Every photograph is developed in its own sense. From your automatically heavily processed phone photo, to the landscape photograph you see in your local bar.
This is a guide that is meant to provide you with the principles of photo editing software, so that you can pick any one of them up and start editing, not because you know that specific software, but because you understand the basics. As a - let’s call it - backbone of this guide is the new up & coming open source photo editing software called RapidRAW which I believe is great for beginners with a low entry barrier and the best of all, no cost. No cost not at the cost of performance.
You may notice up there a little toggle between darkroom and neutral. To capture your attention, this guide has styling that’s a little obnoxious, but it’s funky and I like it.
To get used to your editing software I recommend you try out the neutral theme. It’s grey. Just grey. This has a good reason. It promotes “colour accuracy” in your eyes. Yes, your eyes lie to you. Get used to this grey, you will be seeing it a lot. Unless you decide to change the theme of your editor to be a little more fun instead. (Hey! It’s your decision, I am here only to recommend you new stuff)
Every photograph in history that mattered went through the careful hands of a photographer that wasn’t capturing just a scene. They were capturing so much more… stories, emotions… As such, editing photographs was always an integral part of the craft. The emotions of today’s folk about “not wanting to edit photos, I want them to be as they are, I want them to be real” are a little disingenuous. The photograph you have received from your phone, from your point-and-shoot camera, from your scanned film negatives… all of those photos have gone through some sort of “post-processing”. They were never “real” the way you imagined them to be. Instead, the raw data captured by the sensor, or the photon sensitive crystals of film were just that. Raw data. No emotion. No story. No “human”.
The first pioneers of photography always edited their photos and created stunning works of art in their respective era. Back then, they required whole “labs” selectively “dodging” and “burning” (the equivalent of today’s local exposure adjustments), using chemicals and light in a specific way, combining two different negatives to capture a bright sky and the dark sea in one picture and many other means just to bring to the world what they “really” saw not just with their eyes. But with their heart.
Photography has become - let’s say - democratised. Everyone of us has at their disposal tools those legendary photographers starting from the 1850s could have only dreamt of; yet they still managed to create works of art that stun our breaths ‘till this day.
Here is the part that actually follows up on our photography field notes. The exposure triangle and the result of using it let’s say.
“How bright it is”. Let’s keep it simple and look at it that way. When you are taking a photo you are playing with the amount of photons that land on a photon sensitive layer (film) or that get turned into signal (digital). By doing this we capture a lot of data about the visible world around us (visible to us humans).
Now if you take the exposure compensation and crank it up by +2, you have a very bright image. This is - let’s call it - the equivalent of what exposure does in editing software.
What it is mechanically. In a scene-referred pipeline, exposure is the simplest operation there is: a multiplication of linear light. One stop = ×2 or ×0.5. output = input × 2^EV. Because it's a pure multiply on linear values, it behaves exactly like opening the aperture or lengthening the shutter would have — it's the "virtual" continuation of the exposure triangle into editing. This is why it's measured in EV/stops, not an arbitrary 0–100.
You f*cked up the exposure settings in the field, now you can fix it 😍.
When operating with exposure, stick to these principles:
If you are playing around with the exposure slider, pay a close attention to skin tones. This usually is the middle grey (we will talk about that later). A useful rule of thumb for getting the exposure “good enough” are the skin tones of people.
While adjusting for the midtones, watch out to not “clip” any information at the top end and the bottom end - highlights and shadows. Anything that touches the histogram at the right is pure white, anything at the left is pure black. There is no information here, there is no detail here. There are just regrets.
Pushing clipped highlights anywhere will usually lead to weird purple or orange colours. If you managed to do this, congratulations, The photo “can’t be saved”. Now you have to pretend that that was your artistic vision and sell that. Similar concept goes for blacks.
One way to use this to your advantage is while taking the photo do what’s called ETTR (“expose to the right”). Essencially if you know you will be editing the photo and want as much detail captured, while taking the photo use exposure settings that will push the histogram as much to the right without reaching the right border, that is without clipping the information and capturing pure white.
