RawJPEG Adapter: Invertible Raw-to-JPEG Pipeline

• RawJPEG Adapter is a method that transforms demosaiced raw RGB images using invertible operations to preserve linear raw fidelity during JPEG compression.
• It employs channel-wise 1D LUTs, optional global blockwise DCT scaling, and spatially adaptive gamma mapping to mitigate quantization and compression artifacts.
• The approach improves metrics like PSNR and SSIM over direct JPEG storage and maintains backward compatibility with standard JPEG decoders via metadata embedding.

RawJPEG Adapter, published as Raw-JPEG Adapter, denotes a lightweight, learnable, and invertible preprocessing pipeline that adapts raw images for standard JPEG compression while preserving substantially higher raw reconstruction fidelity than direct raw-to-JPEG storage. In its canonical form, the method applies compact invertible transforms to a black-level-normalized demosaiced raw RGB image before standard JPEG encoding, stores the transform parameters inside the JPEG comment field, and inverts those transforms after JPEG decoding. The result is a standard JPEG file that remains widely interoperable, while a raw-aware decoder can reconstruct a raw-domain image with much higher fidelity than ordinary JPEG applied directly to raw data (Afifi et al., 23 Sep 2025).

1. Definition and representational scope

In the literature, RawJPEG Adapter is primarily a storage-and-reconstruction method rather than a new image codec. The central object is a black-level-normalized demosaiced raw RGB image $\mathbf{I}$, and the direct-storage baseline is written as

$$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$

The adapter inserts an invertible transform $F$ with parameters $\theta$: $$\mathbf{\hat{I}} = F^{-1}\left(Dec\left(Enc\left(F\left(\mathbf{I}; \theta\right)\right)\right); \theta\right).$$ This formulation makes the JPEG file itself the transport format, while the learned adapter is an auxiliary reversible map around the codec (Afifi et al., 23 Sep 2025).

A critical representational choice is that the method operates on demosaiced raw RGB rather than the Bayer mosaic directly. The paper states, “Given an input RGB demosaiced raw image $\mathbf{I} \in \mathbb{R}^{H \times W \times 3}$...”, and in experiments Bayer raw images are demosaiced using Menon’s algorithm. This means the method targets linear raw-domain RGB after sensor normalization, not native CFA storage. It therefore preserves raw-domain linearity and editability better than ordinary JPEG, but it is not a drop-in lossless replacement for Bayer-container formats such as DNG (Afifi et al., 23 Sep 2025).

The method is also explicitly not a new codec standard. It does not alter JPEG syntax, entropy coding, or decoder semantics. A conventional viewer can decode the file as a normal image, but it will display the adapted JPEG image rather than the reconstructed raw. A raw-aware decoder reads the metadata stored in the JPEG COM segment and inverts the adapter to recover the raw-domain representation (Afifi et al., 23 Sep 2025).

2. Problem setting: RAW, ISP, and JPEG mismatch

The motivation for RawJPEG Adapter is the structural mismatch between raw sensor data and standard JPEG. Raw data is typically 12–14 bits/channel, linear, and shaped by sensor and illumination statistics. Standard JPEG stores 8-bit image data, uses YCbCr conversion, usually 4:2:0 chroma subsampling, 8×8 DCT, and fixed quantization tables designed for natural sRGB-like images. Direct raw→JPEG therefore causes “banding in dark regions, clipping/highlight distortion, color shifts, and irreversible quantization artifacts,” even at high quality (Afifi et al., 23 Sep 2025).

Related work sharpens the distinction between compression loss and earlier image-signal-processing loss. In underwater imaging, the main scientific problem with JPEG is stated to be “not due to JPEG compression itself, but to the in-camera processing that precedes it,” because proprietary photofinishing destroys the linear relationship between pixel values and scene radiance before compression occurs. This makes JPEG→RAW recovery fundamentally ill-posed for quantitative color recovery, especially underwater (Akkaynak et al., 21 Mar 2026). A plausible implication is that RawJPEG Adapter addresses a different problem: it does not attempt to invert arbitrary consumer JPEGs after unknown ISP, but instead preserves raw-domain information by controlling the transform before JPEG encoding.

A separate but related line of evidence comes from direct RAW vision. “Raw Instinct” reports that RAW-to-RGB conversion “does not add new capture information” for classification, and the total computation time from RAW image data to classification results can be up to 8.46 times faster than RGB on its controlled dataset when conversion is bypassed (Kantas et al., 2024). This suggests that the RAW/JPEG boundary is not merely a storage issue; it is also an interface problem between sensor-domain data and downstream computation.

