Every conversation about camera formats eventually circles back to the same debate: Big sensor vs. little sensor. And while I personally do like bigger sensors, I think peoples obsession of them are actually more to do with optical characteristics and not the sensor size itself. One of my goals as a journalist in this space, writing for ProVideo Coalition and doing the Frame & Reference podcast, is to help filmmakers make educated decisions whether it be financial or creative. To make that sensor/optics distinction, I want to look past the marketing hype and focus strictly on physics.

This is going to involve a lot of numbers and HOPEFULLY I’m able to make it easy to understand by over-explaining (and in some cases over-simplifying) everything, but we’ll see how this goes.

To start, if you take a Full Frame sensor and a Super35 sensor, and both of them are native 4K, the physical surface area of the Full Frame sensor means its individual photosites are much larger.

Think of photosites as buckets. Larger buckets catch more photons in the same amount of time that smaller buckets would. This gives a Full Frame sensor a higher signal-to-noise ratio, resulting in cleaner images in low-light scenarios. These larger photosites also have a higher capacity to hold electrical charge before they clip, which inherently expands the camera’s dynamic range, giving you more latitude in your highlights and shadows.

But the benefits of the larger format become even more pronounced when you look at a high-resolution Full Frame sensor (like an 8K sensor that downsamples its image to output a 4K file) compared to a Super35 sensor recording native 4K.

When you capture a scene with an 8K grid of photosites and mathematically combine them at the camera-level to make a 4K deliverable, you are executing a perfect four-to-one integer downsample (or supersample). Every single pixel in your final 4K file is calculated by averaging a two-by-two cluster of native 8K photosites. Because random electronic and photon shot noise across those photosites is completely random, averaging them together boosts your signal-to-noise ratio and effectively hands you one extra stop of clean shadow detail.

This averaging math directly improves color fidelity and gradient smoothness; by blending those four adjacent photosites into one, the scaling algorithm minimizes digital quantization errors, interpolating the subtle values between hard digital steps and eliminates color banding in low-contrast areas like skies or out-of-focus backgrounds, effectively boosting the perceived bit-depth of your final master. Because an 8K sensor captures the optical image at double the spatial frequency of 4K, it delivers superior micro-contrast and per-pixel sharpness, often avoiding the jagged artifacts or moiré that can plague a smaller, native-resolution sensor.

However, if you attempt to cram that same 8K resolution down onto a smaller Super35 sensor, the laws of physics push back. To fit 33 million pixels onto a smaller piece of silicon, the individual photosites must be shrunk to microscopic fractions of their original size. These tiny photosites collect far fewer photons and fill up their electrical capacity much faster, resulting in less DR. Even though both formats get the same 4-to-1 downsampling math, the 8K Full Frame sensor is calculating its averages from a cleaner pool of source data. Silicon surface area always dictates the purity of the signal before the math ever touches it.

What if you don’t have an 8K sensor though? If you take a 6K image like my C500mkII and downsample it to 4K, you’re only oversampling by a factor of 2.7x which isn’t a clean 4-to-1 drop like with an 8K sensor and the math has to deal with fractional pixel mapping (e.g., merging 1.5 photosites into 1 pixel). This requires complex scaling algorithms that can introduce faint mathematical interpolation artifacts. With an 8K sensor dropping to a 4K file, the math is perfectly clean division. There is no spatial shifting or blurring across pixel boundaries.

Most digital cinema sensors use a Bayer pattern color filter array, where each photosite only captures one color (either Red, Green, or Blue). The camera must use a “debayering” algorithm to guess the missing color data for every pixel. But with a 2×2 cluster from an 8K sensor, that grid naturally contains two Green photosites, one Red, and one Blue. Because a 4K pixel requires exactly one R, one G, and one B value, an 8K sensor provides a nearly direct, un-guessed color readout for the final 4K master, eliminating color aliasing at its source.

So, because of all these aforementioned advantages, filmmakers using smaller formats often look for a way to bridge that gap. This is where the focal reducer (aka “Speedbooster”) comes in, and where the widespread misconception that putting a focal reducer on a smaller sensor “gives your camera the Full Frame look.”

Any lens projects a circle of light inside the camera body. The focal length remains the same, it’s just the imaging circle that changes when you use a lens made for “Full Frame” or “Medium Format” or what have you. When you mount a Full Frame lens natively to a smaller Super35 sensor, that smaller sensor only captures the center portion of the image circle, discarding the rest of the light. This creates a narrower angle of view, commonly called a “crop factor.” We never used to even consider Crop Factors when it pertained to cinema use because essentially all film was shot on Super35, but with the advent of digital photography people who were used to shooting “Full Frame” 35mm film photography (same thing as Vistavision in Cinema parlance) needed a shorthand to figure out what their field of view would look like compared to what they were used to. So we got crop factors.

A focal reducer makes that mental math moot by introducing an optical element between the lens and the sensor that acts as a kind of reverse-magnifying glass (Same thing as when we used to use magnifying glasses to burn leafs on the playground.) It takes that large, Full Frame image circle and optically compresses it down into a smaller, denser circle of light that matches the physical size of the smaller sensor.