Measured again in stops, this tells you how much “range” there is between pure black and pure white. In practice this means taking photos in bright harsh sun and being able to brighten the dark areas and vice versa. This is where modern cameras have not gained much technical progress since let’s say 2014 (source: trust me bro).
Fun fact: HDR is this but on steroids - “High Dynamic Range”. Meaning, you can have a scene with very bright and very dark parts without any “clipping”.
So what exactly is the difference between exposure, brightness, lightness, luminance and brilliance? Honestly, think of the movie with the iconic line “We’re same same.. but different!” There are technicalities, each does something else, best to just crank the slider from one end to another and see what it does. Here you get some AI generated definitions:
Quick recap first: exposure is a multiply on the actual light in the scene — global, physical, measured in stops (one stop = ×2 or ×0.5). It behaves exactly like opening the aperture or lengthening the shutter would have: the virtual continuation of the exposure triangle into editing. The next four are all answers to the same question, “how light does this look?”, just measured differently.
Brightness is the naive, subjective one — usually a blunt lift applied evenly across the whole image. Luminance is the physical one — the real, perceptually-weighted amount of visible light (green counts most, blue least). Lightness is that luminance bent onto the curve of human perception, so middle grey lands mid-scale instead of near the bottom (the “L” in HSL is a rough version of it). Brilliance is darktable’s term for changing how bright a colour reads while keeping the colour itself intact — brightening without bleaching.
Why it’s worth telling them apart: exposure rescales the light itself, while the others reshape how that light is presented. Reach for the right one and your colours and contrast stay honest instead of washing out. Same same, but different.
So now we have got the data exposed properly or the way we want to. What do we do now with the data? Well we won’t be really doing much by ourself, but the software will.
Adding onto what was already mentioned about dynamic range: A camera can capture more than a display can.. well.. display (pun intended). The software now needs to figure out what the “brightest brights” and the “darkest darks” are. More on that later, now let’s talk about how you can work with this.
Ok but what the hell do I need this information for?
Because usually this is the first thing that pops up in your face when you open Lightroom, RapidRAW, Darktable, or any other photo editing software. You are working in (usually) 5 “zones” established by the tone mapper. These are the blacks, shadows, midtones, brights/highlights and whites plus contrast. In human speak, all it does is you are selectively editing parts of the image that are “this bright”.
If your image is the lovely pure raw flat grey something, you can start off like this.
You are essentially setting the “clipping point”. Pull Whites up until the brightest meaningful highlight just reaches white; Blacks down until the deepest shadow just reaches black. Or however else your eye pleases. It is your photo.
Recover detail in the bright end (pull Highlights down to bring back a sky) or open the dark end (lift Shadows). Or fuck it up on purpose
In my own words: create more “distance” between the dark tones and the bright tones.
You can up the blacks so that they are not that black in an artistic manner. It may resemble “film” be “aesthetic” give off “vibes”. That way the picture is less “harsh” and can be a bit more pleasing to the eyes. As a rule of thumb, pushing blacks to be really black can be the exact opposite of what you are looking for. And again, mind that small nudges get you far. If you are a beginner, you probably already moved the sliders way too far. Tone it down (pun intended). If you are a professional, you maybe moved the sliders way too far. For both of you; take a 10 minute break, go outside, look out the window, come back and look at it again.
Tone equalizer!!! In it’s essence you just point your mouse at the parts you want to “brighten” or “darken”, scroll your mouse and bam! That’s it. Under the hood it is essentially the same.
There are 2 main technical directions currently:
This demo is here to make you understand the concept of tone mappers and a comparison. Play around with the slider at AGX and Basic and notice how the highlights clip and how the highlights “roll-off” without clipping. AGX here is trying to emulate how actual analogue film captures light.
