Concept · Video remastering
What Is Inverse Tone Mapping?
Tone mapping squeezes HDR into SDR so standard screens can show it. Inverse tone mapping (ITM) does the reverse: it expands SDR luminance, color and detail back toward HDR. The quality depends entirely on how the expansion is guessed.
Background
Why SDR needs an inverse of tone mapping
HDR mastering encodes up to 10,000 nits of luminance in 10-bit code values; SDR holds about 100 nits in 8-bit. Tone mapping is the lossy squeeze from HDR down to SDR. The original HDR values are gone — inverse tone mapping must re-create them from context. It is a restoration-with-hallucination problem, and honest ITM tools say so.
Three generations of ITM
| Approach | How it works | Failure mode |
|---|---|---|
| Global curve | One expansion curve applied to the whole frame | Lifts everything — washed-out look, clipped highlights |
| Local/region methods | Expansion adapts per region, often using luminance histograms | Halos around bright objects, noise amplification in shadows |
| Deep learning | A model trained on SDR/HDR pairs predicts per-scene expansion | Needs good training data; can invent texture on low-quality sources |
What a good ITM model gets right
A learned model does three things at once, per scene:
1. Brightness — maps the 0.1–100 nit SDR range toward the full HDR range so highlights and shadows both breathe.
2. Color — expands BT.709 toward BT.2020 while preserving hue relationships, so skin tones and brand colors do not shift.
3. Detail — rebuilds fine texture in shadows and highlights that SDR compression flattens, without amplifying noise.
The result is 10-bit output with HDR10 metadata — a genuine HDR master, not a brightened SDR file.
Where ITM hits its limits
ITM cannot recover information that was destroyed before it was captured: clipped highlights stay clipped, severe compression artifacts stay visible, and a 240p VHS rip will not become 4K HDR. The honest framing: ITM is for lifting professionally-shot SDR into HDR range, not for conjuring quality that is not there.
Source quality is the single biggest success factor. Clean 8-bit BT.709 masters convert far better than re-encoded, heavily-compressed copies.
ITM in production
- Per-scene analysis, not one global curve
- 10-bit output with HDR10 metadata
- Async job API for whole libraries
- Postman collection included
FAQ
Common questions
Is inverse tone mapping the same as upscaling?
No. Upscaling adds resolution (more pixels); inverse tone mapping expands dynamic range (brighter highlights, wider color). They are complementary — see our SDR2HDR vs AI upscaling comparison.
Can inverse tone mapping create true HDR?
It creates genuine HDR deliverables (10-bit, BT.2020, HDR10 metadata). It is not native HDR capture — information lost in SDR mastering cannot be perfectly recovered, which is why we are explicit about the limits.
What sources convert best?
Clean, professionally-graded SDR masters. Heavily compressed or severely clipped footage gains less.
Does the API support batch conversion?
Yes — async jobs with per-file webhooks, so you can remaster a whole library without babysitting it.
Keep exploring
Related guides
SDR2HDR API
Convert SDR video to HDR with restored brightness, gamut and detail.
Learn moreHdr10 vs Dolby Vision
HDR10 vs Dolby Vision, explained for video engineers: static vs dynamic metadata, 10-bit vs 12-bit, brightness limi...
Read guideSDR vs HDR
SDR vs HDR explained for video creators: brightness (100 vs 1,000+ nits), color space (BT.709 vs BT.2020), bit dept...
Read guide