I have been circling a new phone for a few weeks, and I caught myself doing the thing. The thing where you assemble a justification you already know is decorative - in this case that the always-with-me snapshot camera might be the actual reason to buy it. So I went looking for permission on the internet, and the internet was delighted to provide: this phone now genuinely takes photos as good as a mirrorless camera.

Then I fell over a video making the identical claim about the Pixel 3, seven years ago.

It was not true then and it is not true now. I still have RAW files from a Canon 20D that hold up perfectly well next to what my current cameras produce, and I can see hard quality differences between those cameras and the snapshots from a reasonably recent Pixel without squinting. Both of those things at once. The sensor in your pocket got extraordinary and it did not close the gap, and the industry has been announcing that it closed the gap on a roughly annual cycle for a decade.

Which is the shape almost every gear story comes in, once you start noticing.

I said at the start of all this that photography keeps being sold to us as something done for an audience, a platform, or a purchase. The first two I have already had my say about. This is the third one.

Vinyl crackle

My favourite specimen is one I own and love. Fuji's film simulations are marvellous, I use them constantly, and the claim that they simulate film is marketing nonsense of a high order.

Put a bit of noise and the occasional pop underneath a digitally produced track and you have not simulated a vinyl record. You have added a reference to one. The grain slider in Lightroom has the same relationship to actual film grain: it gestures at a physical process it does not reproduce and was never going to. It is the same move as the 50mm myth I went after the other day, a decision handed to us as a fact, except that here somebody is holding an invoice. What Fuji has built is a set of very good, very characterful colour renderings with evocative names, and if they had called them that, nobody would have loved them less.

The Leica Look lives in the same drawer for me, along with 3D pop and micro-contrast and whatever the current word is. I am not saying the images do not have a character. I am saying that the character is a lens design and a colour profile and, most of all, the person who chose to buy that camera and therefore shoots a certain way - and that "look" is doing the work of hiding all three behind something that sounds like physics.

None of this stops me enjoying any of it. The gap I care about is between the marketing story and what is true, not between good gear and bad.

Three kinds of Leica commentator

Which brings me to the place where the stories get loudest.

There are three kinds of Leica commentator. One, people with photographic or technical reasons for liking them. Two, people who like them for emotional reasons. Three, people who do not own a Leica, would never buy one, and are extremely keen to tell everybody else how they ought to think about Leica.

Full disclosure so you can weight everything below accordingly: I like Leica, and I put myself in category one. I shoot an M10-D, which needs a word of explanation if you have never handled one.

The M is Leica's rangefinder line, unchanged in concept since the 1950s. You focus it by hand, by lining up a second faint image inside a small bright patch in the viewfinder until the two coincide. There is no autofocus. You are looking through a window at the actual world with a frame drawn on it, rather than at a small screen relaying what the sensor currently sees. The D in mine goes a step further: no screen on the back either, so there is no way to look at a photograph until it is off the camera. That combination exists nowhere else.

Then Leica announced the M EV1, an M with the rangefinder taken out and an electronic viewfinder put in - a tiny screen, in other words, showing you the sensor's view, the way essentially every modern camera does.

The forums were a perfect demonstration of all three categories at once, and honestly quite entertaining. Category one asking why this costs more than a Sony while doing less than Leica's own SL with an M-adapter on the front. Category two answering with Look, Design, Prestige, Legacy. Category three explaining that Leica is a Rolex for dentists with more money than ability. Nobody talking to anybody.

My own reaction was that I did not understand it, and it took me a while to work out why. Every so often the community starts hoping aloud that Leica will build exactly this, and the wish always sounded to me a bit like the sports car community hoping Ferrari will finally offer soft suspension and a cruise setting for the engine management. The constraint is the product. Take the rangefinder out and you have removed the reason the thing exists, and what remains is an expensive camera that does less than cheaper ones.

What the camera does to you

And that is the part the marketing never sells, because it cannot be photographed for a product page.

Shooting fully mechanically does something to you. Not to the files - to you. No screen means you cannot check the shot you just took, which means the walk never stops, which means the next frame comes out of what you were still looking at instead of out of the last one. The rangefinder puts a bright frame in the world rather than a preview of a picture, so you are looking at a scene with an intention laid over it and not at a small television showing you what you already have. I cannot demonstrate any of this in an image. It changed how I shoot anyway.