Don’t mind this too much right now, but remember this if you ever see these terms:
Quick version before the deep end: “display-referred” and “scene-referred” describe what numbers the editor is really pushing around. Display-referred works straight in screen values — 0 is black, 1 is white, and nothing is allowed to be brighter than white, so highlights just hit the ceiling. Scene-referred works in the actual light of the scene, where a lamp can be far “brighter than white,” and only squeezes that huge range down to the screen at the very last step — which is why modern highlights roll off gently instead of snapping to white.
Display-referred vs scene-referred — the fault line.
The tone mappers.
“Yeah I’ve got those! 😘 “ Not those curves unfortunately 😞. This is probably the most surgical tool for tones and will be your biggest source of practical and artful changes to the photo. For illustration purposes curves are that diagonal line that by dragging your mouse on you are creating the curves (In reality, the tone mapper already created a curve, you are in a way creating “another curve”). But be very careful, slight shifts do big changes. If you feel that you have done a good job with the curves, “tone it down” (pun intended). Usually as a beginner you already did far too much. Usually as a professional you maybe did too much.
A gentle S curve is the “photography standard” and you can start off there.
In theory anything that has a “value”. In practice tones again. You can also do stuff with colours though.
Opening a photo editing software and seeing the “curves section” you may see 4 circles, one is kinda white, the other 3 are red green and blue. In the L (“luminance”)/white circle you are once again just playing around with tones. The fun part begins now. You can selectively edit specific colours based on how “bright” or how “dark they are.
There is also such a thing as RGB curve present in darktable. I imagine this as “changing colours along with curves, not just tones”. This has to do something with some nerdy shit you can read down below.
By adding a point juuuuust barely almost at the lower end and pulling it up a teeeeny tiny bit, you can create something of a “pleasing film look”, which just means you have “brightened” the blacks and shadows a little, and made them “less harsh”. This is similar to the tip prior in tones about creating “aesthetic” blacks. You can look at it as doing the same thing in a different way.
One more word you’ll bump into: splines. In human speak, a spline is just the smooth curved line the editor threads through the points you drop — instead of joining them with straight segments — so your tones flow softly from one to the next. This demo even lets you feel the two flavours real editors use, with the Smooth / Monotone toggle on the curve explained by AI:
Flip between them on the very same points and you’ll see it right away. Neither is “correct”. Just another tool in the tool box.
The most powerful and surgical tonal tool, and the one that rewards understanding.
The mental model. X-axis = input brightness, Y-axis = output brightness. The identity is a 45° diagonal (output = input, no change). Raise a point and that input range gets brighter; lower it and it darkens. The slope of the line at any point = local contrast there: steeper than 45° = more contrast (tones pushed apart), flatter = less (tones compressed together). That's the whole tool.
RGB curves — colour, not just tone. Editing the individual R, G, B channels shifts colour, and does it per tonal zone, which is the superpower:
Luminance (L) curve vs RGB curves. RapidRAW's Curves panel exposes L, R, G, B. An L-only curve changes brightness while largely preserving hue — good for clean contrast. RGB-channel curves shift colour. Note the saturation side effect again: adding contrast via an RGB (or the L applied naively) curve raises saturation because it stretches channel separation; some editors let the curve run in a mode that preserves saturation/hue. Worth a line: a contrast S-curve is also a saturation boost — factor that in before you also reach for the saturation slider.
Parametric vs point curves. Point curves = you place the points. Parametric curves (Lightroom's Highlights/Lights/Darks/Shadows region sliders) = you move fixed tonal regions with sliders that can't create kinks — safer, less precise.
Deeper (encoding matters). A curve behaves very differently depending on whether it sits on linear or log/gamma-encoded values. The same visual "S" applied to linear-scene data vs display-encoded data gives different results; this is part of why darktable's rgb curve and tone curve modules warn you about working space, and why scene-referred contrast is often done with the tone mapper + tone equalizer rather than a raw curve.
The next logical step to figure out is colour and here is where it kind of gets complicated (at the technical level).
