Video Edit Preview Colors Look Right but the Exported File Looks Different? Color Space, Profile & Playback Diagnosis Guide

2026-11-17 | GPU Repair | 41 minutes | HK GPU Fix Editorial Team

Video Edit Preview Colors Look Right but the Exported File Looks Different? Color Space, Profile & Playback Diagnosis Guide

Key Summary

The preview in the editing software looks right, but the exported file comes out washed out, grey, warmer, or colder. This guide separates a playback reading fault from an export labelling fault, using player and file-tag comparisons to find which layer the difference belongs to. Work through the five layers in order, changing one setting at a time.

The Shape of the Problem: Preview Right, Export Wrong

This guide covers one specific shape, and the shape itself is the whole diagnosis. The timeline inside the editing software looks right. The colours on the preview monitor match what the editor intended: skin looks natural, whites look white, and the grade looks the way it was set. The export then runs to completion without an error. The finished file is opened, and the colours are different. The picture may look washed out, or grey, or warmer, or colder, or darker, or lighter than the preview. The picture itself is intact — there is no broken image, no coloured block, no flicker, and no black frame — only the colour is wrong.

Three details define the shape, and all three have to be present. First, the preview is right. The colours in the editing software, on the timeline, are the colours the editor intended. Second, the export completes. There is no error message, no stop halfway, and no failed render. Third, the finished file is wrong in colour only. The picture is there, the sound is there, and the length is right, but the colours do not match the preview. A file that fails to export, a file with a broken picture, or a preview that was already wrong are all different shapes, and they belong with different guides.

That combination is what makes this symptom different from its neighbours. One neighbouring shape is a finished file with a broken picture — green blocks, a torn image, a flash, or a black frame — where the preview colours were right but the picture itself is damaged. That shape is covered in the guide numbered 493, and it is not this one: here the picture is whole and only the colour has shifted. Another neighbouring shape is the exact mirror image: the preview itself is damaged, with a green or purple screen, while the finished file comes out normal. That shape is covered in the guide numbered 488, and it runs in the opposite direction from this one. A third neighbour is two monitors showing the same image in different colours, which is covered in the guide numbered 201 — that is a difference between two displays, not a difference between a preview and a finished file. A fourth neighbour is a high dynamic range screenshot or recording that turns grey, covered in the guide numbered 228 — that is a capture and viewing problem, not an editing export. Each of those guides has its own shape, and this guide starts from a different one: preview right, export finished, colour wrong, picture intact.

The first conclusion most editors reach is that the grade itself must be wrong — that the colour correction was set badly, and the export is simply showing the truth. That conclusion jumps too far. The preview and the export do not travel the same road, and a correct preview does not guarantee a correct export, just as a wrong export does not prove the grade was wrong. The preview can use proxy files, cached frames, or a lower-resolution render to keep playback smooth, and the preview monitor can present the image under the software's own display settings rather than under the file's own colour tags. The export has to render the timeline again at full resolution, hand the result to an encoder, and write colour tags into the file. A fault in any of those later stages changes the colour of the finished file while leaving the preview untouched. The rest of this guide is about testing which of those stages the fault belongs to, instead of going back and re-grading a grade that may never have been wrong.

One practical point has to be settled before anything else. The direction of the colour error matters, and it should be written down before any setting is changed. Is the finished file paler than the preview, or greyer, or warmer, or colder, or darker, or lighter? Different directions point at different layers: a picture that is washed out overall leans toward a dynamic range mismatch, while a picture that has shifted warmer or colder leans toward a colour temperature or gamma reading difference. A direction remembered loosely is no evidence at all, so the comparison should be made side by side — the preview on one screen and the finished file on the same screen, or the same frame frozen in both — and the direction should be written down in plain words before the first test is run.

How a Preview and an Export Take Different Roads

To understand why a preview can be right while an export is wrong, it helps to know what the editing software is doing in each case. The preview and the export are two separate paths through the same software, and they do not share the same steps.

During preview, the software is trying to play back in real time. To keep playback smooth, it is allowed to take shortcuts. It may play proxy files — lower-resolution copies of the original footage — instead of the original files. It may show frames it rendered earlier and stored in a cache, rather than rendering each frame again. It may render at a lower resolution and then scale the result up for the preview monitor. None of those shortcuts changes the timeline, and none of them changes the grade, but all of them mean the picture on the preview monitor is not necessarily the picture the export will produce. The preview monitor itself adds another difference: it presents the image under the software's own display settings, which may not follow the colour tags that will be written into the finished file.