The same thing runs in the other direction and is easier to miss. I bought an OM-3 a while back and get spoken to about it constantly, everywhere, by strangers. Stealthy street photography this is not. That is not a flaw in the camera, it is a property of carrying it, and it changes what kind of work is available to me on any given afternoon as surely as a focal length does.

Or take the constraint I imposed on myself for a while, shooting out of camera and skipping the edit. I still do not know whether that was a self-inflicted handicap or a development step. My results only ever pleased me moderately and I wanted to fiddle with every single one - which could mean the discipline was wrong for me, or could mean I have years of conditioning to shed and that the itch to fiddle is the conditioning talking. That question is worth more to me than any comparison of JPEG engines, and it is a question about me.

So I will keep wearing it out

A dealer told me recently that my Q2 is worth almost nothing on the used market. Not because anything is broken - because it has marks on it. Apparently Leica buyers, unlike buyers of every other brand, expect a used camera to have been kept in cotton wool and taken out only in good weather, and the moment there is any evidence of a life the price falls through the floor.

Sit with that for a second, because it is the whole argument standing in the open. The resale market is offering me money to treat my camera as an object to be protected rather than a tool to be used. That is a piece of gear reaching all the way into how I behave, and in exactly the wrong direction. Everything else in this post is a story someone told about a picture. This one is a story about my hands.

I was only curious anyway. I will carry on wrecking it.

There is more where this came from. The whole series collects under Photo Thoughts.

My phone has been set to greyscale for months now. It started as self-defense - the standard advice about making the slot machine in your pocket less rewarding - and I expected to last about a week. Instead I turn colour back on maybe once a fortnight, for a map or a photo I actually need to judge, and every single time I flinch. The thing is candy-bright. Loud in a way I had completely stopped registering. Then I switch it back off and the room gets quiet again.

The mental hygiene worked. What I did not expect was the second effect, which turned out to be the more interesting one, and which is really what this whole post is about: it changed what I could see.

Scroll a feed in greyscale and most of the images collapse. Not some of them - most. It becomes very obvious very fast how many photos you encounter in an ordinary day are running purely on signal colour: a saturated orange, a branded blue, a sunset cranked until it hums. Take the colour away and there is nothing underneath. No composition worth the name, no contrast doing any work, no reason for the eye to stop anywhere in particular. We have collectively agreed to use colour as an attention mechanism rather than a compositional one, and the result is a diet of pure sugar with no finesse in it anywhere.

And then every so often something does stop you, and it stops you harder than it ever did in colour, because now you are seeing structure instead of being flagged down. Those images are worth studying, because whatever is holding them up is doing so without the loudest tool in the box. Greyscale turns out to be a decent instrument for sharpening your own eye, which is not what I installed it for.

Try it for two weeks. Worst case you learn which pictures in your own archive were only ever carried by a colour.

Half-truths we repeat until they sound like physics

The greyscale thing is one small correction to a much larger problem, which is how much of what we believe about photography arrived without ever being examined. Somebody says it on YouTube, it gets repeated, and after a while it stops sounding like an opinion and starts sounding like a property of the world.

My favourite specimen, because it is the most confidently wrong: 50mm is the focal length that matches human vision.

I would love for some photography student somewhere to write a proper paper on how this claim managed to convince so many people, because as far as I can tell it is repeated far more often than it is examined. And the giveaway is that everybody agrees on the conclusion while nobody agrees on the reason. Ask five people why it is true and you get four different mechanisms, each of which falls apart when you poke it.

The most literal version goes via the focal length itself, which is easy to check and immediately wrong. The eye's physical focal length is somewhere around 17mm, and it is not even fixed - the lens in your head changes shape when you focus on something close. Whatever else 50 is, it is not that number.

So people move to field of view, which sounds much more promising. A 50mm lens on full frame covers roughly 40 degrees horizontally and about 27 vertically. Human vision runs to something like 135 degrees. That is not a near miss, that is a different order of thing. And this is where the comparison stops being merely inaccurate and starts being category confusion, because our eyes do not work like lenses at all. Only a tiny patch of the retina resolves sharply. If we assessed our vision the way we assess a lens, on the basis of what is actually sharp on the sensor, we would be walking around with a two degree field of view. We do not notice this because our eyes are in constant motion and the brain assembles the result into a stable picture that never existed on the retina in one piece. Which is a further reason the comparison fails: there is no exposure. You are not looking at a frame, you are looking at a composite being continuously rebuilt.