This compression does two things. First, it widens the angle of view by allowing you to see the entire Full Frame imaging circle (as well as the full “character” of that lens, which is often considered preferable). Second, because it concentrates that exact same volume of light into a much smaller physical area, it increases the concentration of light hitting the sensor, making the final image brighter.

To understand exactly how this separately alters exposure and depth of field, we need to explain a critical industry distinction: f-stops versus t-stops.

An f-stop is purely a geometric measurement: the focal length divided by the diameter of the lens’s entrance pupil (in other words, the size of the aperture opening). It dictates your depth of field (the area in focus) and the physical size of your blur circles (the “blurriness” of your background). These are two separate things but are often discussed interchangeably, which adds to the confusion we’re talking about.

A t-stop measures actual light transmission: the real amount of light that makes it through the glass layers to hit the sensor. We use t-stops in filmmaking so that when we’re using a set of lenses we can effectively set exposure. Because light hits internal resistance as it passes through glass, a lens set to a geometric aperture of f/2.8 might only transmit a light value of t/3.0, so we don’t want to have one camera or setup look different than the other just because of a lens difference. Hence t-stops.

When you introduce a focal reducer, it takes your physical f-stop and virtually widens it. If you attach a 50mm f/4 Full Frame lens to a 0.71x adapter, the system is physically re-engineered into a 35.5mm lens with a geometric value of f/2.8 (even though the lens itself is set to f/4). Because that same volume of light is now squeezed into a tighter space, the light transmission also jumps by one full stop, meaning a potential baseline transmission of roughly T/4 is concentrated down into an effective exposure of T/2.8.

While a focal reducer changes your depth of field and t-stop (making it a wider lens, but a stop faster), it obviously does not transform your camera into a Full Frame sensor: Your photosites are not physically growing larger, they are not gaining a higher native well-capacity, and your sensor’s intrinsic dynamic range remains completely unchanged. Even if you feed it a brighter image circle, it also lacks the downsampling math and low-frequency noise floor of that potential “higher-K” oversampled alternative. You are simply projecting a brighter, more concentrated optical image onto the exact same small photosites, and your calculations on DoF need to be done assuming this new, wider lens.

Also, because you are pushing that lens through extra glass, the character of the background blur will not look identical to a native lens and color shift or certain types of aberration might be introduced. The speedboosted lens does not create a “blurrier” background, it just creates a different one. Again, your background blurriness depends on more factors than just “speedboosted-ness”.

All of this can be found in various books, but this distinction between geometric blur and optical rendering is thoroughly documented for free by Steve Yedlin, ASC, in some of the technical essays on his website. Yedlin’s extensive testing demonstrates that what we perceive as background blur is entirely defined by mathematical ratios; specifically the physical size of the “blur circle” relative to the subject, determined by the physical diameter of the lens’s entrance pupil.

If you match these ratios by compensating with your lens selection, aperture, and distance to subject, the geometric rendering of the image is completely interchangeable. The viewer cannot look at a final image and identify the sensor size based purely on the depth of field. This means that you largely don’t need a Speedbooster to achieve a given look, and there may very well be a native lens you could use instead, especially considering the constant deluge of new lenses hitting the market seemingly every week.

To see how this works, let’s take the physical aperture diameter of a native 50mm lens shot at f/4: that opening is 12.5mm (50mm÷4).

If you put that same 50mm lens on a 0.71x Speed Booster on a Super35 camera, it literally becomes a 35.5mm lens: by adding the glass you’re essentially just building a new lens, right? So to maintain that exact 12.5mm physical opening with our new 35.5mm focal length, our target f-stop becomes an f/2.8 (35.5mm÷12.5mm). Because the focal reducer automatically concentrates the light by one full stop, leaving your physical lens ring set right at f/4.0 delivers that exact effective f/2.8 geometry (and whatever the extra “t-stop” of exposure brightness is) directly to the sensor.

Here’s how this math applies to the two setups filmmakers constantly try to compare:

First, what happens if we shoot both of these lens combinations on the exact same Super35 camera body from the same position? Because the speedboosted setup has a wider FoV, it captures more of the background, making those blur circles look smaller relative to the overall frame. To make them match, you have to physically move the camera closer to the subject with the speedboosted setup. Moving closer increases the magnification ratio, which narrows the depth of field and expands those background blur circles until the images match. But again, this is just “matching” two images, this isn’t inherently a “full frame” look and can be achieved by simply undoing our math and shooting a native 35mm lens at “true” 2.8 (but you will have a stop less exposure, which can be compensated with your lights or bumping the ISO a stop, depending on your situation).

But what about the real-world comparison filmmakers tend to want to make: a native 50mm lens on a Full Frame camera versus that exact same 50mm lens on a speedboosted Super35 camera.

Because the focal reducer has already compressed the image circle down to the size of the Super35 sensor, this new 35.5mm focal length automatically yields a field of view that matches the original 50mm lens on the Full Frame camera. If you keep both cameras in the exact same physical position, your subject distance is identical, your field of view is identical and your physical aperture diameter is identical at 12.5mm. By leaving the aperture ring at f/4.0 on both cameras, the optics handle that math for you: the size of the geometric background blur circles will match the Full Frame shot without moving the camera at all.