3. Invertible architecture and transform components

The Raw-JPEG Adapter pre-encoding pipeline consists of three invertible components applied in sequence: channel-wise 1D look-up tables (LuTs), an optional global blockwise DCT coefficient scaling transform, and a pixel-wise gamma map. These parameters are predicted from a thumbnail of the input image by a compact encoder-decoder network, then serialized into the JPEG comment field (Afifi et al., 23 Sep 2025).

For each channel $c \in \{R,G,B\}$, the adapter applies a monotonic 1D LUT with 128 samples: $$\mathbf{I}^{LuT}_{c(x,y)} = LuT_c\!\left(\mathbf{I}_{c(x,y)}\right), \quad c \in \{R,G,B\}.$$ Monotonicity is enforced through cumulative positive increments: $$\mathbf{L}(i) = cumsum\!\left(softplus(h_\theta)\right)_i.$$ This redistributes raw intensities globally so that dark-region values are less vulnerable to 8-bit JPEG quantization (Afifi et al., 23 Sep 2025).

The optional frequency-domain stage is aligned with JPEG’s own block structure. The image is divided into non-overlapping $8\times 8$ blocks; for each block $$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$0, the adapter computes a 2D DCT, multiplies by a single global learned 8×8 scaling matrix $$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$1, and reconstructs: $$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$2 The scaling matrix is constrained as

$$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$3

which restricts values to approximately $$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$4. This stage captures camera-specific frequency biases and better aligns raw statistics with JPEG’s fixed quantization tables, although its generalization across unseen cameras is weaker than the spatial-only variant (Afifi et al., 23 Sep 2025).

The final adaptive stage is a spatially varying gamma field. The network predicts a 100×100 gamma map $$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$5, upsamples it bilinearly to image resolution, and applies

$$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$6

Gamma values are constrained by

$$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$7

mapping to approximately $$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$8. The inverse pipeline performs inverse gamma, inverse DCT scaling if present, and inverse LUT application. Decode-time inversion is lightweight and deterministic; the decoder does not run a neural network (Afifi et al., 23 Sep 2025).

Per image, the stored metadata consists of the 100×100 gamma map, three 128-entry 1D LUTs, and optionally one global 8×8 DCT scaling matrix. These parameters are compressed with zlib, Base64-encoded, inserted into the JPEG COM segment, and kept under 64 KB, typically around 40 KB. The predictive network itself has about 37K parameters (Afifi et al., 23 Sep 2025).

4. Training objective and codec coupling

Training is self-supervised: the target is the original demosaiced raw image itself. The pipeline predicts $$\mathbf{I}' = Dec\left(Enc\left(\mathbf{I}\right)\right) \approx \mathbf{I}.$$9, applies the forward adapter, passes the result through a differentiable JPEG simulator, inverts the adapter, and compares the reconstruction $F$0 to $F$1. The loss is

$F$2

with $F$3, $F$4, and $F$5 (Afifi et al., 23 Sep 2025).

The differentiable JPEG simulator models the standard JPEG path: RGB to YCbCr, 4:2:0 chroma subsampling, blockwise 8×8 DCT, quantization using luminance/chroma tables scaled by quality factor $F$6, and inverse steps back to RGB. The quality scaling is

$F$7

Coefficient quantization is modeled as

$F$8

and rounding is approximated by a truncated Fourier series: $F$9 An ablation replacing this JPEG simulator with simple RGB 8-bit uniform quantization performs slightly worse, indicating that modeling actual JPEG internals matters (Afifi et al., 23 Sep 2025).

Training uses the S24 training set, comprising 2,619 raw images from the Samsung S24 Ultra main camera, with a 400-image S24 test set. Training is patch-based on non-overlapping 512×512 patches, for 100 epochs, with Adam, learning rate 0.001, $\theta$0, weight decay $\theta$1, and cosine decay to $\theta$2. Separate models are trained for each JPEG quality $\theta$3. Augmentations include standard geometry, intensity scaling by a random factor in $\theta$4, and a random $\theta$5 color transform with a small +0.05 boost to the green row before normalization (Afifi et al., 23 Sep 2025).

5. Empirical behavior and performance

On the S24 test set, Raw-JPEG Adapter consistently outperforms direct JPEG-on-raw across all reported qualities. Representative results are shown below (Afifi et al., 23 Sep 2025).

