SESSION 08 / 20 Color & Tone Phase 3

Color Correction

Fixing exposure is half the job β€” colour is the other half. This session covers GIMP's three dedicated colour tools: Hue-Saturation for global and range-specific colour shifts, Color Balance for intuitive cast correction across tonal zones, and per-channel Curves revisited for precision colour matching between images.

1hSession Length
+15mReview
4Phases
πŸ’‘
0:00 – 0:10  Β·  ⏱ 10m
Concept Brief β€” The Colour Wheel and Cast Diagnosis
β–Ό

A colour cast is an unwanted tint that shifts neutral areas (whites, greys, black) away from true neutral. It has a source β€” tungsten light is orange, open shade is blue, fluorescent is green β€” and a fix. Every cast is neutralised by adding its opposite on the colour wheel. The colour wheel for correction purposes has three opposing pairs: Red ↔ Cyan, Green ↔ Magenta, Blue ↔ Yellow. Adding more of a colour's opposite removes that colour's influence.

Orange/warm cast

Source: Tungsten / incandescent light, golden hour. Fix: reduce Red + Yellow, or increase Blue + Cyan.

Blue/cool cast

Source: Open shade, overcast sky, wrong white balance. Fix: reduce Blue, or increase Yellow + Red (warm up).

Green cast

Source: Fluorescent lighting, foliage reflection. Fix: reduce Green, or increase Magenta.

Magenta cast

Source: Some LED sources, pink reflectors. Fix: reduce Magenta / increase Green.

Cyan cast

Source: Water reflection, some fluorescents. Fix: reduce Cyan / increase Red.

Yellow cast

Source: Old fluorescents, certain LED types. Fix: reduce Yellow / increase Blue.

To identify a cast: find the most neutral area in your image (a white wall, a grey card, a white shirt) and sample its colour with the Color Picker tool. If the RGB values are unequal β€” e.g. R:240 G:215 B:190 β€” the image has a warm/orange cast (Red is highest). The correction strategy always targets neutral areas first: when the whites are neutral, the rest of the image falls into place.

βœ…

Skill unlock: After this session you can identify any colour cast by name, choose the right tool to fix it, adjust saturation globally or by colour range, and match the colour temperature between two photographs β€” the essential skill for compositing.

  • 01
    Hue-Saturation β€” Global Controls

    Open Colors β†’ Hue-Saturation. Three sliders act on all colours simultaneously when the "All" preset is selected at the top of the dialog. Hue rotates every colour in the image around the colour wheel β€” at +30 all colours shift 30Β° toward warmer tones; at βˆ’30 toward cooler tones. This is rarely useful for correction (it changes every colour including skin tones) but is powerful for creative colour shifts or recolouring objects. Saturation scales the intensity of every colour: positive values increase vividness, negative values reduce it. Zero saturation produces a greyscale image while preserving luminosity. Lightness (distinct from Levels/Curves brightness) uniformly shifts all pixels toward white or black β€” a cruder tool than Curves and best avoided for exposure correction. Use it only for deliberate stylistic lightening or darkening where tonal precision is not required.

    Colors β†’ Hue-Saturation β†’ "All" selected β†’ adjust Hue / Saturation / Lightness
  • 02
    Hue-Saturation β€” Per-Colour-Range Targeting

    The real power of Hue-Saturation is its colour range selector. Instead of "All", choose a specific colour: R (Reds), Y (Yellows), G (Greens), C (Cyans), B (Blues), or M (Magentas). The sliders now affect only pixels in that colour range. Practical applications: select Reds, reduce Saturation to βˆ’20 to desaturate ruddy or overly red skin tones without affecting other colours. Select Blues, shift Hue by +10 to push a sky from a flat blue toward a more saturated cyan-blue. Select Yellows, increase Saturation to boost autumn foliage without over-saturating the sky. This per-range targeting is the most precise saturation control available in GIMP without using selections.

    Colors β†’ Hue-Saturation β†’ click colour range button (R, Y, G, C, B, M) β†’ adjust sliders for that range only
  • 03
    Hue-Saturation β€” Overlap and Overlap Width

    When targeting a colour range, the Overlap slider controls how smoothly the adjustment transitions at the edges of that colour range. A low Overlap produces a sharp cut β€” only pure Reds are affected, and pixels near the boundary between Red and Orange are mostly unaffected. A high Overlap blends the adjustment gradually into neighbouring ranges β€” useful when the target colour grade into adjacent colours you also want to shift. Increase Overlap when your subject colour is near a range boundary (for example, an orange-red that falls between R and Y) and the adjustment is cutting off too soon.