During export, the software has to do the work properly. It renders the timeline again at full resolution, frame by frame, applying every effect and every grade setting. It then hands the rendered frames to an encoder, which compresses them into a file. It also has to write colour space tags into that file — metadata that tells a player how the numbers in the file should be read. A player opening the file reads those tags and then shows the picture according to its own reading method. So the finished file passes through three separate stages after the timeline: full-resolution rendering, encoding, and tagging, and then a fourth stage outside the software altogether, which is the player reading the tags. A preview that looks right only proves that the first road — the immediate playback road — works. It proves nothing about the tagging road or the playback reading road.

The colour space tags are the heart of this guide, and they need to be understood clearly. A colour space tag is metadata inside the file. It tells a player what the numbers in the file mean: which colour primaries the file uses, which transfer characteristic it follows, and which matrix coefficients it was encoded with, together with whether the data range is Full or Limited. The numbers themselves — the brightness and colour values stored in the file — can be exactly right, and the picture can still look wrong, because the tag or the reading is wrong. That is the counter-intuitive core of the whole problem: the values in the file can be completely correct while the picture is still wrong, because the fault is in the label and the reading, not in the values themselves. Raising the saturation or the brightness to compensate only makes the values wrong, so the file then looks even more wrong on a system that reads the tags correctly. Compensation is not a repair; it writes the error into the file itself.

There is a well-known special case on Mac computers that has to be stated plainly. A Mac displays Rec.709 media using a gamma of 1.96, while the rest of the industry uses 2.4. The same file, viewed in QuickTime on a Mac and on a broadcast-standard system, will look different, and that difference is not a fault in the file and not a fault in the editing software. It is a difference in the gamma standard of the display system. This point has to be kept in mind throughout the guide, because otherwise an editor will conclude the file is damaged when the file is behaving exactly as the system it is viewed on dictates.

The timeline settings add a further layer. A sequence has its own colour space setting, and many editing programs also have automatic detection for logarithmic footage and automatic tone mapping for high dynamic range material. If the sequence is set to one colour space while the footage was recorded in another, or if automatic detection is switched on for footage that does not need it, the timeline can apply a conversion that the editor never asked for — or fail to apply a conversion that was needed. The export preset matters too: an export preset set to a high dynamic range colour space will tag the file differently from a Rec.709 preset, and the player will then read the file differently. None of these settings shows up in the preview as an obvious error, because the preview monitor can hide the difference under its own display settings.

The source footage is the last piece of the mechanism. Footage can be recorded in many colour spaces: standard Rec.709, logarithmic formats that hold more dynamic range, hybrid log-gamma, or perceptual quantiser formats for high dynamic range. Each of those needs to be converted correctly into the timeline colour space. If one clip is logarithmic and the timeline treats it as Rec.709, that clip will look wrong while the other clips look right. If every clip looks wrong in the same way, the fault is more likely in the timeline or the tags than in any one clip. That distinction — some clips wrong against all clips wrong — is one of the most useful observations in the whole guide, and it is why the steps below check the footage clip by clip.

The Deciding Forks: Three Questions That Narrow the Field

The whole guide turns on three forks. Each fork is a comparison, and each comparison removes a whole set of possibilities. They should be done in order, before any setting is changed, because a setting changed before the fork is tested destroys the evidence the fork would have given.

Fork One: Does the Colour Follow the Player?

The first fork is the player comparison, and it is the cheapest test in the guide. Take the same finished file and open it in two or more players: inside the editing software by importing it back onto a timeline, in VLC, in the operating system's built-in player, and on a phone. If the colour changes from player to player — right in one player and wrong in another — the fault follows the player, and it belongs in the playback reading layer. The file itself may be fine, and the players are simply reading its tags differently. If every player shows the same wrong colour, the fault does not follow the player, and it belongs in the export labelling layer or further back. This fork decides whether the problem is in the file at all, and that is why it comes first. No tag should be changed and no grade should be re-done before this fork has been run, because changing the file before knowing whether the file is at fault destroys the comparison.

Fork Two: Do the File Tags Match the Timeline?

