Introducing Myself and ProbRAW: Reproducible RAW Processing for Scientific and Forensic Imaging

Processing
, ,
1

Hello everyone,

my name is Alejandro Maestre . I am a photographer, forensic expert specialized in digital and multimedia evidence analysis, CIO and co-founder of Probatia Forensics , and president of the Spanish Association of Scientific and Forensic Imaging .

My professional work is located at the intersection of photography, scientific imaging, digital evidence, technical documentation and forensic analysis. I am especially interested in imaging workflows where the photograph is not treated merely as a creative or aesthetic output, but as a source of information, measurement, documentation or evidence. In those contexts, the way an image is captured, processed, exported and preserved becomes critically important.

I am currently developing an open-source project called ProbRAW . The goal of the project is to build a RAW/TIFF processing environment focused on reproducibility, traceability and technical auditability . ProbRAW is not intended to be another general-purpose creative RAW editor. Instead, it is being designed as a controlled processing tool for scientific, forensic and cultural heritage photography, where every relevant processing decision should be explicit, repeatable and reviewable.

Some of the main technical ideas behind ProbRAW are:

  • reproducible RAW-to-TIFF development;
  • non-destructive, parametric processing recipes;
  • ICC-based color management and session profiling;
  • support for controlled workflows using color targets and calibration references;
  • preservation of processing parameters alongside derived files;
  • generation of technically documented TIFF outputs;
  • hash-based traceability of source and derived files;
  • sharpness and image quality analysis using methods such as ESF, LSF and MTF;
  • future integration of provenance mechanisms such as C2PA manifests;
  • a workflow suitable for forensic, scientific and cultural heritage documentation.

One of the key principles of the project is that a processed image should not be a “black box” result. In a scientific or forensic context, it should be possible to explain how a TIFF derivative was obtained from a RAW file, which parameters were applied, which color profile was used, whether sharpening or noise reduction was introduced, and whether the resulting image remains suitable for interpretation, measurement or evidential use.

I am particularly interested in discussing questions such as:

  • How should RAW development be documented in scientific imaging?
  • What is the most defensible way to combine ICC/DCP-style workflows with reproducible RAW processing?
  • How should sharpening, denoising and tone mapping be limited or documented when the image may later be used as evidence?
  • What metadata and provenance records should accompany a derived TIFF file?
  • How can MTF/ESF analysis be integrated in a way that is useful but not misleading?
  • What practices from scientific imaging, cultural heritage imaging or forensic imaging should be considered essential?

The project is still young and under active development, so my main reason for joining this community is to learn, receive critical feedback and discuss methodology with people who have deep experience in RAW processing, color management, scientific imaging, microscopy, macro photography, digital preservation or forensic documentation.

I would be very grateful for any comments, criticism, references, technical suggestions or methodological concerns that could help improve the project.

Project repository:

Best regards,

Alejandro Maestre

9 Likes
2

Welcome to the forum.

I have a few ideas about what you’re after, embodied in my hack raw processor, aptly and un-imanginitively named ‘rawproc’. It lists the processing toolchain from raw to render in a “toolchain” pane, and saves that toolchain in the render’s metadata. No side-car shenanigans.

Further, if one opens a render for editing, it’ll offer the opportunity to instead re-open the source file and apply the processing from the render metadata, providing the starting point for further edit modification for another render.

https://github.com/butcherg/rawproc

2 Likes
3

Hi Glenn,

thank you very much for your welcome and for sharing your project. I really appreciate it. I will take a look at rawproc , because the idea of preserving the processing toolchain inside the render metadata is very close to some of the issues I am trying to address with ProbRAW.

In parametric RAW processing, the use of sidecar files is of course a common and practical solution, mainly because the original RAW file should not be altered — with the particular exception of workflows based on DNG, where metadata can sometimes be embedded. So it is entirely normal that the processing parameters, development settings and editing decisions are stored externally in a sidecar file.

However, from a forensic or evidential point of view, a sidecar file by itself is not a strong guarantee. It can document the processing decisions, but it does not necessarily prove who created those metadata, when they were created, whether they have been modified afterwards, or whether they actually correspond to the original RAW file under examination. In other words, it is very useful as a technical record, but it is not sufficient as a probative mechanism.

This is why I am interested in exploring C2PA-based provenance metadata for ProbRAW. In that model, the processing history would not only describe the applied changes, but could also provide stronger guarantees about the origin and authenticity of those metadata through digital signatures and trusted timestamps. Ideally, the provenance record should also include the hash of the original RAW file, so that the rendered TIFF can be linked in a verifiable way to its source file and to the documented processing chain.

So, in short, I see sidecars and embedded processing metadata as very valuable for reproducibility, but I am also interested in adding a layer of cryptographic provenance and verification for forensic, scientific and evidential use cases.

Thanks again for the reference. I will review your approach with great interest.

Best regards,

Alejandro

1 Like