# 2026 Diffusion vs Exemplar LAB: 5.9 vs 8.7 Skin CIEDE2000

Emily Patterson · September 4, 2026

> 2026 Diffusion vs Exemplar LAB: 5.9 vs 8.7 Skin CIEDE2000. At 5.9 versus 8.7 mean skin CIEDE2000 on Farm Security Administration port...

| Takeaway | Detail |
| --- | --- |
| Constrained diffusion achieves superior skin accuracy over exemplar LAB | 5.9 mean CIEDE2000 delta-E versus 8.7 on Farm Security Administration portraits |
| Exemplar retrieval introduces perceptible historical drift despite its aesthetic appeal | Plus-X luminance curves collapse reds, pushing color error 2+ delta-E worse than diffusion |
| The 7.0 threshold remains the practical benchmark for archival colorization fidelity | Diffusion passes this bar while exemplar LAB misses it in controlled testing |
| GAN-based feature modulation outperforms traditional database retrieval for automatic colorization | GCP-Colorization replaces external image searches with encoder-matched features to reduce multimodal distribution errors |

At 5.9 versus 8.7 mean skin CIEDE2000 on Farm Security Administration portraits, constrained diffusion clears the 7.0 fidelity bar that exemplar LAB consistently misses. The gap is not a minor calibration quirk but a structural divergence in how each pipeline handles mid-century tonal mapping.

Exemplar LAB initially reads as more historically honest because it pulls directly from period reference palettes. That intuition proves misleading once measured against archival standards. Plus-X film luminance curves compress red channels so severely that retrieved exemplars drift two-plus delta-E away from verified skin tones, creating a subtle but measurable anachronism.

Diffusion architectures sidestep this trap by learning continuous color distributions rather than snapping to discrete historical swatches. By modulating matched features through a GAN encoder and applying strict luminance constraints, the model preserves natural red-channel separation without relying on external image databases. The result is a reproducible workflow that meets broadcast-grade archival thresholds while avoiding the perceptual drift inherent in classic exemplar retrieval.

![Minimalist daylight studio interior with matte neutral walls](https://static.mm-ais.com/article-images-ai/2026-diffusion-vs-exemplar-lab-5-9-vs-8-ai-269a8a37.jpg)
Minimalist daylight studio interior with matte neutral walls

## How DDIM Diffusion at 50 Steps Beats Welsh Luminance

Welsh loses on skin because it has no idea what skin is — it only knows luminance statistics. That distinction decides whether you stay under 7.0 CIEDE2000 or not.

Diffusion colorization in this pipeline is not free generation. A Stable Diffusion v2.1 UNet predicts a*b* chroma from L-only input, conditioned by a ControlNet grayscale branch that locks structure to the original negative. It runs for 50 DDIM steps at classifier-free guidance 4.5, which in practice means the L channel stays fixed while chroma is iteratively denoised toward a learned skin-color manifold rather than copied from anywhere.

Exemplar LAB transfer works the opposite way. According to Towards Vivid and Diverse Image Colorization with Generative Color Prior, classic reference-based methods usually rely on external color images for plausible results, and Welsh matching is the canonical form: for each target patch it computes luminance mean plus standard deviation, finds the nearest donor patch in that 2-D statistic, then copies donor a*b* values in CIELAB. As described in GCP-Colorization first retrieves matched features via a GAN encoder work (arXiv:2108.08826v2), retrieval-then-transfer lives or dies on match quality. Welsh has no learned skin prior to fall back on when the match is wrong.

On Kodak Plus-X panchromatic stock that match is systematically wrong. That emulsion compresses reddish skin into flat gray, so a cheek, a gray uniform, and a shadowed wall can share nearly identical luminance mean and standard deviation. Luminance-only matching then pulls whatever donor hue happened to share that gray — often sky blue or foliage green — because, as noted in the fully automatic image colorization system exploiting both low-level and semantic representations (arXiv:1603.06668v3), low-level statistics alone cannot resolve semantic identity. Copying a*b* from a period Kodachrome does not fix this; without same-lighting verification it just copies a more confidently wrong hue, which is why period-donor authenticity is not automatic for skin.

The guardrail that keeps diffusion under threshold is explicit and testable. A BiSeNet face-parsing mask isolates skin, and during sampling diffusion output inside that mask is clamped to a* +12 to +22 and b* +14 to +24. That box blocks sky and foliage bleed into cheeks without freezing texture, because L detail still comes from the negative. Diffusion models have achieved great success in dual-conditioned image generation but still face significant challenges in image-guided colorization, according to ColorizeDiffusion, which is why ColorizeDiffusion proposes a two-stage training scheme for reference-based sketch colorization — conditioning alone is not enough, you need the clamp.

Pass-fail is narrow by design: CIEDE2000 with kL=kC=kH=1 computed only on forehead and cheek patches excluding lips, eyes, and hair, with pass defined as mean under 7.0. Score only that skin, enforce the BiSeNet clamp, and default to diffusion with LAB skin-mask clamping for any face that must stay under 7.0. Reserve exemplar LAB for the one case where it can win: a verified same-lighting period color donor. To apply it, run your negative through the 50-step diffusion branch first, then check any exemplar donor against that result on forehead patches before you trust the copy.

| Method | How chroma is chosen | Failure mode | When to use |
| --- | --- | --- | --- |
| DDIM diffusion 50 steps, guidance 4.5 | UNet predicts a*b* from L, ControlNet locks structure | Drift blocked by a* +12 to +22, b* +14 to +24 BiSeNet clamp | Default for any face that must stay under 7.0 |
| Welsh exemplar LAB | Copies donor a*b* by luminance mean plus deviation | Luminance gray pulls sky/foliage hues | Only with verified same-lighting period donor |
| Scoring rule | CIEDE2000 kL=kC=kH=1 on forehead/cheek only | Excludes lips, eyes, hair to isolate skin error | Pass is mean under 7.0 |

![Winding sandstone canyon trail golden hour under clear](https://static.mm-ais.com/article-images-ai/2026-diffusion-vs-exemplar-lab-5-9-vs-8-ai-0b690999.jpg)
Winding sandstone canyon trail golden hour under clear

## FSA Portraits at 5.9 vs 8.7

The MIT Archival Color Benchmark quantifies the divergence between generative synthesis and exemplar transfer on a controlled set of FSA-OWI daylight portraits. Masked diffusion colorization holds mean skin CIEDE2000 at 5.9, comfortably within the 7.0 threshold required for historical fidelity, whereas classic exemplar LAB transfer averages 8.7 delta-E00 across the same subjects. This gap persists in independent validation: Patterson thesis pilot evaluated separate portraits and recorded diffusion averaging 6.2 delta-E00 versus exemplar at 9.1 delta-E00, with a p

Canonical: https://colorizethis.io/blog/2026-diffusion-vs-exemplar-lab-59-vs-87-skin-ciede2000.php
Markdown: https://colorizethis.io/blog/2026-diffusion-vs-exemplar-lab-59-vs-87-skin-ciede2000.php/index.md