Point this out and the argument usually retreats to perspective. Never mind the angle, the claim goes, it is about how lines and depth relationships look - a 50mm rendering of a street matches how the street feels to stand in. Except the eye projects onto a curved surface and a lens projects onto a flat one. There is no focal length that makes a flat projection behave like a spherical one; the geometry simply does not have that solution. You can pick a focal length whose distortions bother you least, and that is a real thing worth choosing, but it is a preference about failure modes rather than a match. And whatever "how the street feels to stand in" means, a good part of it is coming from having two eyes about six centimetres apart, which no single lens has ever done anything about.

The last version is the only one with something in it, and it quietly gives the game away. A 50mm image can look right - but only if you also fix how far away the viewer is standing. Move a print closer to your face or further from it and you have changed the angle of your overall shot, which is roughly what zooming does. So the claim holds only if we all agree to look at pictures from the same distance, which we very obviously do not, since the same photograph now gets viewed on a phone at arm's length, on a laptop, and occasionally on a wall.

Notice also that the arguments pull in opposite directions. Match the total field of view and you end up somewhere around an ultra-wide. Match the small cone you are actually attending to at any given instant and you end up in short telephoto territory, 85mm or beyond. Match the projection geometry and there is no answer at all. Nothing here arrives at 50 unless you already knew that 50 was the answer you were looking for.

What is really going on has nothing to do with eyes. The diagonal of a 35mm frame is about 43mm, and a lens roughly matching the diagonal of its format was, in Barnack's day, the cheapest thing to build fast and well corrected. 50 was close enough, so 50 got made, in enormous numbers, for decades, and came bundled with every camera anybody bought. Two or three generations of photographers learned to see through it before they ever chose anything. The "normal" lens is normal because of manufacturing economics and the shape of a rectangle somebody standardised a hundred years ago.

Once you have caught one of these you start finding the rest, and they all have the same shape - a decision presented as a fact:

The good light is golden hour and blue hour. Grey skies are boring. A photograph has to be sharp and full of detail. A photograph comes in a certain format.

Take the sharpness one, since it cost me the longest to shake. Sharpness is the easiest quality to measure and therefore the one the industry has agreed to sell us, review by review, MTF chart by MTF chart. The longer I photograph, the less often I care. My favourite frame from last winter is a low-angle shot of a pair of black ice skates with pink laces cutting across the ice, and essentially nothing in it is sharp. The blur is the subject. Freeze that moment properly and you have a picture of some skates.

The frame is a decision, not a default

Which brings me to the last one on that list, and the one I have thought about most.

Image formats are a convention. Not a law, not a technical constraint, and not even a very old convention. There was a time when oval and round images were about as ordinary as rectangular ones - look at what people hung on walls before industrial production got involved. The rectangle won because it is what rolls off machines efficiently, what fits a film gate, what cuts from a sheet without waste. Then it won so thoroughly that everything else began to feel wrong.

I have been playing with a circular fisheye lately, a lens that produces a genuinely round image with black around it, and the most interesting part has been watching my own reaction. Round pictures feel wrong to me. Not bad. Wrong, the way a misspelled word is wrong. I look at a round photograph of the Toronto skyline in magenta and green, sun flaring off a tower, the whole thing curled into something like a crystal ball, and part of my brain is still filing a complaint about the shape. There is no reason for that beyond a lifetime of rectangles. The complaint is information about me, not about the picture.

So here is what I actually think, and it is the closest thing I have to a rule. A mature photographer chooses the format from what they are trying to say, not from where the sensor collects the most data.

That turns a default into a set of questions worth asking every time. Do you want to confirm what the viewer expects, or surprise them? Where should the eye enter, and where should it be unable to leave? Are you after symmetry or motion? Do you follow the standard - landscapes horizontal, portraits vertical - because the subject wants it, or because that is what everyone does? A square imposes a calm that a 3:2 cannot, simply by refusing to favour any direction. A very long horizontal makes the eye travel and turns viewing into something closer to reading. Extreme formats are wonderful for exactly this reason: they make the viewer feel the shape before they have identified anything inside it.