But again, there’s nothing about the Full Frame camera and nothing about the Speedbooster that’s doing anything special to the look of your image beyond that extra stop of light and additional edge character: To otherwise replicate this exact speedboosted image using purely native, unadapted glass on your Super35 camera, you would simply mount a native 35mm lens and set its aperture to f/2.8. You can get the same “Full Frame look” by simply making a different lens choice.

Now, the ONE time where this changes (and what I think most people are often thinking about when they’re talking about a “Full Frame” look when employing a speedbooster) is if you try to get the ultra-shallow aesthetic of something like a native Full Frame 85mm f/1.2 lens on a Super35 camera from the exact same camera position. In that case, you’d need to find a S35 60mm f/0.85 lens. That’s probably not happening. So in that case, sure, by using a focal reducer with that 85mm you’ve achieved a look on your Super35 sensor that “only Full Frame can achieve” but ONCE AGAIN it’s because the equivalent S35 lens probably doesn’t exist, not that the Full Frame sensor is somehow letting you get that look alone. And if you were to just stop down a bit, the problem goes away and you can use a lens that’s more likely on the market.

It’s mildly pedantic but an important distinction to make because if I’m being honest, who on earth wants to shoot THAT shallow? I just tested out the Fujifilm ETERNA with my Mitakon 65mm f/1.2 and it was certainly a cool look and that lens has beautiful out-of-focus character, but is almost impossible to use practically. Same thing with an equivalent FF 50mm f/1.2 (technically an f/1.1 would be an exact match but who’s counting), or S35 35mm f/0.77 (which you could find closely enough in the X Mount 7artisans 35mm f/0.95 that I own, which Mitakon also makes a version of): all abysmal to attempt to work with in a real-world fashion. If you weren’t around for the “5D Revolution” we really beat that horse to death and while it was novel, it wasn’t practical. Plus, if Reddit is anything to go by, people are sick of completely blown-smooth backgrounds and yearn to see the environments the characters are inhabiting. All of our favorite movies were likely shot on Super35 anywhere from f2 to f5.6, and you can do that with pretty much any lens.

Finally, here’s a scenario where a larger sensor (or the speedbooster) actually is helpful from an operational and “lens reality” sense: you’re in a car, you’re in the passenger seat, you’re filming the driver with the ETERNA and you want a medium shot. You’re at f2.8 for focus reasons. You’ll probably need something like a 32mm lens. That’s easy enough to find. On Full frame you’d need something like a 25mm lens, and on Super35 you’d reach for an 18mm.

To match the look of a 32mm lens on the ETERNA at f/2.8 on Full Frame, you’d need your lens to open up to f/2.0. No worries there. However on Super35, your 18mm would need to be able to open up to f/1.4, which is harder to find and has the potential to be a poorly-calibrated wide angle lens if you’re on a budget and might have some unwanted distortion around the edges. In that case, sure, slapping the nicer FF 24mm on the Speedbooster will get you there and everyone’s happy.

BUT! And this is the other thing that I think REALLY trips people up, not all wide angle lenses distort by default, and (surprise!) it depends on the camera position.

The single most important law of optics to remember is this: Lenses do not create perspective; distance does. If you do not move the camera, the spatial relationship between the driver’s nose, their ears, the steering wheel, and the background is completely locked in stone by the laws of geometry. Because both cameras are sitting in the exact same passenger seat at the exact same distance from the driver, the light rays entering the lens travel along the exact same geometric paths. Assuming a high quality set of lenses, when the ETERNA samples that space with a 32mm lens, and the Super35 sensor samples it with an 18mm lens, they are capturing the exact same spatial grid.

In the real world, we associate wide-angle lenses (like an 18mm) with “egg-head” distortion or stretched noses. But that distortion doesn’t happen because the lens is an 18mm, it happens because of where we choose to place an 18mm lens. On a standard Super35 camera, if you want a tight close-up of a face, an 18mm lens forces you to physically shove the camera six inches from the actor’s nose. That close proximity is what causes perspective distortion (making the nose look massive because it is significantly closer to the glass than the ears).

Because our car example is a medium shot, the camera isn’t six inches from their face; it is a few feet away in the passenger seat. At that distance, the perspective is naturally flattering. Now, in the real world a native 18mm Super35 lens has to bend light lines aggressively to achieve its wide field of view. Even high-end lenses can introduce slight barrel distortion or “stretched” pixels at the extreme edges of the frame, so driver’s shoulder or the car door frame might look pulled, but once again this is a lens issue and has nothing to do with the “look” of a larger sensor, and you may not even notice unless you’re looking for it.

A focal reducer is an incredibly powerful optical tool for altering field of view and maximizing light transmission on smaller formats, and can even get you out of trouble on the extreme ends of the practical shooting spectrum, but it simply optimizes a smaller format; it does not replicate a larger one. Use your tools for what they’re intended and don’t let marketing, professional or otherwise, make decisions for you.