    Colors β†’ Hue-Saturation β†’ select colour range β†’ Overlap slider controls edge blending
  • 04
    Color Balance β€” Intuitive Cast Correction by Tonal Zone

    Open Colors β†’ Color Balance. This dialog presents three sliders for each of three tonal zones (Shadows, Midtones, Highlights), each slider pushing colour between two opposite poles: Cyan ↔ Red, Magenta ↔ Green, Yellow ↔ Blue. The interface is deliberately intuitive β€” you describe the symptom in plain terms ("my highlights are too warm") and push the slider away from the problem. Select Highlights in the tonal range selector, then push the Yellow-Blue slider toward Blue to cool the highlights. Switch to Shadows and push toward Yellow-Red to warm the shadows. The Preserve Luminosity checkbox (enabled by default) prevents the brightness from shifting as you push colour β€” leave it enabled for all correction work.

    Colors β†’ Color Balance β†’ select Shadows / Midtones / Highlights β†’ push sliders toward opposite of the cast
  • 05
    Color Balance vs Curves for Cast Correction β€” Which to Use

    Color Balance and per-channel Curves both fix colour casts, but they suit different situations. Color Balance is faster and more intuitive: push away from the problem colour in the appropriate tonal zone. Use it for broad, visible casts where you know where the problem is (warm highlights, cool shadows). Per-channel Curves (from Session 7) are more precise: you can target a very narrow tonal range within a channel, add a control point at exactly the tonal value causing the problem, and adjust only that region. Use per-channel Curves when Color Balance is overcorrecting (it affects a wider tonal zone than needed) or when you need to correct one tonal zone without disturbing an adjacent one.

    Color Balance: broad cast correction by zone. Per-channel Curves: narrow tonal range within a single channel.
  • 06
    Using the Color Picker to Sample Neutrals

    Before applying any colour correction, use the Color Picker tool (O) to measure the RGB values of a known-neutral area. Activate the Color Picker, set Sample Merged in Tool Options, and click on the whitest or most neutral area visible in the image. The foreground colour swatch updates to the sampled colour, and the exact RGB values are shown in the foreground colour dialog (double-click the swatch to open it). If R, G, and B are unequal, the difference between the highest and lowest channel value indicates the cast. R:230 G:215 B:200 means: too much Red (warm cast), not enough Blue. The correction is to reduce Red by ~30 units or increase Blue by ~30 units β€” now you have a numerical target for your Curves or Color Balance adjustment.

    Color Picker (O) β†’ click neutral area β†’ double-click foreground swatch β†’ read R, G, B values β†’ note which channels are unequal
  • 07
    Colour Matching Between Two Images

    Compositing (Sessions 11–12) requires that the subject and background look like they were photographed in the same light. Colour matching is the process of adjusting one image's colour to resemble another's. The workflow: sample three neutral or representative areas in the background image with the Color Picker and note the approximate RGB values for shadows, midtones, and highlights. Sample the equivalent tonal zones in the subject image. The difference between them is what you need to correct on the subject layer. Use per-channel Curves on the subject layer: in each channel (R, G, B), adjust the curve so the subject's shadow, midtone, and highlight values approximate the background's equivalent values. This is an iterative process β€” adjust, compare, adjust again β€” but three to five Curves adjustments typically produce a convincing match.

    Sample RGB values in both images at equivalent tonal zones β†’ apply per-channel Curves to subject to match values
  • 08
    Selective Colour Correction Using Selections + Hue-Saturation

    Sometimes a colour problem exists only in a specific area β€” a sky that's too saturated while the landscape is fine, or a product that's the wrong shade of red in a photo. Combine a selection from Sessions 4–5 with Hue-Saturation for targeted correction: select the problematic region (Select by Color for a sky, Fuzzy Select for a product), feather the selection by 1–2 pixels to avoid a hard edge, then open Colors β†’ Hue-Saturation and adjust. The change applies only within the selection. Deselect when done. This combination of selections + colour tools is one of the most versatile techniques in all of image editing.