None of that is available while the frame is still a setting you never touched.

Minor White has a line that I keep coming back to, and it is one of my favourite photography quotes: "Don't show me what it is, show me what else it is!" I do not manage it often. But the times it works are the times I decided something on purpose - the colour, the light, the shape - instead of accepting what the camera and the internet handed me.

There is more where this came from. The whole series collects under Photo Thoughts.

Then other people started posting fotogedanken of their own, which made me very happy. The tag has turned into something better than a personal notebook - follow it and you get a steady stream of photographers thinking out loud about the medium instead of about gear, which is rarer than it should be.

Reading a year of my own back in one sitting, the same argument keeps surfacing in different disguises: photography is a practice you do for yourself, in a world that keeps insisting it is for an audience, a platform, or a purchase. Everything below is a variation on that one idea. Each section also has a longer post behind it that I will get to over the coming weeks - this one is the distillation.

The part that isn't pressing the button

Start with something I keep coming back to. Photography is the only art form where the act itself can theoretically be over in a millisecond. A painter is inside the making for hours. A musician is inside it for the length of the piece, at minimum. We can walk around for an entire afternoon and the work, the actual exposure, lasts a five hundredth of a second. Everything else - the walking, the waiting, the noticing, the editing, the sequencing, the moment where you hand the thing to somebody - sits around the act rather than being it. That is exactly why thinking about the medium is worth the time. The interesting part is almost entirely the part that is not pressing the button.

It also means photography reaches a lot further than making pictures with a camera. My son's girlfriend gave me a notebook for Christmas, a Nuuna made from recycled jeans, and it feels wonderful. It also needed work before it could compete with my current favourite, a Dingbats. I transplanted a ribbon marker out of another notebook, glued in three pen loops, built a paper pocket into the back cover. Now I only need the patience to fill the book I am currently writing in before I am allowed to start the new one. I count that as one of my fotogedanken, because people who are into photography tend to be the same people who notice the fine grain of things, who have a soft spot for analog objects even when they shoot digital and publish online, and for whom the way there is usually the point.

Amateurish is a compliment

Somewhere in the first year of putting frames around my images, someone whose work I openly admired told me they looked amateurish and gimmicky. Gimmicky is a matter of taste, fine. Amateurish I will happily take as a compliment, because it says I am doing this primarily out of affection for the medium rather than to satisfy a convention. We live in a world of photographic oversaturation. The only person who has to be happy with the result is you. If that means a frame, a filter, a round image, a deliberately blown-out sky, then you have already won.

A swipe crowd is not an audience

"Instagram is the best way to find an audience for my photography." I hear it constantly and I have doubts. I have never once watched somebody be long and thoughtful on social media. Be honest with yourself - how often does it really happen? And by long I do not mean three seconds instead of half a second. What we are dealing with is a swipe-like-swipe-swipe-like crowd, and whatever that is, it is not an audience. Not the same species of thing.

The marketers make this very legible. Every online presence I have ever run - the blog, the podcasts, the photography, back when I was still feeding a five-figure follower count on Twitter - has eventually attracted someone who wanted me to start selling my audience. Photography seems especially prone to it. The wish for an audience seems to lead directly to the error of treating monetization as the same thing as success. Requests came in via Pixelfed the other week. Nobody has yet offered to monetize my Bookwyrm profile, which is a shame. I guess I keep reading purely for fun then.

Twenty copies, given away in person

The counterweight is not to give up on being seen. It is to put the work somewhere that assumes a viewer with a spare five minutes. Erik's photo zine is the model I keep pointing at, because it stands in a long tradition of people thinking critically about photography as a form of communication and handing those thoughts to others for discussion. That practice has become rare. Most photography publishing today is either a lightly disguised invitation to buy gear or a how-to, and both of those are fine, and neither of them is a conversation.

There is a rule hiding in that which I only formulated recently, and it is uncomfortable enough that I want it in writing: if you put no effort into publishing your images, you do not get to be surprised when the audience puts no effort into thinking about what they saw. Layout is respect. A sequence is an argument. Handing somebody twenty printed pages is a different request than dropping a JPEG into a feed, and people answer the request you actually made.

