Make an SDR Pitch Copy of HDR Footage: Map on Input or Output?
Make an SDR Pitch Copy of HDR Footage: Map on Input or Output?
The problem usually arrives as a symptom. Someone opens the SDR review copy and says the window is gone, or the red jacket has turned brick, or the face has gone flat against the light. The instinct is to fix it where the symptom appears: a curve on the highlights, a saturation bump on the jacket, a LUT applied as an effect because it looks better than nothing.
Sometimes that works. But the choice between mapping on input and mapping on output is not really about how a frame looks when you pause on it. It is about which stage of the pipeline is still allowed to make decisions your corrections depend on. Once you know that, the placement usually chooses itself.
Here is the structural fact the decision rests on. In Premiere's Color Management options — as described in Adobe's documentation for the feature, revision dated 7 January 2026 — tone mapping and gamut compression exist in two positions. Input tone mapping and input gamut compression act on the clip before effects, and can be set per clip. Output tone mapping and output gamut mapping act after effects, applying the selected mapping across the sequence as a whole. One placement puts a fold between the source and your work. The other puts your work between the source and the fold.
For a pitch copy whose story depends on highlight or saturation relationships that a single delivered SDR image has to hold, the usual answer is output mapping. But "usual" is doing real work in that sentence, and the rest of this piece is about when it stops applying.
Declare the source and the review signal
HDR is not one input type. PQ and HLG carry luminance differently. Camera log is not a delivery transfer function at all. An already-graded HDR master has a different problem from an ungraded camera original, because someone has already spent decisions inside that range. Two files that both say "HDR" in a filename can require different algorithms, and treating all of them as one interchangeable source is how a conversion gets judged against the wrong standard.
Premiere's documentation is explicit that the clip's color-space metadata may not be enough on its own, and that interpretation can require manual confirmation. This is worth taking seriously, because a mislabeled input makes everything downstream a lie. If a PQ clip is interpreted as standard range, no mapping occurs, the image looks flat and low-contrast, and the editor reaches for a LUT. The LUT then becomes part of the pipeline, and every conclusion drawn afterward is drawn from a wrong premise.
Before judging either conversion route, record four things: the application version, the working color space, the output space, and the conditions the review copy will actually be watched in. Option structure moves between releases — the absence of a control in your build is not a statement about the feature. Note that a pitch review on a laptop in an office is a different viewing condition from a calibrated room, and both are different from what a festival submission would require.
Then separate the two operations that get collapsed in conversation. Tone mapping is about luminance range: how many stops get folded into SDR, and where the fold is placed. Gamut compression is about color range: how far outside the target gamut the source sits, and how much of that has to be pulled in. A window that flattens into a single white shape is a tone question. A saturated jacket that loses its hue against a warm wall is a gamut question. If you diagnose one and adjust the other, you will spend an afternoon on a control that was never connected to the problem.
If the source interpretation is unknown, that is the task — not a footnote. Ask for the camera report, the delivery spec, or the colorist's notes. A copy whose source is a guess is not color-faithful; it is a guess with a frame rate.
Draw the two pipelines in the order they run
The order is the whole argument, so write it out rather than holding it in your head. Two routes, same stages, different positions.
INPUT MAPPING
source file
→ clip interpretation (declared space; manual override if needed)
→ input tone mapping + input gamut compression (per clip)
→ effects and corrections
→ sequence output (SDR)
→ export
OUTPUT MAPPING
source file
→ clip interpretation
→ effects and corrections (in the working space)
→ sequence-wide output tone mapping + gamut mapping
→ export
The difference is not the number of stages. It is which side of your corrections the fold lands on. In the input route, by the time your curve reaches the image, the image has already been distributed into an SDR range. In the output route, your curve runs while the working space still holds the wider range, and the fold has to accept whatever you made.
That produces two consequences, and they pull in opposite directions.
Reach. In the input route, the highlight range was spent at the top of the pipeline. A correction intended to bring something out of near-white is now re-expanding a range that has already been compressed; the separation between a pale sky and a white railing was decided before you arrived. In the output route, that decision is downstream of your work. You are shaping what gets folded, not arguing with a fold that already happened.
