% gray exposure, and the ISO setting?
- Study the photons/electrons graph below. Does it apply to all or only some sensors?
For more on how to plan your own strategy, study these articles:
A few general thoughts...
The reason photographers should distrust any sensor size references is it's normal for digital cameras to have image noise. What determines this image noise? Most photographers are never told that all sensors come with a native noise floor. Should we trust those reviewers who promote sensor size or write biased camera reviews? This is likely the main reason we don't see discussions about advanced digital photography techniques, like how to use ISO amplification correctly, or how to manage the performance of the Image Sensor. (See this link)
For example, why was the old-school Exposure Triangle never improved? Especially while it's used to train photographers on digital photography? How will they ever master advanced digital camera skills like SNR, sensor saturation, or image signal amplification with an outdated triangle?
Is size a reasonable measure for IQ? We know pixel area (size) is one of many variables to impact the Optical Efficiency of the image sensor. So why focus on only one of many variables? Well, looking for answers is like finding a needle in a haystack. A more reliable way of rating image sensors seems to be Sensor Sensitivity (Optical and Quantum Efficiency).
To illustrate the oversimplicity of the "size and capture" theory, study the illustration below. This illustration offers more information about the image sensor, the noise elements in the sensor noise floor, and the effective dynamic range of the sensor. Other than the "size and capture" theory, which cannot explain shadow noise, those who master the principles illustrated below will have a strong theoretical foundation. They will improve their analyzing and sensor performance skills.
For example, take a moment and consider the graph below. The horizontal axis is the reflected light or photons hitting the sensor. The vertical axis represents the converted electrons. The sensor's full saturation capacity is reached with a fully exposed sensor. Plot the saturation for shadows or low-light scenes. How does this impact the performance of the image sensor? What happens to the SNR in the shadows? What does the histogram look like for an under-exposed sensor? These are simple questions every digital photographer should be able to answer...
Does the size of the sensor backplate "capture" photons? The answer is NO! We know pixels capture photons and pixels (photocells) convert photons into electrons. This is the main reason why scientists improve pixel (photocell) sensitivity and why they don't design bigger sensors. That said, the size of the sensor does play a role. Any idea what? Think of image effects like background blur.
Olympus photographers are familiar with 12MP or 20MP (MFT) sensors. The pixel diameter of 12MP sensors is almost double that of 20MP sensors. We know the EM1 III has one of the most sensitive M43 sensors and delivers far superior IQ to any of the older 12MP MFT sensors. Ever wondered why? Could one of the reasons be, that sensors with lower Temporal Noise have cleaner images?
Study DxO Mark results for the EM1 II sensor.
The more we learn, the more we see what happens with image quality...
Another illustration with info on how to manage the sensor at ISO3200.
Let's talk about the physical size of mirrorless cameras? The size of the image sensor influences the physical size of the camera? The reason is the lens image circle needs to cover the full sensor. This impacts the size of the lenses, the camera energy needs, heat management, and the effectiveness of features like IBIS. Digital cameras are basically built around the image sensor. The penalty for cutting corners is overheating, lower efficiencies, and less reliable cameras and lenses.
Separately from any fixed mechanical design criteria, scientists focus on materials and the electrical design aspects of creating more sensitive image sensors. This represents a better way of designing new cameras and improving Sensor Sensitivity. For example, typical improvements in image sensors include replacing older wired functions with modern software or AI solutions...
As you know, Olympus and Panasonic were the first to introduce mirrorless cameras. Did they also establish the mechanical design benchmark for mirrorless cameras? For example, what is the built-in safety margin on M43 cameras? When you see similarly sized APC or FF cameras, does it mean the M43 camera is over-designed, or are these APC and FF cameras under-designed?
How much image noise is added to the noise floor for each 1-degree increase in temperature..?
Try this quick experiment and point a light source to your PC. Which of these sensors is receiving more light?
If someone says one sensor captures more light than the other, then I cannot help to think, is this statement theoretically correct? I was searching for information when I saw this review. I could not help asking, is this just another Undisclosed Promotion? What if the "more light" benefit was only 0.0002% while those bigger sensors were 10% less efficient? One would like to think, it's all about the efficiency of the sensor when converting photons into electrons, right?
See this discussion. It's a great example of why photographers should push manufacturers for better information. Also, do a quick search on the implications of "Undisclosed Promotions"...
Final comments on the two images in this article
Take a look at the 1st image in this article. I have set the exposure for the bright areas (sky). I wanted the sky with darker shadows. At home, I did a quick test to study the visible shadow noise when I increased the shadow brightness. Editing the raw file in PhotoLab 4, it was possible to extract cleaner image details from those same shadows. Does that mean the image had enough available information in the shadows or is it only PhotoLab doing a great job?
The above example shows the jpeg on the left and the edited raw version on the right. The image was exposed for the shadows, which over-saturated the sensor in the bright areas. It did not clip the highlights while pushing them hard. I tried different editing techniques to get the most from this "data-rich" raw file. The most pleasing result was editing the raw file with Aurora into an HDR image. Did I push the image sensor too hard, or is it OK when we push the image sensor?