So my photos exist almost exclusively in print, in a zine or a book, and even then only if you really want one. Twenty copies, maximum print run, given away in person. This year I want to go a step further and make them by hand. I genuinely do not know yet whether that is a good idea, because I love holding printed photographs about as much as I hate wrestling with modern printing technology. But I want unusual paper, and unusual paper means doing it myself.

There is more where this came from. The whole series collects under Photo Thoughts.

Still here? Okay. Let's get started.

In my last post, I described where standard RAW developers like Lightroom hit their limits when it comes to infrared photography. The takeaway was that automated processes optimized for normal photography get completely thrown off by the quirks of an IR image.

The first step of any image processing is what we call demosaicing. Even if you shoot straight to JPEG, this step still happens first, just directly inside the camera instead.

What is demosaicing anyway? The art of puzzling

A RAW file is not a finished image. You have to imagine it more like an incomplete mosaic. Every pixel on the camera sensor sits under a color filter - usually Red, Green, or Blue (the so-called Bayer filter). So a pixel under a red filter only knows the brightness value for red, and has absolutely no clue about green or blue. Demosaicing is the process where the software solves this puzzle. The algorithm looks at a pixel and its neighbors, then makes an educated guess to calculate the missing color values for every single pixel. Only after that do we get an actual color image.

How the software solves this puzzle depends on the motif, the situation, and the camera. It is less of a fixed mathematical equation and more of a strategic choice. This is where the wheat gets separated from the chaff. There are a bunch of different ways to solve this problem, and since most people want to deal with photos rather than algorithms, programs like RawTherapee have smart processes to choose for us. But they also let us make the choice ourselves - which can actually be better, since we as photographers usually know best what we want to focus on.

Here is a screenshot from RawTherapee showing a 100% crop of one of my RAW images with demosaicing deactivated. You can clearly see the structure of the sensor:

The exact same image, same crop. Only difference: the algorithm.

Broadly speaking, you can divide most algorithms into two philosophies:

1. The "detail-first" philosophy (The Optimists)

These algorithms assume that the image data is clean and that even the tiniest deviation between pixels represents a real detail. They analyze edges and patterns aggressively to reconstruct the sharpest possible image.

• Examples of these algorithms: AMaZE, RCD, DCB
• Strengths: They deliver perfect sharpness with clean, low-ISO images. Ideal for fabrics, architecture, or fine structures.
• Weaknesses: They are easily fooled. Noise looks like a fine pattern to them - and gets promptly "amplified". This leads to things called labyrinth artifacts. You can get false colors (color fringing) on hard edges, or moiré (rainbow patterns) on fine textures.

2. The "artifact-avoidance" philosophy (The Cautious)

These algorithms are more conservative. Their main goal is to avoid making mistakes, even if that means losing a tiny bit of sharpness. They assume the data might be noisy and put more value on smooth surfaces and clean colors.

• Examples: LMMSE, IGV, VNG4
• Strengths: They are ideal for high-ISO images or noisy data. Labyrinth artifacts or moiré are barely an issue here.
• Weaknesses: With an absolutely perfect image, they can look a bit "softer" in a direct comparison with their aggressive colleagues.

Useful parameters to know

Besides choosing the algorithm, there are a few sliders you can use to fine-tune your strategy. The following refers to RawTherapee, but you can find similar options in other tools too.

• Dual Demosaic (e.g., AMaZE+VNG4): A brilliant hybrid approach. You can use a "detail algorithm" for the sharp areas of the image and a "cautious algorithm" for the smooth surfaces (like the sky). The contrast threshold slider defines where that boundary lies.
• False Color Suppression: This tool applies a filter that affects only the color information, not the sharpness. It smooths out those colorful splotches that often appear along edges.
• Border: The pixels at the very edge of the sensor can't be perfectly reconstructed because they lack neighbors. Normally, RAW converters just crop them out. With this slider, you can force the software to try anyway - but at your own risk, as this can create artifacts.

With this knowledge, we can now choose the right strategy for our images on purpose. For standard shots, modern RAW developers are already pretty good at doing this for us. Low ISO shot? => AMaZE. High ISO? One of the others... For photographers, knowing this stuff becomes valuable when you deviate from the standard case. Shot at high ISO by accident? A special light spectrum or heavy filter use that creates a scenario the software can't guess? Choosing the algorithm yourself can really help!

