{"course":{"slug":"image-editing","title":"Editing an Image","url":"https://addaly.com/learn/image-editing","level":"beginner","track":"create"},"lesson":{"slug":"exposure-blending-and-hdr","title":"Two exposures of one scene, and how not to make it look like HDR","url":"https://addaly.com/learn/image-editing/exposure-blending-and-hdr","position":55,"of":78,"minutes":9,"answer":"Exposure fusion picks the best-exposed pixels and needs no tone mapping, while HDR merge plus tone mapping flattens global contrast and restores it locally, which is exactly why halos and the characteristic grey look appear.","keyIdea":"Exposure fusion picks the best-exposed pixels and needs no tone mapping, while HDR merge plus tone mapping flattens global contrast and restores it locally, which is exactly why halos and the characteristic grey look appear, and a hand-masked blend of two exposures beats both."},"body":{"format":"markdown","text":"## The problem: the scene exceeds the sensor\n\nA room with a window. Expose for the room and the window is pure white. Expose for the window and the room is black. The scene's dynamic range (perhaps 14 or 15 stops) exceeds what a single exposure can hold with detail at both ends.\n\nThe answer is more than one exposure. Bracket: three frames at −2, 0 and +2 stops, or five at one-stop intervals for an extreme case. Use a tripod, or at minimum brace the camera, and shoot in rapid succession.\n\nThe important detail: **change the shutter speed, not the aperture**. Changing aperture changes depth of field between frames, so the images will not align in sharpness. Most cameras' auto-bracket does this correctly; check.\n\n## Two ways to combine them, and they give different results\n\n**Exposure fusion** picks the best-exposed pixels from each frame and blends them, weighted by how well-exposed each one is. It never builds a high-dynamic-range image at all; it goes straight to a normal-range result. This is what **enfuse** does, and it is what Photoshop's *Merge to HDR* produces when you use the more conservative settings.\n\n**HDR merge plus tone mapping** first builds a genuine high-dynamic-range file holding the full range in floating point, then compresses that range down to something displayable. The compression is the tone mapping, and it is where the characteristic look comes from.\n\nFor most photographs, **fusion gives a more natural result with less effort**. Tone mapping gives more control and more opportunities to produce something that announces itself.\n\n> Figure: Two ways to combine a bracket, and why one of them has a look. Exposure fusion, as enfuse does it: Picks the best-exposed pixels from each frame, weighted by how well exposed they are; Never builds a high-dynamic-range file at all, so nothing has to be tone mapped down; More natural for less effort, which is why phones do this for every shot; align_image_stack and enfuse are two free command lines, from Hugin. HDR merge plus tone mapping: Builds the full range in floating point, then compresses it to something displayable; The compression flattens global contrast; local contrast is raised to hide that; Halos at every strong edge, because local contrast compares each pixel to a blurred neighbourhood; The cure is less of it and a smaller radius, or not tone mapping at all. The grey, halo-ringed appearance is not a fashion, it is the mechanism: reducing global contrast to fit fifteen stops into a display makes the picture flat, and restoring local contrast to compensate puts a bright band along every roofline. A hand-masked blend of two exposures beats both and takes ten minutes.\n\n## Why tone-mapped HDR looks like that\n\nThe grey, flat, halo-ringed, over-detailed look is not a fashion. It is the mechanism.\n\nTo fit 15 stops into a displayable range, a tone mapper has to reduce global contrast substantially, which is why the image goes flat and grey. To stop it looking flat, it then increases **local** contrast, using exactly the blurred-neighbourhood comparison from the tone module. And local contrast at a large radius produces halos at every strong edge, which is why tone-mapped photographs have bright bands around every roofline and dark bands inside them.\n\nKnowing that tells you the cure: **less local contrast, and a smaller radius**. Or avoid tone mapping and use fusion.\n\n## The manual method, which is better\n\nThe result most professionals actually use, and it takes ten minutes:\n\n1. Load two or three exposures as layers in one document, aligned.