The selected images demonstrate the different technical aspects discussed in this article plus it shows it's safe to work with ISO Low on your Olympus Pen F. The same is true for ALL cameras. Don't we benefit more from working with a fully saturated sensor and resetting our final image "brightness" in Workspace? Why is there a link between the camera (Live View) and Workspace? Why sensor size and then push restrictions like don't use the extended ISOs on your M43 camera..?
Finally, what's better, exposing creatively, or saturating the sensor?
4 comments:
The correct exposure or ETTR have a large effect on the noise as the does the available light. For instance if you shot a well lit scene at 200, 400, and 800, Each correctly exposed, there will be very little noise difference. It is when shooting a darker scene that the ISO differences show up.
About the low ISOs, for Olympus cameras. ISO 200 and above apply a curve to the raw data. This raises the dark levels and lowers the highlights with a smooth curve to display dynamic range. This is similar to negative film. (Although Different in degree) Negative film always had more dynamic range than slides. ISO 100 and 64 remind me of slide film. It does not have the curve applied to sensor data. If the histogram is not end to end, then ISO 100 and 64 have a wider histogram. It requires more precise exposure. What it gives is more data and contrast across the photo giving a touch of that wonderful slide film look. There is also more data to work with than a ISO200 shot. This gives a sharper looking photo and a color pop that I cannot get with ISO200 or above. There are times when I need the flatter look of the higher ISOs. But when I do not, I love the look of the low ISOs. Very many times I do not need more dynamic range, so I chose the more dynamic photo. Just as I would always shot reversal film as a pro in the film days, I shoot 100 for the natural looking dynamic photo with out the processing (either in camera or external) that give that digital look. Although I also reject the RAW profiles that Olympus gives us and use linear profiles to regain the data those profiles lose. Having all of the data to start then gives me the latitude to chose the look and compression of details and dynamic range that I want. And when someone comments on the haloless sharpness of some of my photos, I do not even try to explain about starting with a better RAW because they would get lost unless they could see the 2 different starting points. The sensor gives us more information than the RAWs Olympus provides us. Especially as ISO 200 and up.
Thank you for the very interesting and informative comments. I enjoyed reading more about the link you made to analog photography and how you experienced the ISO options below ISO200. I think your input can be so helpful to readers. One of the reasons is with Part 2 of this series, my aim is to train MFT readers, and therefore my presentation style will be a little different from the hands-on experience you shared.
I was also interested in your description of taking the image and how you remove the Olympus profiles from the raw (I haven't done that myself). I guess we can talk for hours...
So thank you again so much for taking the time to comment.
Siegfried
Siegfried, you brought up some interesting points in this article. The question is simple: how to get the maximum information out of the sensor (regardless the size). To do this in ISO invariant sensors (most of the modern ones), I simply expose for the lowest possible shutter speed I need with the native ISO (at a given needed aperture). If I cannot get the needed shutter speed (at a given needed aperture), I need to increase the ISO (thus, ETTR). However, I've noticed the ISO doesn't matter with ISO invariant sensors upto ISO 6400. When shooting RAW, you can increase in post with the same result. In fact, ISO is irrelevant. If you NEED more light, you simple need to bring it in (with a flash or reflector)... or you need to make a compromise on either the shutter speed or aperture. Is my logic flawed?
Hi John, thank you for your input. I appreciate your (regardless of size) reminder. It's a very difficult subject to discuss. If we look past that (myself included) and we consider only the image sensor as a recording device, then it's so interesting.
I started a new article a while back on ISO invariance. I could not find any mention of ISO Invariance from any manufacturer. I thought that was interesting. I read the Panasonic GX800 used one of these sensors, tested one, and wasn't excited with what I saw...
I like to add one more comment. I think we see two scenarios, one is casual photography. The other is a client demanding the best possible IQ, color accuracy, and more. My two-part series will benefit both scenarios but are leaning towards the second.
A few comments on your image taking process:-
- I fully agree with your process of doing a quick evaluation of the scene. I think this is so important and something that is neglected in the mobile phone age.
- You covered the key elements in preparing your camera for the image. Let's focus on the following:-
- The sensor - we control the sensor with the aperture and shutter speed
- The ISO - Select the ISO you need and optimize the sensor for that ISO
- Using a flash/reflector - this is so relevant thanks, one of my next articles
- John it's good to develop a winning strategy. My journey on this path has been nearly 3 years and I am always excited to see the info I am sharing work. In other words, each week I am discovering something new within the framework I am presenting on this blog.
What about the 2-step exposure technique, the graphs, and the illustrations I share? These all form the basis of mastering the digital camera or the image-taking process.
That said, a quick comment on the challenge in my article, I think you will appreciate it...
- 1st illustration - only the ISO changed demonstration the effect of image signal amplification
- 2nd illustration - only the shutter speed change demonstrating how to manage the sensor
- 3rd illustration - the image on the right was taken using an LED light - ETTR using ext flash :-)
I hope this is of value for you...
Siegfried
Post a Comment