Granularity. The input route's mapping is per clip, which means the HLG insert from a phone and the PQ master from the camera each get the algorithm their source needs. The output route runs one algorithm for the whole sequence. If the shot-to-shot HDR character genuinely varies, that is a real cost. You can compensate shot by shot with corrections upstream — that is precisely what the output route lets you do — but you are now maintaining corrections whose job is to shape what the mapping receives, and anyone reading the timeline later needs to know that.
The per-clip flexibility cuts both ways, too. Two clips mapped by two different algorithms can look like two different shows in the same sequence, and an algorithmic difference between shots is harder to spot than a brightness difference. Flexibility is not free consistency.
This describes one application's documented option structure. It is not a law of color management, and other tools place these stages elsewhere. Where the exact location of these controls matters for your build, confirm it against that version's documentation rather than a memory of a tutorial.
Compare corrections, not only untouched images
Take an invented passage — six seconds, one shot, the kind of material the choice exists for. An already-graded HDR master, PQ in a wide container, with the brightest part of frame around 1000 nits. A person sits with their back to a large window, face in the lower part of the range. The window's upper corner holds two things the story needs: a pale sky and a white railing that must not merge into one white shape. The person wears a saturated red jacket; a warm, less-saturated orange wall sits behind them, and the two must stay distinct. Delivery is a Rec.709 review copy for a laptop pitch, with display conditions unverified.
Apply the same correction to both copies, with the same numbers: pull the window's top end down so the railing separates from the sky, lift the face so it reads as a face rather than a silhouette, and make a modest saturation adjustment on the jacket.
In the input-mapped copy, the window's top end has already been folded to the ceiling of the SDR range before the pull-down reaches it. Pulling it down now means working inside a range that has already been compressed: the distance between railing and sky was set at the fold, and the correction can only lower the flattened result. The railing and the sky may separate by less than the shot needs, and no downstream move restores what the earlier fold collapsed, because the SDR result is all that remains of it. The jacket sits in the same position. Gamut compression has already pulled the saturated red in before the saturation adjustment arrives, so the adjustment is moving colors whose distance from the wall was decided upstream — a trim on a separation that has already been narrowed.
In the output-mapped copy, the pull-down happens while the range is still available, and the fold then distributes the corrected image. The saturation adjustment runs the same way, on a jacket still outside the target gamut, and the compression distributes whatever the corrected value turns out to be. If the railing still merges, or the jacket and the wall still converge, the move is upstream: change the correction so the folded result keeps the separation, and the mapping re-evaluates. The variable you are adjusting is the one the problem actually lives in.
That is the comparison worth making, and it is a paper comparison. No paired export was rendered for this article; what is fixed is the documented stage order and the arithmetic consequence of it. The size of the difference on your footage depends on the algorithm you select and how much range your source actually holds. The ordering is not in question. The magnitude is yours to measure.
There is a counter-case, and it deserves stating plainly. If the sample's corrections are a white-balance match, a midtone lift, and a small saturation trim — nothing that depends on the top of the range, and nothing pushing a color toward the target gamut's edge — then the input route's per-clip control and simpler timeline may win outright. Choosing output mapping because it sounds more rigorous is its own kind of error, and it produces the same waste as the curve applied to the wrong symptom.
Whatever you choose, judge it after the full effect order. The frame you open first is the one the mapping decisions have not been stress-tested on — usually the easiest frame in the piece. Look past it to what the same fold also has to carry: the cut into and out of the shot you are judging, and the darker material next to it. A sequence-wide fold serves all of them with one set of numbers. That is where the constraint gets tested, not in the opening frame.
Find accidental extra transformations
Four places can insert a conversion: clip interpretation, an effect, the sequence's color management, and the export settings. No single one of them is suspicious. Together they are where the second conversion hides.