    Make selection (S4/S5 tools) β†’ feather 1–2px β†’ Colors β†’ Hue-Saturation β†’ adjust within selection only
  • 09
    Desaturation β€” Greyscale Conversion Approaches

    GIMP offers multiple ways to remove colour, each producing different luminosity results. Colors β†’ Hue-Saturation β†’ Saturation βˆ’100: removes colour while using the HSL lightness formula β€” fast but often produces flat-looking greyscale with compressed tonal range. Colors β†’ Desaturate β†’ Luminosity: uses the perceptual luminosity formula (weighted RΓ—0.2126 + GΓ—0.7152 + BΓ—0.0722), matching how the human eye perceives brightness. Produces the most natural greyscale with the best tonal separation. Colors β†’ Desaturate β†’ Luma: similar to Luminosity but uses the linear light formula β€” slightly different result, particularly in very bright or very dark tones. For professional greyscale conversion, Luminosity is the default recommendation. For creative control over the greyscale mix, use the channel mixer: Colors β†’ Components β†’ Channel Mixer to manually weight each channel.

    Colors β†’ Desaturate β†’ Luminosity (best default). Or: Colors β†’ Components β†’ Channel Mixer for manual channel weighting.
  • 10
    Tool Decision Reference β€” All Colour Tools

    A summary of when to reach for each colour correction tool:

    ToolBest forLimitation
    Levels eyedroppers One-click white balance on images with a clearly neutral white or grey area No tonal zone targeting; affects the full image uniformly
    Color Balance Intuitive cast correction separated into Shadows / Midtones / Highlights Broad tonal zones; less precise than per-channel Curves
    Hue-Saturation (All) Global saturation boost/reduction; creative hue rotation Affects every colour equally; can shift skin tones undesirably
    Hue-Saturation (range) Targeting one colour range only β€” sky blues, skin reds, foliage greens Range boundaries are fixed to the six primary/secondary colours
    Per-channel Curves Precise cast correction targeting any specific tonal range within a channel; colour grading Steeper learning curve; requires understanding of channel colour relationships
    Selection + any tool Isolating a colour correction to one region without affecting the rest Requires making an accurate selection first
⚠️

Perceptual note: Your monitor's calibration directly affects how you perceive colour. An uncalibrated monitor may make correct images look warm or cool, leading to over-correction. If you notice your corrections consistently going too far in one direction, view the file on a second screen or share it and check it on another device. A hardware calibrator (X-Rite, Datacolor) is the definitive solution for critical colour work.

Open a portrait or interior photograph with a visible colour cast β€” ideally one shot under mixed or artificial lighting. You will diagnose the cast, correct it, then refine saturation selectively.

  • A
    Diagnose the cast numerically

    Use the Color Picker (O) to click on the most neutral area visible β€” a white wall, white clothing, or light grey surface. Double-click the foreground colour swatch and read the R, G, B values. Which channel is highest? Which is lowest? State the cast: "This image has a warm/orange cast β€” Red is highest at 238, Blue is lowest at 195. The correction needs ~43 units of Blue increase or equivalent Red reduction."

  • B
    Correct the cast with Color Balance

    Open Colors β†’ Color Balance. Select Midtones. Push the Yellow-Blue slider toward Blue by approximately half your diagnosed difference (e.g. if Blue is 43 units low, push to +20 Blue and evaluate). Switch to Highlights and make a smaller adjustment in the same direction (Β±10). Switch to Shadows and make an even smaller adjustment (Β±5). Preserve Luminosity must stay on. Sample the neutral area again with the Color Picker β€” are the R, G, B values now closer to equal? Iterate until the neutral area measures within Β±8 units across all three channels.

  • C
    Refine skin-tone saturation with Hue-Saturation

    Open Colors β†’ Hue-Saturation. Select the Reds range. Reduce Saturation by βˆ’10 to βˆ’20 β€” this takes the edge off any redness in skin without affecting other colours. Switch to Yellows and check if the skin yellow tones also need slight reduction or if the overall skin colour looks natural. Toggle preview on/off (the Preview checkbox) to compare against the pre-adjustment state.

  • D
    Boost environmental colour selectively

    If the image contains a sky or foliage, select the Blues or Greens range in Hue-Saturation and increase Saturation by +10 to +20. The environmental colour becomes more vivid without touching the skin tones or neutral areas. Compare the final corrected layer against the original by toggling layer visibility.