One such scenario is infrared photography, so here is an explanation of what that looks like in practice.

Filter Group 1: "Super Color" / Low Wavelength (550nm, 585nm, 625nm)

• The Challenge: Here, a wild mix of strong infrared light and a portion of the visible spectrum hits the sensor. The result is very intense false colors and a high risk of artifacts. However, we still have usable detail information across several channels.
• The Strategy: We need a compromise. A "detail-first" algorithm is good, but it has to keep false colors under control.
• The Choice: I almost always start with DCB here. It's just as sharp as AMaZE, but it's better at handling false colors, especially for cameras without an AA filter (like my RP). An alternative is a Dual Demosaic approach to treat the sky and foliage separately. To help things along, I set the false color suppression slider to 2 or 3.

Filter Group 2: "Classic IR" / Mid Wavelength (720nm)

• The Challenge: The signal is now almost pure infrared. The red channel is the undisputed king, while the green and blue channels are weak and extremely noisy. The ISO 100 in your metadata is a flat-out lie - for G and B, the image is closer to ISO 12,800.
• The Strategy: We have to force the software to treat the image like a high-ISO shot. An "artifact-avoidance" strategy is mandatory so we don't interpret and amplify the noise as detail.
• The Choice: For false-color images, LMMSE or IGV are the top choices here. They handle the noisy channels gently and prevent those dreaded labyrinth patterns. With a 720nm filter, though, you might also just decide to go straight to a black-and-white conversion. Which brings us to our next option...

Filter Group 3: "Deep IR" / High Wavelength (850nm) & Black-and-White Conversion

• The Challenge: There is virtually no color information left here. The G and B channels are basically pure noise. Traditional demosaicing would be pointless and counterproductive here.
• The Strategy: We should focus on the strongest, cleanest signal we have and ignore the rest.
• The Choice: This is the perfect use case for the Mono algorithm. It is the "anti-demosaicer". It ignores the broken G and B channels and builds the black-and-white image directly from the data of the clean red channel. The result is the maximum possible sharpness and tonality without any interference.

So, that's it for today. At some point, I'll extend this series to cover file formats and color spaces, but that's enough for now. Ultimately, for the pragmatic case, you can always just switch your camera to "Sport" mode and trust the tech 😉

Anyone getting into infrared photography quickly notices: the usual image editing tools hit their limits here. Lightroom in particular, my tool of choice for years, is really not built for IR. It gets in your way at the very first step - setting the white balance. IR images are very warm (red, basically), and to compensate you have to push the white balance slider dramatically to the left into the "cold" range. Lightroom unfortunately has a hard-coded lower bound that is "too warm".

Resourceful users found a workaround: camera profiles! Turns out you can create and register camera-specific profiles in Lightroom. Getting there is a bit fiddly though, especially if you're trying to sidestep Lightroom's built-in limits. The solution? A free DNG tool from 2012 (!!) by Adobe, originally meant for converting manufacturer RAW files to DNG (more info here). Open a sample file in the tool, drag the color temperature all the way down under "Color Matrices", export the whole thing as a camera profile, drop it into Lightroom's profile directory, then either pick it manually or configure it as the default when this specific camera shows up as the author of a RAW file. Convoluted, sure (here's a tutorial), but at least you can now set a proper white balance for these files. One important thing: NEVER trust the auto setting. It's hard-tuned in LR. Great for normal shots, not great for special cases.

And the pain doesn't stop there - a lot of IR photographers eventually want to swap color channels. Lightroom is not exactly intuitive for that, and at the latest at this point you'll want to move to a "big" image editor.

The next issue: "intelligent" software like Lightroom often produces noisier IR images than necessary. Why? Because neither the development algorithms (in particular the demosaicing) nor Lightroom's famous denoising fit our use case. You don't always spot it immediately, especially not if you look at a downsized image instead of at 100%. But in practice, even when the image looks clean at first glance, you have much less headroom for corrections and less detail than you could have. And especially when adjusting color channels you'll hit limits and clip channels faster than you'd like.