\n2. Put the one exposed for the shadows at the bottom, the one exposed for the highlights on top.\n3. Add a **luminosity mask** to the top layer so it applies only where the underlying image is bright.\n4. Refine the mask by hand where needed.\n\nThat is it. You are choosing, per tonal region, which exposure to show, with a mask that has no edges anywhere, which is why it produces no halos at all. This is the technique from the luminosity mask lesson doing the job it is best at.\n\nFor a window in a room specifically: mask the window exactly (a rectangular selection usually) rather than by luminosity, because the boundary is a real hard edge and pretending otherwise gives a glow around the frame.\n\n## Alignment and ghosting\n\n**Alignment.** Handheld brackets shift between frames. Photoshop's *Auto-Align Layers* handles it; **align_image_stack**, part of Hugin and free, does it from the command line and is excellent:\n\n```bash\nalign_image_stack -m -a aligned_ img1.jpg img2.jpg img3.jpg\nenfuse -o merged.tif aligned_*.tif\n```\n\nThat is a complete free exposure-fusion pipeline in two lines.\n\n**Ghosting.** Anything that moved between frames (leaves, people, cars, water) appears in several positions at once, blended into a smear. Deghosting algorithms pick one frame as the reference for moving regions.\n\nThe manual fix is more reliable: pick one frame as the base for the moving area and mask the others away from it. A person walking through a bracketed scene is a mask, not an algorithm.\n\n## Free tools\n\n- **Hugin** (align_image_stack) and **enfuse**: alignment and fusion, free, excellent, command line.\n- **Luminance HDR**: free, cross-platform, several tone mapping operators including Mantiuk and Fattal. Useful for learning what the operators do.\n- **darktable**: has its own HDR merge, and separately its *exposure* and *tone equaliser* modules can recover a remarkable amount from a single raw file, which is often enough that bracketing is unnecessary.\n- **GIMP**: blend layers with luminosity masks manually, exactly as described above.\n- **Phones** already do this. Every modern phone camera bracket-and-fuses automatically for every shot, which is why phone photographs of backlit scenes look better than a dedicated camera's single frame.\n\n## When one raw file is enough\n\nA modern raw file holds 12 to 14 stops. For a scene with 11 or 12 stops of range, a single well-exposed raw plus shadow recovery gives a clean result with no alignment, no ghosting and no bracketing.\n\nThe test: expose so the highlights are just below clipping, then lift the shadows in the developer and look at the noise. If it is acceptable, you did not need a bracket. If the shadows are noisy and blotchy, you did: the sensor's noise floor is the real limit, and it is why bracketing still exists on cameras that claim 14 stops.\n\n## The position\n\nThe purpose of combining exposures is to produce an image that looks like the scene looked to you, which a single exposure could not. That is a defensible reason and it is what the eye does continuously by adapting as it scans.\n\nWhere it stops being defensible is the same place as everywhere else: when the combination asserts something the scene did not contain. A blended interior showing the real view through the window is honest. A blended interior with a view from a different day is a composite, and the estate agency conversation from the previous lesson applies."},"check":{"question":"A photographer merges a three-frame bracket of a cathedral interior with an HDR tone mapper and gets bright bands along every column edge and a flat grey overall look. What produced both symptoms?","options":[{"text":"Misalignment between the handheld frames, which shows as edge artefacts and reduced contrast after merging.","correct":false,"tempting":"Names a real bracket problem that produces doubled or ghosted edges rather than symmetrical bright bands, and does not explain the global flattening."},{"text":"The frames were bracketed by aperture rather than shutter speed, so the depth of field differs between them.","correct":false,"tempting":"Identifies a genuine bracketing mistake whose symptom is mismatched sharpness across the frame, not halos tracking every high-contrast boundary."},{"text":"Compressing fifteen stops into a displayable range flattens global contrast, and the local contrast added to compensate creates halos wherever the blurred neighbourhood spans a strong edge.","correct":true},{"text":"The merged file was 8-bit, so the tonal compression exhausted the available levels and banded along the highest-contrast edges.","correct":false,"tempting":"Diagnoses quantisation, which appears as stepped terracing across smooth gradients rather than as bands hugging every edge."}]},"prev":{"slug":"sky-replacement","title":"A new sky changes the light in the whole photograph","url":"https://addaly.com/learn/image-editing/sky-replacement"},"next":{"slug":"panoramas-and-stitching","title":"Many frames, one view, and the pivot point that matters","url":"https://addaly.com/learn/image-editing/panoramas-and-stitching"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'Two exposures of one scene, and how not to make it look like HDR', in Editing an Image. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/image-editing/exposure-blending-and-hdr (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"Two exposures of one scene, and how not to make it look like HDR","courseTitle":"Editing an Image","module":{"index":6,"slug":"compositing","title":"Compositing: making two pictures one"},"answer":null,"keyIdea":"Exposure fusion picks the best-exposed pixels and needs no tone mapping, while HDR merge plus tone mapping flattens global contrast and restores it locally, which is exactly why halos and the characteristic grey look appear, and a hand-masked blend of two exposures beats both.","hero":{"kind":"figure","figure":{"t":"figure","label":"Figure","figure":{"left":{"title":"Exposure fusion, as enfuse does it","items":["Picks the best-exposed pixels from each frame, weighted by how well exposed they are","Never builds a high-dynamic-range file at all, so nothing has to be tone mapped down","More natural for less effort, which is why phones do this for every shot","align_image_stack and enfuse are two free command lines, from Hugin"]},"type":"compare","right":{"title":"HDR merge plus tone mapping","items":["Builds the full range in floating point, then compresses it to something displayable","The compression flattens global contrast; local contrast is raised to hide that","Halos at every strong edge, because local contrast compares each pixel to a blurred neighbourhood","The cure is less of it and a smaller radius, or not tone mapping at all"]},"title":"Two ways to combine a bracket, and why one of them has a look","caption":"The grey, halo-ringed appearance is not a fashion, it is the mechanism: reducing global contrast to fit fifteen stops into a display makes the picture flat, and restoring local contrast to compensate puts a bright band along every roofline. A hand-masked blend of two exposures beats both and takes ten minutes."},"alt":"Figure: Two ways to combine a bracket, and why one of them has a look. Exposure fusion, as enfuse does it: Picks the best-exposed pixels from each frame, weighted by how well exposed they are; Never builds a high-dynamic-range file at all, so nothing has to be tone mapped down; More natural for less effort, which is why phones do this for every shot; align_image_stack and enfuse are two free command lines, from Hugin. HDR merge plus tone mapping: Builds the full range in floating point, then compresses it to something displayable; The compression flattens global contrast; local contrast is raised to hide that; Halos at every strong edge, because local contrast compares each pixel to a blurred neighbourhood; The cure is less of it and a smaller radius, or not tone mapping at all. The grey, halo-ringed appearance is not a fashion, it is the mechanism: reducing global contrast to fit fifteen stops into a display makes the picture flat, and restoring local contrast to compensate puts a bright band along every roofline. A hand-masked blend of two exposures beats both and takes ten minutes."}},"blocks":[{"t":"h2","id":"the-problem-the-scene-exceeds-the-sensor","runs":[{"text":"The problem: the scene exceeds the sensor"}]},{"t":"p","runs":[{"text":"A room with a window. Expose for the room and the window is pure white. Expose for the window and the room is black. The scene’s dynamic range (perhaps 14 or 15 stops) exceeds what a single exposure can hold with detail at both ends."