πŸ’‘

Expected result: A corrected duplicate layer where neutral areas (sampled with the Color Picker) read within Β±8 units across R, G, and B. Skin tones look natural without over-saturation. Any sky or foliage is more vivid. Toggle the layer against the original and the improvement should be immediately obvious.

Open two photographs with different colour temperatures β€” one warm (golden hour, tungsten) and one cool (shade, overcast). Place one above the other. Using only Color Balance and per-channel Curves, make the top image's colour closely match the bottom image's colour temperature. No step-by-step guidance.

  • β†’
    Sample, compare, and correct

    Use the Color Picker to sample equivalent neutral or mid-tone areas in both images. Note the R, G, B difference. Apply Color Balance to the top layer to bring its overall temperature in line with the bottom. Use per-channel Curves for fine-tuning: if Color Balance gets the midtones right but the shadows are still mismatched, fix the shadow region of the Red or Blue channel curve independently. The goal: sample the same equivalent area in both images and achieve R, G, B values within 15 units of each other.

  • β†’
    Evaluate the composite result

    Set the top layer to 50% opacity temporarily. The two images should blend without a jarring colour seam. If the blend looks unnatural β€” one side noticeably warmer or cooler β€” your colour match is not yet close enough. Identify which tonal zone is mismatched (is it the shadows, the midtones, or the highlights?) and target it specifically with Color Balance on that zone. This preview test is the same evaluation used in professional compositing before final output.

πŸ’‘

Hint: A 50% opacity overlap test works best when both images share some tonal content at the overlap point. If the images are compositionally very different (e.g. a portrait and a wide landscape), place them side-by-side in a single wide canvas at 100% opacity and compare a neutral area sampled from each.

Your progress is saved locally in your browser.
βœ“   Session 8 complete. Up next: Session 9 β€” Retouching I β†’
πŸ”
+ 15 minutes

Session 8 Review

15:00 review
Key Concepts
  • Every colour cast has an opposite on the colour wheel β€” the fix is always to add more of the opposite colour
  • Diagnosis: sample a neutral area with Color Picker; unequal R, G, B values identify the cast and its magnitude
  • Hue-Saturation "All": global hue rotation, saturation scaling, and lightness shift
  • Hue-Saturation range selector: adjusts only pixels in a specific colour family (R, Y, G, C, B, M)
  • Color Balance: Cyan↔Red, Magenta↔Green, Yellow↔Blue sliders across Shadows / Midtones / Highlights zones
  • Preserve Luminosity in Color Balance prevents brightness shift during colour correction β€” keep it on
  • Colour matching: sample neutrals in both images, use per-channel Curves to bring values within ~15 units
  • Greyscale conversion: Desaturate β†’ Luminosity produces the best perceptual tonal result
Common Mistakes
  • Correcting by eye without sampling neutrals β€” visual adaptation makes the eye unreliable for cast detection after a few minutes
  • Over-saturating β€” increasing global saturation by +30 or more often produces garish, unnatural colours; +10 to +15 is usually the limit
  • Using Hue-Saturation Lightness instead of Curves for brightness β€” produces flat, tonally compressed results
  • Forgetting Preserve Luminosity in Color Balance β€” colour shifts then also change perceived brightness
  • Applying global colour correction when the problem is localised β€” use a selection to restrict the correction to the affected area
Quick Quiz
  • You sample a white wall and get R:240 G:240 B:200. What colour cast does this indicate, and what is the correction direction?
  • How do you restrict Hue-Saturation to affect only the blue sky without touching skin tones?
  • What does the Preserve Luminosity option in Color Balance prevent?
  • Which desaturation method produces the most perceptually accurate greyscale β€” Desaturate β†’ Luminosity or Hue-Saturation βˆ’100? Why?
  • What numerical target do you aim for when colour-matching two images using the Color Picker?
Next Session Preview
  • Session 9 begins Phase 4 β€” Retouching β€” with the Clone Tool and Heal Tool
  • These tools remove distractions, blemishes, and unwanted objects from images
  • The colour and tonal correction skills from Sessions 7–8 are prerequisites: retouching a poorly exposed or colour-cast image produces inconsistent results
  • Prepare a photograph with something to remove: a spot, a background distraction, a wire, a piece of rubbish β€” anything you wish wasn't there
πŸ“š Best Resources