More specific about white balance. Or in human speak: “How exactly white is this white?”
This is an extremely important step in working with colour. “Get the white right or go black and white!”
Basics already mentioned in photography field notes. Here’s a TL;DR
Auto WB will be probably weirdly wrong under low quality artificial light (which is almost every artificial light source nowadays, since no one is going to pay 30 euros for a single light bulb). Same thing may happen when the scene has large TV screens or a scene is “monotone” i.e. a green wall of plants. Under such conditions WB is always going to be an uphill battle.
When outside, WB changes with the passage of every cloud so your whites may be off every couple of minutes or even seconds when WB is set to manual. Sometimes auto WB doesn’t notice the small changes and may be a little off.
You may not want to have the WB always “accurate”. If you want to give a scene “warm” feelings → make it warmer. If you want the scene to be “dark and gloomy” make it cold.
Here's some extra nerdy AI generated information about different light spectra coming from different light sources and other WB related pearls:
Blackbody radiators. An idealised object that glows purely because it’s hot — its colour depends only on its temperature. This is where “colour temperature in Kelvin” comes from. Incandescent bulbs are nearly true blackbodies (~2700–3200 K), and the sun is close (~5800 K). Both emit a smooth, continuous spectrum — every wavelength present. That’s exactly why a simple temperature slider works on them: the light is well-behaved.
Daylight isn’t quite a blackbody — the atmosphere scatters it, so it follows a slightly different track (the CIE “daylight locus”, which gives us D50 and D65). Close enough that the Kelvin model still holds.
LED and fluorescent are not thermal radiators at all. They make light by other means, and their spectra are spiky and full of holes — whole bands of wavelengths are simply missing. For those we can only quote a Correlated Colour Temperature (CCT): the blackbody whose colour the light happens to resemble. Which leads to the killer fact: two lights can share the identical colour temperature and still render your subject completely differently. CCT tells you what colour the light looks, not what it’s made of.
Why colours go wrong under cheap light. Picture three spectra side by side: sunlight is a smooth continuous curve; incandescent is smooth but tilted warm; a cheap LED is a violent blue spike, a broad hump, and a canyon where the deep reds and cyans should be. Now the punchline — if the light never emitted a wavelength, then an object that only reflects that wavelength has nothing to reflect. It comes back muddy, grey, wrong. And no white-balance slider can recover it, because white balance is just three channel multipliers; it cannot invent light that was never there. This is metamerism biting you, and it’s why skin looks sallow under bad bulbs.
CRI (Colour Rendering Index, 0–100) is the score for this: how faithfully a light renders colour versus a reference. Cheap bulbs sit around 70–80. Worth knowing it’s a flawed and gameable metric — the modern replacement is TM-30, which reports both fidelity and whether the light secretly over-saturates.
The key is to be shooting in raw. JPEGs don’t hold as much information (as such are much smaller files). Upon coming to the white balance sliders (temperature and tint) there is also usually a colour picker (or somewhere else in the interface, depends on the software). With this click on something that is supposed to be really white or exactly middle grey.
If the colours still look off, try to click around other white and/or grey areas of the image and settle on the one that feels the closest to reality/your vision. Then nudge the sliders ever so slightly until it feels right. White balance can be fiddly, so be patient. It is possible that by looking again at the photo after a couple of minutes you will nudge it in the other direction, think to yourself “Now it’s perfect!”, then come back even later do the same (and land on the exact same value you were at prior) and think to yourself “Now it’s perfect!”. If you want to know precisely here’s some nerdy techy AI generated info:
Vectorscopes (these get their own chapter later, so just the headline here). A vectorscope is a circular plot of hue and saturation only — brightness is thrown away entirely. Hue is the angle, saturation is the distance from the centre. Its killer feature: the skin-tone line, a diagonal reference running through the plot. Here’s the great fact — human skin of every ethnicity falls along that same line. What differs between people is only how far out along it they sit (saturation), not the angle. So if your subject’s skin cluster drifts off that line, your white balance or your grade is off. It is the most objective skin check that exists.