The second fork is the file-tag comparison. Use MediaInfo to read the finished file's colour tags — Colour primaries, Transfer characteristics, and Matrix coefficients — together with the data range, Full or Limited. Then compare those tags against the timeline settings: the sequence colour space, the output colour space, and the export preset. If the tags do not match the timeline, the fault belongs in the labelling layer: the file was written with tags that do not describe what the timeline produced. If the tags match the timeline and the colour is still wrong, the investigation moves on to the preview path layer, because the tags are not the source of the difference. MediaInfo is a free tool that reads the metadata inside a media file without changing it, and it is the right tool for this fork because it shows what was actually written into the file, rather than what the export settings were supposed to write.

Fork Three: Which Direction Is the Colour Error?

The third fork is the direction of the error, recorded in the opening section. A finished file that is paler or greyer overall than the preview leans toward a dynamic range mismatch — the data range, Full against Limited, or a tone mapping step that was applied or missed. A finished file that has shifted warmer or colder leans toward a colour temperature or gamma reading difference — the player is reading the transfer characteristic differently from the way the editing software displayed it. A finished file that is darker overall on one system and right on another leans toward the Mac gamma difference described above. The direction does not prove a layer on its own, but it narrows the search, and it should be written down alongside the player comparison result, because a direction that changes from player to player is itself evidence for the playback layer, while a direction that stays the same everywhere is evidence for the labelling layer.

ForkComparisonResult AResult B
Player comparisonSame finished file in two or more playersColour follows the player — playback reading layerSame wrong colour in every player — export labelling layer
File-tag comparisonMediaInfo tags against timeline settingsTags do not match the timeline — labelling layerTags match but colour is still wrong — preview path layer
Direction of the errorFinished file against preview, side by sideWashed out overall — dynamic range mismatchWarmer or colder shift — gamma or colour temperature reading

The Five Suspicion Layers

A preview that is right and an export that is wrong in colour can fail at five layers, and they must be kept separate. Changing the timeline setting, the export preset, and the player all at once and finding the colour right proves nothing about which change did the work, and the error will return the next time only one of those conditions is present. The layers below run from the player back to the source footage, and each one has its own deciding evidence. The order itself is the diagnosis: the player comparison is cheapest and decides whether the file is at fault at all, so it comes first; the tag comparison is the second fork and decides whether to work on the labels or on the preview path; the footage layer comes last, because it needs a clip-by-clip check.

LayerWhat it coversDeciding evidence
1. Playback reading layerThe same finished file read differently by different playersPlayer comparison: the colour follows the player
2. Export labelling layerThe colour space tags and data range written into the file do not match the timelineMediaInfo comparison: tags and Full or Limited range do not match the timeline settings
3. Preview path layerThe preview used proxies, cached frames, or a lower-resolution render, or the preview monitor settings differPreview comparison: with proxies off and cache cleared, the preview matches the finished file
4. Timeline settings layerThe sequence colour space, automatic log detection, and automatic tone mapping do not fit the footageSettings comparison: the sequence is not Rec.709, or automatic detection and tone mapping are switched on for footage that does not need them
5. Source footage layerThe footage itself was recorded in different colour spaces — logarithmic, hybrid log-gamma, perceptual quantiser, or Full range — and the timeline did not convert it correctlyFootage comparison: only some clips are wrong, and the wrong clips share the same source colour space

Layer One: The Playback Reading Layer

The first layer is the player reading the file. Different players read colour tags differently, and different operating systems apply different display profiles. A file that looks right in VLC can look washed out in the system player, and a file that looks right in the editing software can look different again on a phone. The signature of this layer is the player comparison: the colour follows the player. That result means the file itself may be sound, and the difference sits in how each player interprets the tags. This layer is also where the Mac gamma difference lives. A Mac displays Rec.709 media at a gamma of 1.96, while the rest of the industry uses 2.4, so the same file will look different on a Mac in QuickTime and on a broadcast-standard system. Neither view is wrong in its own system; the file has not changed, and the display standard has. The temptation at this layer is to change the file so that it looks right in one particular player. That temptation should be resisted: a file changed to suit one player will look wrong on a system that reads the tags correctly, and the change writes the error into the file. The correct response to a playback reading fault is to identify which player is reading differently and why, not to re-grade the timeline.