Where does this come from? Simple: if you point a RAW developer at IR photos, you're basically forcing it to get confused. Unlike "normal" images, the image information is not evenly distributed across the four filters of a Bayer sensor. Instead the red channel is loaded with information while the green and blue channels are weak and noisy. The job of a RAW developer is to compute the "right" colors from those signal values using a suitable algorithm. Especially with high-ISO images, i.e. lots of noise, the software picks different algorithms than for low ISO. An IR image, particularly one taken during the day, often comes in at a low ISO (100, 200) but still has enormous noise in two of the three channels. So if you want to squeeze out the best result you need to tell the software up front how to develop the RAW. Many tools only allow that in a very limited way. Denoising is very similar: modern software comes with several specially trained AI models and picks them based on image metadata. Automatic denoising in Lightroom or in the DxO tools I usually love is a "hit and miss" affair. You just can't trust it blindly. In many of my shots these tools actually make the noise worse if I don't watch out.

All of these problems can of course be worked around in an image editor. Pros use channel-specific luminosity masks and do denoising the "classic" way, without AI tools. That gets tedious fast, and at least I sometimes lose track of what I'm even trying to do. Image editing gets mixed up with image development, and the result suffers.

So, what to do? I've been at this for a while and by now I've found decent solutions for a lot of these things. What I find interesting in hindsight is the journey I went on - from "blindly click through a tutorial and learn what to do to get a certain look" to "dig deeper and deeper into sensors, file formats, base processing". Especially around IR there's a lot written on the internet that is only half-true, or where the explanations don't quite match the effect. Hence this blog post. I don't claim to be error-free, but maybe these notes help someone who is past the simple basics and just wants to do it properly for once.

One thing to know up front: I don't see myself as a particularly strong image editor. Once my RAWs are properly developed I'm more of a tinkerer than a targeted editor. Photoshop & co are part of my toolbox, but 80% of the functions in there feel esoteric and dangerous to me. But at least the front end of the processing chain I think I've now understood well enough, so here's my workflow, as of 2025:

My workflow for maximum quality (my tools of choice: RawTherapee & DxO)

After a lot of trial and error I've built a workflow that combines the strengths of different programs. The basic idea: use RawTherapee for the technical groundwork on the RAW file, and only then hand it off to my favorite tool, DxO PhotoLab, for optical corrections and creative optimization. Both tools can of course be swapped for alternatives. The role of DxO PhotoLab in particular is interchangeable. But after years of experimenting I do consider RawTherapee the most powerful option, simply because the tool is basically arbitrarily configurable when it comes to the base algorithms. Close behind: Darktable. Personally I don't love DT, but I have to give it that it also offers lots of knobs and levers that are often optimized away or lumped together in the big productivity tools. And last but not least: my IR workflow is different from my workflow for "normal" images. For those I just stay in DxO from start to finish, and before that I mostly lived in LR. But back to the IR workflow.

The goal is to produce, in RawTherapee, a technically perfect but neutral 16-bit TIFF - a "digital negative". Only then does the actual editing begin. Why? Because the point is to interpret the sensor data in a controlled way and then lock that interpretation in. TIFFs are painfully large files, just as a word of warning. But they are lossless and give you maximum image information for further work.

Phase 1: The technical foundation in RawTherapee

This is where you make all the decisions that automated software would get wrong because of IR's quirks. We deliberately override the software's wrong assumptions.

Demosaicing - the single most important step

Here we decide how an image gets computed from the raw sensor data. At 100% zoom the differences become obvious.

• For 720nm filters (classic IR): I usually start with LMMSE or IGV. These algorithms are optimized for high noise and prevent the typical "maze artifacts" that noise in the G and B channels can produce. If you're aiming for a pure black-and-white image, Mono is worth a try - it often gives the sharpest results because it focuses on the clean red channel.
• For 550nm filters ("Super Color"): For me DCB has proven a good starting point. My EOS RP doesn't have an anti-aliasing filter, and the documentation says DCB is good at avoiding false colors on such sensors.
• False color suppression: For 550nm shots this slider is worth its weight in gold. A value of 2 or 3 removes the typical pink/green blotches without hurting sharpness. For 720nm I leave it at 0 or 1.