The neutral surround. The relevant standards (ISO 3664 for viewing conditions, ISO 12646 for monitors) call for a neutral grey surround — not white, not colourful. A bright or tinted surround makes your eye adapt wrongly and you will mis-judge every colour decision that follows. A neutral border around the image also helps your eye lock onto a true white reference. This, by the way, is the entire reason this guide ships a Neutral grey theme.
Monitor calibration (also its own chapter later). An uncalibrated screen lies to you — too blue, too bright, too contrasty — and every colour decision you make inherits the lie. A hardware colorimeter measures the screen and builds a profile so it tells the truth instead.
This is an integral part of getting the colours to look right, having nice skin tones, white dress, white snow, no weird sickly looking green people, no pink people and so on.
To be frank with you, You will edit 100 photos and won’t have to change the WB or just nudge it by a couple of degrees and then 10 where you have to really nail it down that will make you hate colour.
There is still the option of entirely skipping this step by choosing to keep the photo B&W and calling it art. Job done. Thanks for listening to my TedTalk. (I do it too 🤫)
Here we will be talking about HSL, Saturation X Vibrance, colour grading and RGB primaries / colour calibration
First we need to clarify a couple of terms. These are:
Short answer: no, and it’s worth thirty seconds. Chromaticity is the colour coordinate itself — hue plus colourfulness, with brightness stripped out entirely (“what colour is it, ignoring how bright”). Colourfulness is the raw perceived amount of colour. Chroma is colourfulness judged relative to a white lit the same way — roughly “how colourful is this, for this lighting”. Saturation is colourfulness judged relative to its own brightness — roughly, saturation ≈ chroma ÷ lightness.
Why anyone cares: because saturation is brightness-relative, pushing saturation drags brightness around with it — bright things get more colour dumped in and can bleach or clip. Pushing chroma adds colourfulness while leaving lightness where it was. That’s exactly why darktable’s color balance rgb hands you both separately, and why chroma is usually the cleaner of the two. For everything in this chapter, “Saturation” is the word you need — just know the others exist and aren’t interchangeable.
Same story one floor down: luminance is the physical, perceptually-weighted amount of light; lightness is that luminance bent onto the curve of human perception. And the “L” in HSL is honestly neither — it’s a cheap formula that isn’t perceptually correct at all, which is precisely why darktable avoids the term and talks about luminance and brilliance instead.
These 3 get abbreviated to HSL and since they are usually sliders, that comes out to "HSL sliders". Your one stop shop for (most of the things) colour. You can do a lot with this. Try to just play around.
These are the 3 main “concepts” you are working with when touching up colour. At this step you are working with - let's call them - "finished" colours. You are not changing the building blocks of the individual colours (that will come later). Only the final product of the process that creates the individual colours you see (explanation later).
Less is always more (yet again). If you dragged a slider somewhere and you can see that you dragged it, it's probably already too much. Back it off until you're not sure it's on, then leave it. Once you know, you know.
Most of us understand clearly what saturation is. If I tell you something is “vibrant” you will probably also thing of “rich” colours. But there is a difference. In Human speak: Saturation is dumb and “boosts” everything (even things you don’t want like skin tones). Vibrance is more sofisticated (it should protect already saturated tones). In practice that means reach for the vibrance slider first.
In human speak: Change the saturation of the highlights, or shadows, or mid tones selectively. Each editor has a different implementation and name for this. This is very useful for example for protecting skin tones when photographing portraits. How exactly? You can read here in a sophisticated AI generated explanation (this time not for nerds):
Lowering the saturation of the highlights pulls the orange/yellow cast out of the bright side of skin — the lit cheek, forehead, nose — where skin tends to go too warm/saturated. It reads as cleaner, more natural portraits instantly. Highlights desaturated slightly, shadows/mids left alone.