Layer Two: The Export Labelling Layer

The second layer is the tags written into the file at export. The export writes colour primaries, transfer characteristics, and matrix coefficients into the file, together with the data range. If those tags do not match what the timeline actually produced — for example, the timeline is Rec.709 but the file is tagged as something else, or the timeline is Limited range and the file is tagged Full — then every player that reads the tags will show the wrong colour, and the error will be the same in every player. That is the signature of this layer: the same wrong colour everywhere, with tags that do not match the timeline. The deciding evidence is the MediaInfo comparison in fork two. The tags are metadata, and metadata can be wrong while the picture data is right, which is why this layer is so easy to misread as a grading fault. The fix, when this layer is confirmed, is to make the export write the tags that describe the timeline honestly — not to write the tags that make one particular player look right, because tags written to suit one player will mislead every other player. The export preset deserves close attention here: a preset that carries a high dynamic range colour space will tag the file differently from a Rec.709 preset, even when the timeline itself is Rec.709, and that mismatch is enough to shift the colour of the whole file.

Layer Three: The Preview Path Layer

The third layer is the preview path itself. The preview can use proxy files, cached frames, or a lower-resolution render, and the preview monitor can present the image under the software's own display settings. Any of those can make the preview look different from the finished file, even when the timeline, the tags, and the players are all correct. The signature of this layer is the preview comparison: turn the proxies off, clear the render cache, preview at full resolution, and compare again. If the preview now matches the finished file, the fault was in the preview path — the preview was showing a shortcut version of the picture, and the export was showing the real one. If the preview still differs from the finished file after those steps, the preview path is cleared, and the investigation moves on. This layer is often overlooked because the preview feels like the truth, but the preview is a convenience for smooth playback, not a reference for the finished file. A preview that was built from cached frames can also be out of date: a grade changed after the cache was built will not show in the cached preview until the cache is cleared, so the preview and the export can disagree simply because the preview is showing an older version of the grade.

Layer Four: The Timeline Settings Layer

The fourth layer is the timeline settings. The sequence colour space, the automatic detection of logarithmic footage, and automatic tone mapping all decide how the footage is converted into the timeline. If the sequence is set to Rec.709 but the footage is logarithmic and automatic detection is switched off, the logarithmic footage will not be converted, and it will look flat and grey. If automatic tone mapping is switched on for footage that does not need it, the software can apply a conversion that was never asked for, and the picture will shift. The signature of this layer is the settings comparison: check whether the sequence colour space is Rec.709, whether automatic log detection is switched on, whether automatic tone mapping is switched on, and whether the export preset is Rec.709 rather than a high dynamic range format. A change here must be followed by a fresh export of a short clip and a fresh comparison — never by a claim from memory that the setting was changed. One caution from the published sources applies here with particular force: the reference manual for DaVinci Resolve warns against changing the timeline colour space in order to repair an export, because the timeline colour space describes what the timeline is, not what the export should pretend it is. Changing it to chase a player is the same error as changing the tags to chase a player.

Layer Five: The Source Footage Layer

The fifth layer is the source footage itself. Footage can arrive in many colour spaces, and a timeline that mixes them has to convert each one correctly. Logarithmic footage, hybrid log-gamma footage, perceptual quantiser footage, and Full range recordings each need their own conversion into the timeline colour space. If the conversion is missing or wrong for one kind of footage, only the clips recorded that way will look wrong, while the other clips look right. That pattern — some clips wrong, others right — is the signature of this layer, and it is found by checking the footage clip by clip, noting the original colour space of each clip and whether the wrong clips share one source format. If every clip is wrong in the same way, the fault is more likely in the timeline or the labelling layer than in any one clip, and the investigation should not spend its time on the footage. This layer comes last because the clip-by-clip check is the most labour-intensive step, and because a footage fault should only be concluded after the cheaper comparisons have cleared the player, the tags, the preview, and the timeline.

The Seven Diagnostic Steps

The steps below are comparison tests. Each one changes one variable and records the result. Changing two things at once and finding the colour right proves nothing about which change mattered, and the error will return the next time only one of those conditions is present. Write down each result before moving on.