JPEG quality	Direct JPEG PSNR	Raw-JPEG Adapter PSNR
25	36.58	39.82
75	40.75	43.58
100	46.00	49.04

At Q=100, direct JPEG yields 46.00 PSNR, 98.47 SSIM, 99.72 MS-SSIM, 4.36 BPP, CR 6.51, whereas Raw-JPEG Adapter with DCT and augmentation reaches 49.04 PSNR, 99.35 SSIM, 99.89 MS-SSIM, 5.28 BPP, CR 5.41. Across qualities 25, 50, 75, 95, the gain over direct JPEG is typically about 2.6–3.2 dB on the training camera. The paper identifies fixed gamma as a strong simple baseline, but the full adapter remains better (Afifi et al., 23 Sep 2025).

Cross-camera evaluation changes the ranking of components. At Q=75, the spatial-only variant (without DCT) generalizes best:

Dataset at Q=75	Direct JPEG PSNR	Best adapter PSNR
S7	40.94	43.46
MIT-Adobe FiveK	43.91	46.75
NUS	42.32	45.97

This pattern supports the paper’s conclusion that the DCT scaling matrix captures camera-dependent biases and helps most in-domain, while the spatial transforms carry better across unseen sensors (Afifi et al., 23 Sep 2025).

The method also improves downstream rendering. For example, on S24 with LiteISP at Q=75, direct JPEG raw gives 20.78 PSNR / 78.99 SSIM, fixed gamma gives 24.00 / 85.80, and Raw-JPEG Adapter without DCT gives 24.60 / 87.99. The paper further reports that at JPEG quality 50, the method can reduce storage to under 2 MB while producing Lightroom-rendered images perceptually similar to those from original DNGs larger than 30 MB (Afifi et al., 23 Sep 2025).

The framework is not strictly JPEG-specific. It also improves LIC-TCM and JPEG 2000, though the paper treats JPEG as the central practical target because of ecosystem compatibility. Conventional BPP rises relative to direct JPEG because of both transformed image statistics and metadata, but the paper introduces a weighted BPP,

$\theta$6

to reflect preserved tonal richness (Afifi et al., 23 Sep 2025).

6. Position within related adapter research

RawJPEG Adapter belongs to a broader family of interface methods that bridge raw sensor data, JPEG structure, and pretrained vision systems. One adjacent direction is invertible ISP: InvISP learns a bijective mapping between demosaiced RAW and RGB, includes a differentiable JPEG simulator, and is strongest when the forward ISP itself is under control, rather than for arbitrary legacy JPEGs (Xing et al., 2021). Another direction is RAW-to-pretrained-model adaptation: RAW-Adapter combines input-level learnable ISP stages with model-level adapters to connect RAW/ISP priors to sRGB-pretrained backbones, addressing recognition rather than storage or reconstruction (Cui et al., 2024).

JPEG-specific front ends appear in several forms. JPEG-DL prepends a trainable JPEG layer as the first layer of an underlying DNN, replacing hard quantization with a differentiable soft quantizer; however, the downstream model still consumes a reconstructed RGB image, so it is a pixel-domain adapter rather than a raw-storage mechanism (Salamah et al., 2024). By contrast, “RGB no more” trains ViTs directly from minimally decoded JPEG features—specifically dequantized DCT coefficients—and reports up to 39.2% faster training and 17.9% faster inference with no accuracy loss, illustrating a compressed-domain adapter that bypasses full RGB decoding (Park et al., 2022).

Backward-compatible JPEG augmentation provides another neighboring line. JPNeO adds learned operators at both encoding and decoding while maintaining full backward compatibility with the current JPEG format, and JPEG XL defines a distinct losslessly transcoded JPEG mode that reconstructs the original JPEG1 codestream. These works preserve legacy JPEG interoperability, but they target JPEG-domain fidelity and compatibility rather than raw-domain reconstruction from a JPEG-carried transform (Han et al., 31 Jul 2025, Rhatushnyak et al., 2019).

7. Limitations and implications

Raw-JPEG Adapter is explicitly not lossless raw compression. Artifacts remain, especially at low quality. File size increases relative to direct JPEG because of both metadata and transformed-image statistics. The method is camera-sensitive: the optional DCT component helps on the training camera but hurts cross-camera generalization. The published system trains a separate model for each JPEG quality, works on demosaiced raw RGB rather than native CFA, and requires a special decoder to recover the raw-domain representation; generic viewers only show the adapted JPEG image (Afifi et al., 23 Sep 2025).

These limits clarify its proper scope. The method is best understood as a learned, invertible front-end/back-end for standard JPEG, not as a universal inverse-ISP system and not as a raw-native replacement for all camera containers. A plausible implication is that its most durable contribution lies in the adapter principle itself: raw data can be “bent” into JPEG’s operating regime by compact invertible transforms, then “unbent” after decoding, allowing standard JPEG infrastructure to carry substantially more raw-domain information than ordinary direct storage would preserve (Afifi et al., 23 Sep 2025).