Daniel sent us this one — he's been digging into the manual controls on his phone's camera app, and he's staring at those three sliders: ISO, shutter speed, and exposure. The question is how to balance them, and whether smartphone photographers need to think about this differently than someone with a DSLR. And the answer to that second part is yes — emphatically yes — because most of what people learn about the exposure triangle from traditional photography will steer them wrong on a phone.
It's one of those things where the interface looks familiar, so your brain imports all the rules you've heard — keep ISO low, follow the reciprocal rule for shutter speed, expose to the right — and then you look at your photos and they're somehow worse than what the auto mode was giving you.
Which is the quiet tragedy of pro mode. Someone hears "real photographers shoot manual," they flip the switch, they apply DSLR logic to a sensor the size of a lentil, and they walk away thinking they're bad at photography when actually they were just using the wrong playbook.
The timing on Daniel's question is perfect, because the hardware has quietly drifted into territory where the old rules are not just slightly off — they're actively destructive. Smartphone sensors have been shrinking physically while megapixel counts have ballooned. The iPhone 15 Pro uses a 1 over 1.28-inch sensor — that's roughly one-tenth the area of a full-frame sensor. Meanwhile it's packing 48 megapixels onto that tiny slice of silicon. Each individual photosite is minuscule, which means it collects far less light per pixel at any given exposure setting.
When someone sets their ISO to 100 and their shutter to 1 over 500 because that's what they'd do on their Sony, they're starving an already light-hungry sensor, and the computational stack has to scramble to pull something usable out of the noise floor.
And that computational layer is the other half of why this is different. On a dedicated camera, manual mode gives you direct control over three physical parameters — aperture, shutter duration, and sensor gain. On a phone, you're moving sliders that feed into a pipeline that's also running HDR stacking, multi-frame noise reduction, semantic segmentation, and tone mapping. The number you set isn't necessarily the number you get.
It's more like making a suggestion to a committee.
A very opinionated committee that thinks it knows better than you, and honestly, sometimes it does. Apple's Deep Fusion and Google's HDR Plus are doing things in auto mode that manual mode literally cannot replicate, because they're blending multiple exposures taken at different settings before you even see the preview.
The project here isn't "learn to shoot manual like a real photographer." It's "understand what these three parameters actually do on a phone sensor, so you know when manual control helps and when you should just trust the robot.
That's what we're going to walk through. The physics of tiny sensors, why ISO behaves differently, what shutter speeds you can actually hand-hold, and how to balance all of it for both photos and video.
Let's define the three parameters as they actually work on a phone. ISO is analog gain — the sensor amplifies the signal before it hits the analog-to-digital converter. Shutter speed is how long each photosite collects light. And exposure compensation is a brightness offset applied after the metering system has already made its guess. On a DSLR you might also have aperture, but on a phone that's fixed — your lens is permanently wide open at something like f/1.
Which already tells you the phone is making a different tradeoff. A fixed wide aperture means shallow depth of field isn't really an option without computational portrait mode fakery, but it also means the sensor is getting as much light as physically possible through that tiny lens.
And that tiny lens feeds a tiny sensor. The iPhone 15 Pro's 1 over 1.28-inch sensor has roughly one-tenth the surface area of a full-frame 36 by 24 millimeter sensor. When you set both cameras to the same ISO and shutter speed, the phone is collecting a fraction of the light — not because the settings are wrong, but because the bucket is just smaller.
The same ISO number means very different things on the two devices. ISO 800 on a full-frame Sony is a mild inconvenience. ISO 800 on a phone sensor is already introducing chroma noise and banding in the shadows, because you're amplifying a much weaker signal.
This is where the computational layer gets interesting. In auto mode, your phone isn't taking one photo — it's taking a burst of underexposed frames and merging them, or capturing a long exposure and a short exposure simultaneously and blending them. That's HDR stacking, Deep Fusion, Night Mode — they're all variations on "collect more total light across multiple captures and combine intelligently." Manual mode typically gives you a single frame, which means you're opting out of the phone's best trick.
Which is the paradox. You switch to manual because you want more control, but you've just disabled the multi-frame pipeline that makes the tiny sensor viable in the first place. It's like firing your entire engineering team and trying to build the bridge yourself with a hammer.
"correct exposure" becomes a moving target. On a traditional camera, you expose for the subject and hope the dynamic range of the sensor captures the rest. On a phone, the computational stack is making per-region brightness decisions — it might expose for the sky in one part of the frame and the face in another, compositing them after the fact. Your manual exposure setting is more like a suggestion about which version of reality you prefer before the algorithm starts its work.