Step 1: Record the Baseline

Before anything is changed, record the baseline. Where the preview was viewed, and where the finished file was viewed. What the colour error is, in plain words — paler, greyer, warmer, colder, darker, or lighter — and in which direction, measured side by side on the same screen if possible. The editing software and its version, the sequence colour space, the export settings including the format, the codec, and the data range, and the preview settings including whether proxies are switched on and what the monitor settings are. Without this record, later comparisons have nothing to compare against, and a colour error that changes form halfway through the investigation will look like a new problem when it is the same one wearing a different shape. Also note anything else that changed at the same time — a software update, a new monitor, a new export preset — because a second change at the same moment weakens every later conclusion.

Step 2: The Player Comparison

Take the same finished file and open it in two or more players: inside the editing software by importing it back onto a timeline, in VLC, in the operating system's built-in player, and on a phone. Write down what each player shows, using the same words for the direction of the error each time. If the colour follows the player, the fault belongs in the playback reading layer, and the investigation stops changing the file and starts identifying which player reads differently. If every player shows the same wrong colour, the fault belongs in the export labelling layer, and the investigation moves to step three. This step is a comparison test, not a repair: finding a player that shows the colour right does not fix the file, it only proves the file can be read correctly somewhere, and the file still has to be usable in the players it was made for.

Step 3: The File-Tag Comparison

Use MediaInfo to read the finished file's colour tags — Colour primaries, Transfer characteristics, and Matrix coefficients — together with the data range, Full or Limited. Then compare those tags against the timeline settings recorded in step one. If the tags do not match the timeline, the fault belongs in the labelling layer: the file was written with tags that do not describe what the timeline produced. If the tags match the timeline and the colour is still wrong, move on to step four. MediaInfo reads the metadata without changing the file, so this step is safe to repeat as often as needed, and the tags should be written down exactly as MediaInfo reports them, because a tag remembered loosely is no evidence at all. Pay particular attention to the data range: Rec.709 YUV material at 8 or 10 bit should be encoded as 16 to 235, not Full, and Full range applies to 12 or 16 bit RGB material rather than to ordinary camera recordings. A file tagged Full when it should be Limited, or Limited when it should be Full, will look washed out or crushed across every player that reads the tag.

Step 4: The Preview Path Comparison

Turn the proxies off, clear the render cache, and preview at full resolution, then compare the preview against the finished file again. If the preview now matches the finished file, the fault was in the preview path — the preview had been showing a proxy, a cached frame, or a lower-resolution render. If the preview still differs from the finished file, the preview path is cleared, and the investigation moves on. Do not change any timeline or export setting in this step: the only variables changed here are the preview settings, so that the result can be read cleanly. If the cache is cleared and the preview changes, write down that the preview was out of date, because that fact matters for the next time the same shape appears — a stale cache is a recurring cause, not a one-off event.

Step 5: The Timeline Settings Comparison

Check the sequence colour space, and confirm it is Rec.709 unless the project is genuinely a high dynamic range project. Check whether automatic detection of logarithmic footage is switched on, and whether automatic tone mapping is switched on, and consider whether each one fits the footage in the timeline. Check that the export preset is Rec.709 and not a high dynamic range format. Change one setting at a time, then export a short clip and compare it against the preview, writing down the result before changing anything else. Do not change the timeline colour space in order to make one player look right: the published reference manual warns against exactly that, and a timeline setting changed to chase a player will mislead every other system that reads the file honestly. If a short clip exported after a single settings change matches the preview, the timeline layer is indicated, and the full export can be repeated with that setting. If no single settings change moves the colour, move on to step six.

Step 6: The Footage Clip-by-Clip Comparison

Go through the timeline clip by clip, noting the original colour space of each clip and whether that clip is wrong or right in the finished file. If only some clips are wrong, and the wrong clips share the same source colour space — for example, only the logarithmic clips — the fault belongs in the source footage layer, and the conversion for that kind of footage needs attention. If every clip is wrong in the same way, the fault belongs in the timeline or the labelling layer, and the footage check has cleared the source layer. This step is the most labour-intensive in the guide, and it should not be started before the cheaper comparisons have been run, because a footage fault concluded too early will send the investigation into re-converting clips that were never the problem. Write down the clip, its source colour space, and whether it is wrong or right, because that table is the evidence for this layer.