A common failure goes like this. The clip's metadata does not identify the HDR transfer function cleanly, the application interprets it as standard range, and the image looks flat. The editor does the sensible-looking thing and adds a conversion LUT as an effect. Later, someone corrects the interpretation and switches on the mapping. Now the LUT is operating on material that has already been mapped, against a working-space assumption that is no longer true. The image gets worse. The natural next move — another curve on top — treats the symptom and buries the cause one layer deeper.
Repair the configuration before compensating with a grade. If two mapping stages are both present, that is allowed — an input mapping with a clip-specific reason plus a sequence output mapping to the delivery space is a coherent arrangement. But you should be able to say why each one exists. A stage you cannot justify is a stage you did not choose.
Two things to keep. The untouched source media, always. And a record of the route back: a saved project state or written settings note for each version, enough that the comparison can be rebuilt, and enough that you are not the only person who can find it.
Choose and verify the bounded SDR result
Verify through the declared route, not only inside the editing application. The viewer is the condition the copy was made for. If the reviewer will watch on a laptop at a kitchen table, a calibrated grading display tells you what the material can be, not what they will see. Both are legitimate; they answer different questions, and neither substitutes for the other.
In the export, inspect four things rather than the overall impression. Bright regions: does the railing separate from the sky at the actual delivery size? Saturated regions: does the jacket hold its hue and stay distinct from the wall, or do they converge? Transitions: does the fold behave consistently across the cut and through any fade, where a sequence-wide algorithm has no per-shot comfort? And latitude: if the window needs to come down a stop after the review, does this route still have room, or did it spend everything already?
Then write down why the route serves this sample and which uncertainties remain. Unknown viewing conditions and an uncalibrated display cap the claim. Neither route certifies final mastering, and neither one proves color fidelity — the SDR copy is an interpretation of the HDR source, made under stated conditions, and saying so costs nothing.
For the passage above, the choice lands on output mapping, for one reason: the story-critical relationships live at the top of the range and near the gamut's edge, and output mapping keeps both available while the correction is being made. If the passage had needed only midtone work and the sources had been more varied, the answer would have flipped, and it would have been the right answer for the same kind of reasoning.
What no amount of documentation settles is whether your export holds up. That takes the paired run on your own passage — the same correction, the same numbers, both routes, inspected through the route your reviewer will actually use. Run it before the pitch, not after.
Frequently asked questions
What is the structural difference between input and output mapping?
In Premiere's Color Management options, as described in Adobe documentation revision dated 7 January 2026, input tone mapping and input gamut compression act on the clip before effects and can be set per clip. Output tone mapping and output gamut mapping act after effects and apply across the sequence. The choice determines which side of your corrections the fold lands on.
When is output mapping usually appropriate for an SDR pitch copy?
When the story depends on highlight or saturation relationships that a single delivered SDR image must hold, output mapping keeps the wider range and gamut available while corrections are made. In the invented six-second example, the railing separating from the sky and the red jacket staying distinct from the wall are story-critical, so output mapping is the choice. If the work is only midtone and the sources are varied, the answer can flip.
What should be recorded before judging either conversion route?
Record the application version, working color space, output space, and the conditions the review copy will be watched in. Also establish source type and interpretation: PQ, HLG, camera log, graded versus ungraded, and manual confirmation where metadata is insufficient. Option structure moves between releases, so a missing control in your build is not a statement about the feature.
How do tone mapping and gamut compression differ in diagnosis?
Tone mapping concerns luminance range, how many stops are folded into SDR and where the fold sits. Gamut compression concerns color range, how far outside the target gamut the source sits. A window flattening into one white shape is a tone question; a saturated jacket losing hue against a warm wall is a gamut question. Adjusting the other control wastes time.
What accidental extra transformations can undermine an HDR-to-SDR copy?
Four places can insert a conversion: clip interpretation, an effect, the sequence's color management, and export settings. A common failure is misinterpreting HDR as standard range, adding a conversion LUT as an effect, then later correcting interpretation and switching on mapping, so the LUT operates on already-mapped material. Repair the configuration before compensating with a grade; keep untouched source and a record of the route back.