Now. Let’s call this the “mood generator”. If you wish the “vibe generator”. You may have noticed this in modern film making. What am I talking about? Teal shadows and amber highlights, which means “cool shadows, warm lights). Very trendy on YT in the recent years as well (source: it was revealed to me in a dream).
Pushing hues selectively in the different tonal zones. that is Shadows / Midtones / Highlights. And since the industry standard is for this to be in “wheels” (exception = darktable’s color balance rgb module), that’s where the name comes from. There is another terminology (the old school classic) lift / gamma / gain. If you ever see.. it’s the same. Try it out for yourself!
Really tiny changes make massive changes, tread carefully, if you can see it - you probably overdid it.
Here are the exact nerdy definitions generated by AI:
When writing this section I specifically asked AI to fact check me. I failed, so here's the full explanation (in human speak)
Remember what I said about HSL? “At this step you are working with - let's call them - "finished" colours. You are not changing the building blocks of the individual colours”. I promised, I shall deliver. This is the building blocks. (this is still the human me speaking)
You were probably taught that red, green and blue mix together to make every colour there is. They don’t. Three colours can only ever mix into the colours that sit inside the triangle formed by those three corners — and human vision is not a triangle. It’s a big rounded shape, and any triangle you draw inside it will always leave colours out in the corners. So RGB doesn’t make “all the colours there are”. It makes all the colours your screen or camera can manage — and that’s a smaller set. (This is the whole reason different colour spaces exist, and we’ll come back to it.)
They’re the three corners of that triangle — the specific red, the specific green, the specific blue that everything else in your image gets mixed from. Not “the red things in your photo”. The red that all reds are built out of.
And that’s the difference from everything we’ve done so far:
You have at your disposal the hue and the saturation (sometimes called purity) of each the red, green and blue primary colours, here you can try to play around with it a little. Other software may use “colour calibration” work in a different way, but in its essence, it is still the same in principle.
Each primary has a hue (rotate that corner around the wheel) and a saturation — sometimes called purity (drag that corner outward toward pure colour, or inward toward white). darktable’s primaries module literally calls it purity.
What’s happening under the hood, one sentence and then we’re done: it’s a channel mixer. Instead of “the new red = the old red”, it says “the new red = mostly the old red, plus a squeeze of green, minus a touch of blue”. Rebuild the corners, and every colour mixed from them shifts in one consistent motion. That’s why it feels so different from a normal slider — nothing is being selected, everything is being re-derived.
Where to find it. RapidRAW: Color Calibration → Primaries (R/G/B, each with Hue/Sat). darktable: the color calibration module (the channel mixer) and the newer primaries module. Lightroom: the old Calibration panel — which, fun fact, is the secret sauce behind half the famous presets people pay money for.
The honest word of warning. This is a blunt instrument at the root of the pipeline. A tiny nudge here does more than a big nudge anywhere else, and if you break it, you break everything downstream. Most photos never need it. Learn it, know it’s there, and use it when nothing else will do the job.
You can use the RGB primaries / colour calibration. to either:
Here’s some nerdy AI generated info for extra nerdyness:
The deepest colour layer.
What it does. Every colour in the image is mixed from three primaries — the red, green and blue the pipeline is built on. HSL retargets one hue band; grading pushes colour per tonal zone; but moving a primary's hue/saturation remaps every colour derived from it, at the root. Rotate the blue primary and every blue-descended tone (skies, shadows, teals) rotates with it, coherently.
Where. RapidRAW: Color Calibration → Primaries (R/G/B, each with Hue/Sat) + a Shadows Tint. darktable: the color calibration module (a channel mixer built on a chromatic-adaptation transform) and the newer primaries module. Lightroom: the legacy Calibration panel (the basis of half the famous preset "colour sciences").
Three real uses. (1) Fix a stubborn cast at the root when WB + HSL can't. (2) Build a signature colour science that carries across a whole body of work. (3) Feed the tone mapper cleaner primaries so filmic/AgX shift hues less as they compress — a pro move for controlling how brights render.