Step 7: Decide and Prepare for Inspection

Once a layer has been found, write down the comparison results that point to it, and do not change several settings at once. A single settings change, followed by a short test export, is the only way to confirm that the layer found is the layer that matters. Then assemble the conclusion in three sentences: what the symptom is, which variable made the symptom appear or disappear, and which possibilities are still not ruled out. Write it down with the full record: the direction of the colour error, the player comparison result, the MediaInfo tags, the preview path result, the timeline settings, and the clip-by-clip result. If all seven steps have been run and the layers still cannot be separated, or if the colour error is accompanied by a broken picture, coloured blocks, a flicker, or black frames, stop and bring the record to inspection — a colour error with a broken picture is a different shape, and it belongs with the guide numbered 493 rather than this one.

Six Misconceptions That Mislead This Diagnosis

Misconception 1: A Correct Preview Means the Export Must Be Correct

A correct preview only proves that the immediate playback road works. The preview can use proxies, cached frames, or a lower-resolution render to keep playback smooth, and the preview monitor can present the image under the software's own display settings. The export has to render at full resolution, hand the result to an encoder, and write colour tags into the file. Those are different roads, and a fault on the export road leaves the preview untouched. The preview and the export must be tested separately, and neither can stand in for the other.

Misconception 2: A Wrong Export Means the Grade Was Wrong

The values in the file can be completely correct while the picture is still wrong, because the fault can sit in the tag or in the reading rather than in the values. A grade that looks right in the preview and wrong in the finished file has not necessarily been set badly. Until the player comparison has been run, going back to re-grade is guessing, and re-grading a correct grade only makes the file wrong in a new way. Test the layers first, and touch the grade only when the layers have cleared it as a suspect — and even then, change one thing at a time.

Misconception 3: Raising Saturation or Brightness Will Compensate

Raising saturation or brightness changes the values in the file. If the fault is in the tag or the reading, the values were right, and changing them makes the file wrong on every system that reads the tags correctly. Compensation writes the error into the file itself, so that a file which was only misread in one player becomes genuinely wrong everywhere. The correct response to a colour shift is to find which layer shifted it, not to push the values in the opposite direction and hope the two errors cancel out. They will not cancel out on a different player.

Misconception 4: Switching the Renderer or Changing the Codec Counts as a Repair

Switching to software-only rendering, or changing to a different codec, is a comparison test, not a repair. It can show whether the fault follows the renderer or the codec, and that is valuable evidence, but it does not repair anything. Using software-only rendering permanently gives up graphics acceleration, and the neighbouring guide numbered 488 has already stated that position plainly. A comparison step that changes the result is evidence about where the fault lives, and it should be written down as a comparison result, not reported as a fix.

Misconception 5: A File That Looks Wrong on a Mac Must Be Wrong

A Mac displays Rec.709 media at a gamma of 1.96, while the rest of the industry uses 2.4. The same file will look different on a Mac in QuickTime and on a broadcast-standard system, and that difference is a property of the display system, not a fault in the file. Concluding that the file is damaged because it looks wrong on one system skips the player comparison entirely. The file should be judged by the player comparison and the tag comparison, not by how it looks on a single system with its own gamma standard.

Misconception 6: Changing the Tags Will Make Every Player Right

Tags can only be written to describe what the file genuinely is. They cannot be written to describe what the editor wishes a particular player would see. Changing a tag to suit one player will mislead every other player that reads the tag honestly, and the file will then look wrong on systems that were reading it correctly. The Rec.709-A tag is a case in point: it is a Mac and QuickTime scoped practice, effective inside the Mac colour management system, and it is not a universal repair. A tag that makes a file look right in one place and wrong everywhere else has not fixed the file; it has moved the error.

Comparison Table

The table below puts the comparison results side by side. Each comparison changes one variable at a time; a different result points in a different direction.

ComparisonResult AResult BWhat to record
Player comparisonColour follows the playerSame wrong colour in every playerWhat each player shows, in the same words
MediaInfo tag comparisonTags do not match the timelineTags match but colour is still wrongColour primaries, transfer characteristics, matrix coefficients, Full or Limited range
Preview path comparisonProxies off and cache cleared, preview matches the finished filePreview still differs from the finished fileProxy setting, cache state, preview resolution
Timeline settings comparisonOne settings change makes a short test export matchNo single settings change moves the colourSequence colour space, automatic detection, automatic tone mapping, export preset
Footage clip comparisonOnly clips from one source colour space are wrongEvery clip is wrong in the same wayEach clip, its source colour space, wrong or right

Each comparison has to be run separately and recorded separately. Saying from memory that the export looked a little better after a change has no diagnostic value — run the comparison again with the same frame and write both results down before comparing them.