}]},{"t":"p","runs":[{"text":"The answer is more than one exposure. Bracket: three frames at −2, 0 and +2 stops, or five at one-stop intervals for an extreme case. Use a tripod, or at minimum brace the camera, and shoot in rapid succession."}]},{"t":"p","runs":[{"text":"The important detail: "},{"text":"change the shutter speed, not the aperture","b":true},{"text":". Changing aperture changes depth of field between frames, so the images will not align in sharpness. Most cameras’ auto-bracket does this correctly; check."}]},{"t":"h2","id":"two-ways-to-combine-them-and-they-give-different-results","runs":[{"text":"Two ways to combine them, and they give different results"}]},{"t":"p","runs":[{"text":"Exposure fusion","b":true},{"text":" picks the best-exposed pixels from each frame and blends them, weighted by how well-exposed each one is. It never builds a high-dynamic-range image at all; it goes straight to a normal-range result. This is what "},{"text":"enfuse","b":true},{"text":" does, and it is what Photoshop’s "},{"text":"Merge to HDR","i":true},{"text":" produces when you use the more conservative settings."}]},{"t":"p","runs":[{"text":"HDR merge plus tone mapping","b":true},{"text":" first builds a genuine high-dynamic-range file holding the full range in floating point, then compresses that range down to something displayable. The compression is the tone mapping, and it is where the characteristic look comes from."}]},{"t":"p","runs":[{"text":"For most photographs, "},{"text":"fusion gives a more natural result with less effort","b":true},{"text":". Tone mapping gives more control and more opportunities to produce something that announces itself."}]},{"t":"h2","id":"why-tone-mapped-hdr-looks-like-that","runs":[{"text":"Why tone-mapped HDR looks like that"}]},{"t":"p","runs":[{"text":"The grey, flat, halo-ringed, over-detailed look is not a fashion. It is the mechanism."}]},{"t":"p","runs":[{"text":"To fit 15 stops into a displayable range, a tone mapper has to reduce global contrast substantially, which is why the image goes flat and grey. To stop it looking flat, it then increases "},{"text":"local","b":true},{"text":" contrast, using exactly the blurred-neighbourhood comparison from the tone module. And local contrast at a large radius produces halos at every strong edge, which is why tone-mapped photographs have bright bands around every roofline and dark bands inside them."}]},{"t":"p","runs":[{"text":"Knowing that tells you the cure: "},{"text":"less local contrast, and a smaller radius","b":true},{"text":". Or avoid tone mapping and use fusion."}]},{"t":"h2","id":"the-manual-method-which-is-better","runs":[{"text":"The manual method, which is better"}]},{"t":"p","runs":[{"text":"The result most professionals actually use, and it takes ten minutes:"}]},{"t":"ol","start":1,"items":[[{"text":"Load two or three exposures as layers in one document, aligned."}],[{"text":"Put the one exposed for the shadows at the bottom, the one exposed for the highlights on top."}],[{"text":"Add a "},{"text":"luminosity mask","b":true},{"text":" to the top layer so it applies only where the underlying image is bright."}],[{"text":"Refine the mask by hand where needed."}]]},{"t":"p","runs":[{"text":"That is it. You are choosing, per tonal region, which exposure to show, with a mask that has no edges anywhere, which is why it produces no halos at all. This is the technique from the luminosity mask lesson doing the job it is best at."}]},{"t":"p","runs":[{"text":"For a window in a room specifically: mask the window exactly (a rectangular selection usually) rather than by luminosity, because the boundary is a real hard edge and pretending otherwise gives a glow around the frame."}]},{"t":"h2","id":"alignment-and-ghosting","runs":[{"text":"Alignment and ghosting"}]},{"t":"p","runs":[{"text":"Alignment.","b":true},{"text":" Handheld brackets shift between frames. Photoshop’s "},{"text":"Auto-Align Layers","i":true},{"text":" handles it; "},{"text":"align_image_stack","b":true},{"text":", part of Hugin and free, does it from the command line and is excellent:"}]},{"t":"code","lang":"bash","label":"Bash","text":"align_image_stack -m -a aligned_ img1.jpg img2.jpg img3.jpg\nenfuse -o merged.tif aligned_*.tif"},{"t":"p","runs":[{"text":"That is a complete free exposure-fusion pipeline in two lines."}]},{"t":"p","runs":[{"text":"Ghosting.","b":true},{"text":" Anything that moved between frames (leaves, people, cars, water) appears in several positions at once, blended into a smear. Deghosting algorithms pick one frame as the reference for moving regions."}]},{"t":"p","runs":[{"text":"The manual fix is more reliable: pick one frame as the base for the moving area and mask the others away from it. A person walking through a bracketed scene is a mask, not an algorithm."}]},{"t":"h2","id":"free-tools","runs":[{"text":"Free tools"}]},{"t":"ul","items":[[{"text":"Hugin","b":true},{"text":" (align_image_stack) and "},{"text":"enfuse","b":true},{"text":": alignment and fusion, free, excellent, command line."}],[{"text":"Luminance HDR","b":true},{"text":": free, cross-platform, several tone mapping operators including Mantiuk and Fattal. Useful for learning what the operators do."}],[{"text":"darktable","b":true},{"text":": has its own HDR merge, and separately its "},{"text":"exposure","i":true},{"text":" and "},{"text":"tone equaliser","i":true},{"text":" modules can recover a remarkable amount from a single raw file, which is often enough that bracketing is unnecessary."}],[{"text":"GIMP","b":true},{"text":": blend layers with luminosity masks manually, exactly as described above."}],[{"text":"Phones","b":true},{"text":" already do this. Every modern phone camera bracket-and-fuses automatically for every shot, which is why phone photographs of backlit scenes look better than a dedicated camera’s single frame."}]]},{"t":"h2","id":"when-one-raw-file-is-enough","runs":[{"text":"When one raw file is enough"}]},{"t":"p","runs":[{"text":"A modern raw file holds 12 to 14 stops. For a scene with 11 or 12 stops of range, a single well-exposed raw plus shadow recovery gives a clean result with no alignment, no ghosting and no bracketing."}]},{"t":"p","runs":[{"text":"The test: expose so the highlights are just below clipping, then lift the shadows in the developer and look at the noise. If it is acceptable, you did not need a bracket. If the shadows are noisy and blotchy, you did: the sensor’s noise floor is the real limit, and it is why bracketing still exists on cameras that claim 14 stops."}]},{"t":"h2","id":"the-position","runs":[{"text":"The position"}]},{"t":"p","runs":[{"text":"The purpose of combining exposures is to produce an image that looks like the scene looked to you, which a single exposure could not. That is a defensible reason and it is what the eye does continuously by adapting as it scans."}]},{"t":"p","runs":[{"text":"Where it stops being defensible is the same place as everywhere else: when the combination asserts something the scene did not contain. A blended interior showing the real view through the window is honest. A blended interior with a view from a different day is a composite, and the estate agency conversation from the previous lesson applies."}]}],"check":{"question":[{"text":"A photographer merges a three-frame bracket of a cathedral interior with an HDR tone mapper and gets bright bands along every column edge and a flat grey overall look. What produced both symptoms?"}],"options":[{"text":[{"text":"Misalignment between the handheld frames, which shows as edge artefacts and reduced contrast after merging."}],"correct":false,"tempting":[{"text":"Names a real bracket problem that produces doubled or ghosted edges rather than symmetrical bright bands, and does not explain the global flattening."}]},{"text":[{"text":"The frames were bracketed by aperture rather than shutter speed, so the depth of field differs between them."}],"correct":false,"tempting":[{"text":"Identifies a genuine bracketing mistake whose symptom is mismatched sharpness across the frame, not halos tracking every high-contrast boundary."}]},{"text":[{"text":"Compressing fifteen stops into a displayable range flattens global contrast, and the local contrast added to compensate creates halos wherever the blurred neighbourhood spans a strong edge."}],"correct":true,"tempting":null},{"text":[{"text":"The merged file was 8-bit, so the tonal compression exhausted the available levels and banded along the highest-contrast edges."}],"correct":false,"tempting":[{"text":"Diagnoses quantisation, which appears as stepped terracing across smooth gradients rather than as bands hugging every edge."}]}],"order":[0,2,1,3],"why":null},"prev":{"slug":"sky-replacement","title":"A new sky changes the light in the whole photograph","minutes":8},"next":{"slug":"panoramas-and-stitching","title":"Many frames, one view, and the pivot point that matters","minutes":9}}}