White balance - manual control

Auto is essentially useless here, because infrared is invisible by definition. The images we're working with are a form of false color photography, and it's up to the photographer which effect to aim for. That said, you can of course click the auto button and depending on the tool something you like might come out. Just make sure you don't accidentally clip one of the channels. That's exactly why I set WB manually as a rule.

• Eyedropper tool: A click on healthy green foliage (which in IR reflects almost white) or on clouds is usually my first try. But again: you decide what color shift you want.
• Fine tuning: Nudge with the Temperature and Tint sliders.
• The power tool: If the image still has a color cast, RawTherapee has a special tool that feels like it was made for us: the Red/Blue equalizer. This slider directly controls the ratio of red to blue and can eliminate stubborn casts. Why is it made for us? Because in IR the red channel is overrepresented at the expense of blue. The white balance shifts values between red and blue back and forth, and the equalizer lets you calibrate that ratio.

3. Exposure & noise reduction

• Exposure: I set exposure and black point so that the histogram clips neither on the left nor on the right. Important: I leave highlight recovery off, because it's optimized for visible light and produces weird results in IR. I also expose to the right when shooting (ETTR): I deliberately expose so bright that the highlights just barely don't clip. The reason: in IR the shadow detail drowns in noise very quickly. The only cure is to capture more signal, i.e. ideally to expose longer. Cranking up ISO is a bad idea, by the way, because that amplifies the noise too. So you want the lowest possible ISO with the most possible signal. Which means: aperture wide open, or long exposure time.
• Noise reduction: Here I focus exclusively on color noise (Chroma). Luminance noise I don't touch. A manual pass on the Chroma slider very effectively kills the color blotches in the weak channels.
• What NOT to do in this phase: sharpening, contrast, saturation, or any other creative tools. The goal is a "flat", neutral, but technically clean image.

Phase 2: Export as a digital negative

Once everything fits, the image gets prepped for handoff.

• File format: TIFF
• Bit depth: 16-bit (absolutely essential for quality!)
• Color profile: A wide color space like ProPhoto RGB or Rec2020. This makes sure none of the unique IR colors get lost on export.

Phase 3: The fun part in DxO PhotoLab (or another tool)

The exported TIFF is now a "well-behaved", clean image. The problematic IR artifacts have already been handled manually and correctly. Now DxO can play to its strengths without getting confused.

• Optical corrections: Apply DxO's first-class corrections for distortion, vignetting, etc.
• Contrast & clarity: Tools like ClearView Plus and the microcontrast slider now work beautifully.
• Creative work: If you want to swap channels for the "false color look", rotate the slider in the HSL tool by 180 degrees, or pass the TIFF over to Photoshop, Affinity or GIMP. Go wild.
• If some noise is still left: you'll notice right away - working from TIFF, all those almost magical AI denoising tools disappear, because they mostly work on RAW data and need it. But there are exceptions. The "classical" tools are still available of course, and some AI tools work too. Topaz Denoise AI, for example, handles TIFFs just fine, and many of the Photoshop automatics are still available.

This workflow may look convoluted at first, but it makes sure every decision is made by the tool best suited to make it. The result is visibly higher image quality and more flexibility for the creative side. It also optimizes for marginal utility. If you want to be pragmatic: take a look at the automatics and keep an eye on noise levels. Depending on the subject it may just not matter and a simple round with the auto sliders is enough. My final recommendation: if you shoot IR, seriously look at other RAW tools, even if your heart is with Lightroom. Even if you don't want to pick algorithms yourself, just the handling of white balance and IR-specific color shifts is often solved much better elsewhere. Tools like RawTherapee and Darktable are also free, so at least it doesn't hurt your wallet.

More reading?

• Demosaicing article on RawPedia
• Custom DNG Tutorial for setting a white balance for infrared photography
• Bayer sensor (Wikipedia)
• Advanced Infrared Photography Processing: The Power of Luminosity Masks

How do white balance and tint tools actually work?

Every pixel is a mix of a red, green, and blue value. To compute that value, demosaicing algorithms make assumptions about the usual distribution and combine the sensor values accordingly. White balance then shifts the balance between the R(ed) and B(lue) values toward cooler (toward blue) or warmer (toward red). Tint does the same thing but for the green value, which white balance actually leaves untouched.

If you want to play around, here's a little simulator that does this for ONE pixel. Just set the base color and mess with the sliders. Have fun 🙂