What the Sources Disagree About

The published sources do not agree with each other, and both sides are given below.

First, the Mac gamma of 1.96 against the industry gamma of 2.4. No single setting can make the same file look right on every system. Changing the settings to suit Mac QuickTime — using the 1.96 gamma or the Rec.709-A tag — will make the file look darker on a broadcast-standard system. Leaving the settings alone will leave the file looking wrong on a Mac. This is a trade-off, not a repair, and this guide follows the position that the trade-off should be stated plainly rather than hidden behind a setting presented as a universal standard.

Second, the Adobe output look-up table against the DaVinci Resolve Rec.709-A tag. Both are ways of making a file look right on a Mac, and both are effective only inside the Mac colour management system. On other systems they will make the file look darker. Neither can be written as a general repair, and this guide follows the position that both should be described with their scope stated — Mac and QuickTime scoped — rather than presented as a setting every editor should use.

Third, Full against Limited range. Some tutorials say switching to Full range holds more data, but Rec.709 YUV material at 8 or 10 bit should be encoded as 16 to 235, and Full range does not make the dynamic range wider — it only changes how the range is encoded. Full range applies to 12 or 16 bit RGB material, not to ordinary camera recordings. This guide follows the position that Full range should not be written as advanced or correct, and that the range should match what the timeline and the footage genuinely are.

Fourth, there is a disagreement about how to describe the comparison steps. Third-party tutorials often describe switching the renderer, changing the codec, or changing the tags as repairs. This site's position is that all of them are comparison tests. They produce evidence about which layer the fault belongs to, and they are not repairs. This guide follows that position throughout: no comparison step is presented as a fix.

Two further points should be stated plainly. The reference manual for DaVinci Resolve warns against changing the timeline colour space in order to repair an export, and this guide follows that warning. And the preview monitor settings — the options that match the preview to QuickTime or to the Mac display profile — change only the preview, not the file. A preview setting that makes the preview look right does not change what is written into the finished file, and it should never be reported as an export repair.

When to Stop Testing

Colour-difference testing carries little physical risk, but it can still waste a project if it is done in the wrong order. Do not re-grade the timeline before the player comparison has been run: re-grading a correct grade writes a new error into the project, and the original grade may then be unrecoverable. Do not change several settings at once. A test that has already given a clear result does not need to be run again, and a settings change that makes one player look right while making another look wrong is a result in itself — record it and move to the next comparison instead of chasing that player further.

Stop immediately if the colour error is accompanied by a broken picture, coloured blocks, a flicker, or black frames. That is a different shape, and it belongs with the guide numbered 493. Stop also if the export itself starts failing — errors, stops halfway, or never finishing — because that shape belongs with the guide numbered 476. If the preview itself is damaged, with a green or purple screen, while the finished file is normal, that is the mirror shape covered in the guide numbered 488. If the problem appears only in high dynamic range mode, that is a tone mapping layer covered in the guides numbered 139 and 228. If the only difference is between two monitors showing the same image, that is the guide numbered 201. And if the only problem is banding in smooth gradients, that is a banding layer covered in the guide numbered 159. Each of those shapes has its own guide, and this guide's seven steps should not be used to solve them.

Do not delete the original finished file before a corrected export has been compared against it. The original file is the evidence for the comparison, and deleting it removes the baseline. Keep the original, the short test exports, and the written record together until the layer has been found and the corrected export has been checked in two or more players.

What to Bring for Inspection

If the comparisons point nowhere conclusive, or if they point toward a layer that cannot be separated at home, the next step is a professional inspection. Bring the full record: the direction of the colour error, the player comparison result with what each player showed, the MediaInfo tags exactly as reported, the preview path result, the timeline settings, the export preset, and the clip-by-clip result with each clip's source colour space. The three-sentence conclusion from step seven is the summary of that record. Bring the original finished file as well, on the same storage it was exported to, together with a short test export if one was made, so the comparison can be repeated at the workshop without relying on memory.

Inspection and repair pricing follows the site's published three-tier scale of HK$400 to HK$2,800, with the actual quote given after inspection. No price in this guide should be read as a quote for a particular fault, and no fee is stated here for any specific repair.