Let's get specific about ISO, because this is where most people get burned first. On a dedicated camera, ISO is often described as "sensitivity" — as if you're making the sensor more receptive to light. That's a useful fiction for beginners, but physically, it's gain. The sensor collects a fixed number of photons, converts them to a tiny voltage, and the ISO setting amplifies that voltage before it reaches the analog-to-digital converter.
You're not actually collecting more light. You're just turning up the volume on whatever signal you already captured, noise and all.
And on a full-frame sensor, the signal is strong enough that you can amplify it quite a bit before the noise becomes objectionable. A Sony A7 IV at ISO 6400 produces a clean, usable image. But on a phone sensor with one-tenth the surface area, each photosite is capturing far fewer photons to begin with. The signal-to-noise ratio is already worse at base ISO, and when you start applying gain, you're amplifying read noise and pattern noise that's proportionally much larger relative to the signal.
ISO 800 on an iPhone is more like ISO 6400 on a full-frame camera in terms of visible degradation.
Most smartphone sensors have a native ISO — the point where no additional gain is applied — around 100 to 200. Above 800, you start seeing chroma noise, which is those ugly color speckles in the shadows, and banding, which is horizontal or vertical striping from the sensor's readout electronics. By ISO 1600, you're in what I'd call "documentary evidence only" territory. Meanwhile a full-frame mirrorless body is just getting warmed up at 1600.
This is where the "always shoot at the lowest ISO" instinct kicks in for people coming from traditional cameras. They set ISO 100, then compensate with a slower shutter speed, and suddenly everything's blurry because their hands aren't a tripod.
Which brings us to shutter speed. The old reciprocal rule says your shutter speed should be at least one over your focal length to avoid camera shake. For a 24 millimeter equivalent phone lens, that suggests 1 over 24 seconds is safe. But that rule was written for film cameras held against your face with an optical viewfinder providing a third point of stabilization. A phone held at arm's length is a completely different mechanical situation.
Phone optical image stabilization is doing some work, but it's not magic. Most OIS systems on phones provide two to three stops of stabilization. So from the reciprocal rule's 1 over 24, you might get down to about 1 over 8 for a static subject before hand tremor becomes visible.
That's the practical handheld floor. And even then, only if you're braced and the subject isn't moving. The moment your subject moves — a person walking, a dog, a kid doing literally anything — 1 over 8 is useless. You need at least 1 over 125 for casual motion, and 1 over 500 or faster for anything energetic.
The real tradeoff on a phone is between motion blur and noise. You can't have a fast shutter and low ISO in anything but bright sunlight, because the sensor just isn't collecting enough light.
This is where exposure compensation enters as the third variable, though it behaves differently on phones than people expect. On a traditional camera, exposure compensation biases the meter — you tell it to brighten or darken relative to its best guess. On a phone, EC is often just a brightness offset applied after the computational pipeline has already made its decisions. In ProRaw or a manual mode that respects your settings, it adjusts the target brightness, but the phone may still apply tone mapping that crushes highlights or lifts shadows in ways you didn't intend.
Dialing in negative 0.7 to protect a bright sky might work, or the phone might decide it knows better and pull the shadows up anyway, giving you a flat, HDR-looking image that lost the drama you were trying to preserve.
The flip side is even more dangerous. Traditional photography has this rule called "expose to the right" — ETTR — where you intentionally overexpose slightly, pushing the histogram to the right, because that captures more shadow detail with less noise, and you can pull the highlights down in post. On a full-frame sensor with 14 to 15 stops of dynamic range, this works beautifully.
A phone sensor has 10 to 12 stops on a good day.
Those two or three missing stops are almost entirely at the highlight end. When you clip a highlight on a phone sensor, it's gone. Pure white, no recovery. The small photosite fills up — what's called full well capacity — and the excess charge bleeds into neighboring pixels. So ETTR on a phone doesn't give you cleaner shadows. It gives you blown skies, blown skin highlights, and data you can never get back.
Which is the photography equivalent of throwing out the baby and keeping the bathwater. You preserved the shadow detail in the bushes and destroyed the actual subject of your photo.
To put numbers on the comparison Daniel was asking about: take an iPhone 15 Pro at ISO 3200 shooting a dimly lit room. You'll see chroma noise speckles in every shadow region, and if you look closely, faint horizontal banding from the sensor readout. The same scene on a Sony A7 IV at ISO 3200 — the noise is a fine, almost film-like grain, and the shadows retain texture. At ISO 6400 the Sony is still cleaner than the iPhone at 1600.
For shutter speed, the dog-running test makes it concrete. On a Pixel 8 Pro at 1 over 30, even with OIS active, a dog trotting across the frame is a blur. The stabilization can compensate for your hand movement, not the subject's. At 1 over 500, the dog is sharp — individual hairs visible — but you've paid for it in noise, because the sensor had one-fifteenth the light-gathering time and had to crank ISO to compensate. There's no free lunch on a sensor this small.
We've seen how each parameter behaves differently on a phone — but video adds a whole new layer of constraints. The big one is the 180-degree shutter rule. For natural-looking motion blur, your shutter speed should be roughly double your frame rate. Shooting 24 frames per second? You want 1 over 50. Thirty frames per second? 1 over 60.
Which on a dedicated video camera is just what you do. But on a phone, locking your shutter to 1 over 50 means you've removed the easiest variable for controlling brightness. In low light, the sensor now has to sit there collecting light for a full fiftieth of a second per frame, and if that's still not enough, your only option is raising ISO.
Video noise is much more noticeable than photo noise. In a still image, noise sits there and you either accept it or don't. In video, noise dances. It crawls across the frame, flickers between frames, draws your eye. So the penalty for high ISO in video is worse than in stills, at the exact moment you've locked away your shutter speed escape hatch.
It's the exposure triangle with one leg tied behind its back.
Here's the thing most people don't realize — many phones cheat on frame rate to manage this. Android phones especially are notorious for variable frame rate recording. The phone detects low light and silently drops from 30 frames per second to 24 or even 15, which lets it use a slower shutter without you knowing. Saves battery too. But when you pull that footage into an editor, the audio drifts out of sync because the timeline thinks it's 30 frames per second and the actual frames don't match.
Which is a nightmare Daniel knows well, actually. He's mentioned variable frame rate headaches in editing before.
The fix is using a pro video app — Filmic Pro, Blackmagic Cam — that locks both shutter speed and frame rate, forcing constant frame rate recording. You pay for it in battery life because the phone can't coast, but your audio stays synced and your motion blur is consistent.
Then computational video modes throw their own wrenches in. Apple's Cinematic Mode — it records at 30 frames per second with a fixed 1 over 60 shutter, automatically adjusts ISO, and completely ignores any manual exposure settings you've dialed in. Google's Video Boost does real-time HDR and stabilization, same deal — you're along for the ride.
Which circles back to the core question: when do you actually take the wheel? For photos, my practical hierarchy is shutter speed first — set it fast enough to freeze your subject. 1 over 125 for static scenes, 1 over 500 for motion. Then set ISO as low as that shutter speed allows. Then use exposure compensation to fine-tune brightness — and here's the key, don't try to protect highlights manually. Let the computational stack handle highlight recovery. It's better at it than you are on a sensor this small.
In low light, accept the higher ISO. The instinct to drop shutter speed to 1 over 15 to keep ISO low is the single most common way people ruin low-light phone photos. Noise you can sometimes salvage. Motion blur you cannot.
For video, the strategy inverts. Lock shutter to double your frame rate first — that's non-negotiable if you want footage that doesn't look like a security camera. Set ISO to the lowest value that gives you acceptable brightness. And then the trick most smartphone shooters don't know: if you're in bright light and that forces an absurdly fast shutter — like 1 over 2000 at base ISO — you need an ND filter.
Sunglasses for your lens.
Moment, Freewell, and PolarPro all make magnetic ND filters for phones — twenty to fifty dollars, they snap right onto the lens ring. On a Samsung Galaxy S24 Ultra shooting 24 frames per second video in bright sunlight, without an ND filter the phone defaults to something like 1 over 1000 or faster. The motion looks staccato, jittery, unnatural. Slap on a three-stop ND and suddenly you're at 1 over 50, smooth motion blur, cinematic look.
It's one of those things where a twenty-dollar accessory makes a bigger visible difference than a thousand-dollar phone upgrade.
It reveals the second-order point here. Smartphone photographers have to think differently than traditional photographers because the computational layer is both a crutch and a constraint. It's a crutch when it saves you from the tiny sensor's limitations — multi-frame stacking, semantic noise reduction, smart HDR. It's a constraint when it overrides your decisions or when manual mode disables those same crutches and leaves you with raw sensor data you weren't prepared for.
The skill isn't mastering manual. It's knowing which fights to pick with the algorithm and which ones to let it win.
Let's turn all this into something you can actually remember when you're standing in a field at sunset with about three seconds before the light changes. For photos, my rule is: shutter speed first. Set it to freeze whatever you're shooting. One over 125 for a person posing, one over 500 for a kid running. Then let auto ISO do its thing.
Which sounds like heresy to anyone raised on "always control your ISO," but on a phone, manual ISO is rarely the right fight. The sensor is so small that the difference between ISO 100 and 400 is negligible once the computational noise reduction takes a pass, but the motion blur from a shutter speed that's too slow is permanent.
For video, flip the priority. Lock shutter to double your frame rate — one over 50 for 24 frames per second, one over 60 for 30 — and never touch it again. Then set ISO as low as brightness allows. And in daylight, spend twenty bucks on a magnetic ND filter so you're not forced into shutter speeds of one over 2000 that make everything look like a combat footage clip.
The ND filter thing is genuinely the most underrated advice in smartphone video. People spend hours tweaking settings trying to fix staccato motion, and the answer is literally sunglasses for the lens.
Exposure compensation is probably the tool most people ignore entirely. In high-contrast scenes — bright sky, dark foreground — dial in negative 0.3 to negative 0.You're telling the phone to protect the highlights, and on a small sensor, highlights are the first thing to die. In flat, overcast light, go positive 0.3 to 0.7 to lift the shadows without touching ISO.
It's the one adjustment that works with the computational pipeline instead of against it. You're not overriding the algorithm, you're just nudging its priorities.
The final rule — don't fight the computational photography. Learn to work with it. If your phone supports ProRaw or DNG raw, shoot in that format when you want editing latitude. You get the raw sensor data plus the computational metadata — the multi-frame alignment, the depth map — that tools like Lightroom Mobile can actually interpret. You're not getting a single unprocessed frame. You're getting the composite with the training wheels off.
Which is the real sweet spot. Not "manual versus auto," but "computational capture with manual finishing." The phone does what it's good at — merging frames, suppressing noise, aligning handheld bursts — and you do what you're good at, which is deciding what the final image should actually look like.
We've spent this whole episode talking about how to master manual controls on sensors that keep shrinking, and it raises an uncomfortable question — are we optimizing for a tool that's about to disappear? Samsung's latest 200-megapixel sensors are using pixels that are zero point five six microns across. At that size, each photosite is capturing almost no light individually. The entire concept of "manual exposure" starts to feel like adjusting the carburetor on an electric car.
Yet — the rumor mill says the iPhone 17 Pro is getting a variable aperture lens. A mechanical aperture on a phone. That would add a fourth parameter to this whole balancing act. Suddenly you'd have a real depth-of-field control that isn't computational fakery, and a new way to manage light without touching ISO or shutter speed.
Which would make manual control more relevant, not less. Right when the sensors are pushing toward "just let the algorithm handle everything," the lens hardware might give us a reason to stay in the driver's seat.
It's the push and pull of this entire category. The hardware keeps creating new limitations, the software keeps routing around them, and the hardware responds with new capabilities. Manual exposure isn't dying — it's just becoming a different skill than it was five years ago.
The through-line of everything we've talked about is that smartphone exposure is its own thing. The old DSLR rules will burn you. The new rules are: prioritize shutter speed for photos, lock it for video, use exposure compensation as your negotiation tool with the algorithm, and don't be afraid to let the phone do what it's good at.
The rule underneath all the rules — the best setting is the one that gets you to take the photo. If manual mode anxiety is making you miss moments, put it back on auto. The computational pipeline on a modern phone is remarkable. You're not failing as a photographer by using it.
The camera you have with you, set to the mode you'll actually use, beats the perfectly-exposed shot you never took.
Now: Hilbert's daily fun fact.
Hilbert: In the 1950s, a British colonial administrator in Belize noticed that the vivid red ink on a local land deed wasn't fading like standard inks. The dye came from the seeds of the achiote tree — the same plant that gives cheddar cheese its orange color — and had been used as document ink in the region for over a century.
Cheddar and colonial paperwork share a pigment source.
The big open question we're left with is where this all goes as sensors keep shrinking and computational photography keeps getting smarter. Will the variable aperture rumors pan out and give manual control a second life? Or will the algorithm become so good that touching the sliders at all feels like a nostalgia exercise?
Either way, the core insight holds: understand what the tools are actually doing, and you'll make better decisions — even if the decision is to let the phone take the wheel.
This has been My Weird Prompts, with thanks as always to our producer Hilbert Flumingtop. If you enjoyed this, find us at my weird prompts dot com.
Go take some photos. And leave the ISO slider alone unless you've got a very good reason.