The workshop is at Room 1209, Dali Plaza, Shek Mun. The telephone number is 2186 8010, and the WhatsApp number is 6558 6806. There is no Mong Kok shop, and there is no on-site service — inspection is carried out at the workshop.

Frequently Asked Questions

My preview looks right but the exported file looks washed out. Is my grade wrong?

Not necessarily. The values in the file can be completely correct while the picture is still wrong, because the fault can sit in the tag or in the reading rather than in the values. Run the player comparison first: open the same finished file in two or more players. If the colour follows the player, the fault is in the playback reading layer. If every player shows the same washed-out colour, move on to the MediaInfo tag comparison. Do not re-grade before those two comparisons have been run.

The exported file looks different in every player. What does that mean?

It means the fault is probably not in any single player. When every player shows the same wrong colour, the investigation points at the export labelling layer or further back — the tags written into the file, the timeline settings, or the source footage. Use MediaInfo to read the colour primaries, transfer characteristics, matrix coefficients, and data range, and compare them against the timeline settings. Tags that do not match the timeline point at the labelling layer.

The exported file looks right in VLC but wrong in the system player. Is the file damaged?

Not necessarily. Different players read colour tags differently, and a file that looks right in one player and wrong in another is pointing at the playback reading layer, not at file damage. The file may be sound, and the players are simply interpreting its tags in different ways. Write down what each player shows, and do not change the file to suit one player — a change made for one player will make the file look wrong on a system that reads the tags correctly.

The exported file looks wrong on a Mac but right elsewhere. Is the file at fault?

Not necessarily. A Mac displays Rec.709 media at a gamma of 1.96, while the rest of the industry uses 2.4, so the same file will look different on a Mac in QuickTime and on a broadcast-standard system. That difference is a property of the display system, not a fault in the file. Judge the file by the player comparison and the tag comparison, not by how it looks on a single system with its own gamma standard.

Should I raise the saturation or brightness to compensate for the colour shift?

No. Raising saturation or brightness changes the values in the file, and if the fault is in the tag or the reading, the values were right. Changing them makes the file wrong on every system that reads the tags correctly. Compensation writes the error into the file itself. Find which layer shifted the colour first, and change the values only when the layers have shown that the values themselves are the problem.

What is MediaInfo, and what should I look for in it?

MediaInfo is a free tool that reads the metadata inside a media file without changing it. For this problem, look for Colour primaries, Transfer characteristics, and Matrix coefficients, together with the data range, Full or Limited. Compare those tags against the timeline settings: the sequence colour space, the output colour space, and the export preset. Tags that do not match the timeline point at the labelling layer. Write the tags down exactly as MediaInfo reports them.

Should I switch to Full range to get more dynamic range?

No. Full range does not make the dynamic range wider; it only changes how the range is encoded. Rec.709 YUV material at 8 or 10 bit should be encoded as 16 to 235, and Full range applies to 12 or 16 bit RGB material rather than to ordinary camera recordings. A file tagged Full when it should be Limited, or Limited when it should be Full, will look washed out or crushed across every player that reads the tag. The range should match what the timeline and the footage genuinely are.

Only some clips look wrong in the finished file. What does that mean?

It means the fault may sit in the source footage layer. Go through the timeline clip by clip, noting the original colour space of each clip and whether that clip is wrong or right. If only the clips from one source colour space are wrong — for example, only the logarithmic clips — the conversion for that kind of footage needs attention. If every clip is wrong in the same way, the fault is more likely in the timeline or the labelling layer than in any one clip.

Will switching to software-only rendering fix the colour difference?

No, not as a repair. Switching the renderer is a comparison test: it can show whether the fault follows the renderer, and that is valuable evidence. Using software-only rendering permanently gives up graphics acceleration, and it does not repair the tags, the timeline settings, or the player reading. Record the result of the comparison and continue the investigation from the layer it points to, rather than leaving the renderer switched as if the problem were solved.

What should I record before bringing the project in for inspection?

Record the direction of the colour error, the player comparison result with what each player showed, the MediaInfo tags exactly as reported, the preview path result, the timeline settings and export preset, and the clip-by-clip result with each clip's source colour space. Bring the original finished file as well, together with any short test exports. Then put it together as three sentences: what the symptom is, which variable made the symptom appear or disappear, and which possibilities are still not ruled out.

Sources

Related articles

Related services

If you need Chip-level GPU repair, see these service pages: