diff --git a/ci/requirements_gh.txt b/ci/requirements_gh.txt
index f8c33636..55f93a01 100644
--- a/ci/requirements_gh.txt
+++ b/ci/requirements_gh.txt
@@ -14,3 +14,4 @@ Pillow
 lxml>=4.6.3
 pyqt5
 matplotlib
+hdf5plugin
diff --git a/pyproject.toml b/pyproject.toml
index bf090e8a..d55cddb6 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -71,6 +71,7 @@ densify_Bragg = 'fabio.app.densify:main'
 eiger2cbf = 'fabio.app.eiger2cbf:main'
 eiger2crysalis = 'fabio.app.eiger2crysalis:main'
 fabio-convert = 'fabio.app.convert:main'
+hdf2neggia = 'fabio.app.hdf2neggia:main'
 
 [project.gui-scripts]
 fabio_viewer = 'fabio.app.viewer:main'
diff --git a/src/fabio/app/hdf2neggia.py b/src/fabio/app/hdf2neggia.py
new file mode 100755
index 00000000..366f52b0
--- /dev/null
+++ b/src/fabio/app/hdf2neggia.py
@@ -0,0 +1,151 @@
+#!/usr/bin/env python
+
+__date__ = "10/07/2024"
+__author__ = "Jerome Kieffer"
+__license__ = "MIT"
+
+import os
+import sys
+import argparse
+import logging
+application_name = os.path.splitext(os.path.basename(__file__))[0]
+logger = logging.getLogger(application_name)
+logging.basicConfig()
+import numpy
+import fabio
+from fabio.nexus import Nexus
+import h5py
+import hdf5plugin
+try:
+    from pyFAI import load
+except ImportError:
+    pyFAI = None
+    logger.error("Unable to import pyFAI, won't be able to parse PONI-file!")
+
+
+def parse(argv=None):
+    if argv is None:
+        argv = []
+    parser = argparse.ArgumentParser(prog=application_name,
+                    description='Convert any HDF5 file containing images to a file processable by XDS using the neggia plugin from Dectris.'
+                                 'Do not forget to specify LIB=/path/to/plugin/dectris-neggia.so in XDS.INP',
+                    epilog='Requires hdf5plugin and pyFAI to parse configuration file. Geometry can be calibrated with pyFAI-calib2')
+    parser.add_argument("input", nargs='+',  help="Space separated list of input files (HDF5)")
+    parser.add_argument("--verbose", "-v", help="increase output verbosity",
+                        action="count")
+    parser.add_argument("--force", "-f", help="force overwrite output file",
+                        action="store_true")
+    parser.add_argument("--copy", "-c", help="copy dataset instead of using external links",
+                        action="store_true")
+    parser.add_argument("--geometry", "-g", help="PONI-file containing the geometry (pyFAI format, MANDATORY)")
+    parser.add_argument("--output", "-o", help="output filename", default="master.h5")
+    return parser.parse_args(argv)
+
+def process(options):
+    if options.verbose:
+        if options.verbose>1:
+            logger.setLevel(logging.debug)
+        else:
+            logger.setLevel(logging.info)
+
+    if not options.geometry or not os.path.exists(options.geometry):
+        logger.error("Unable to parse PONI-file: %s", options.geometry)
+        return 1
+    try:
+        poni = load(options.geometry)
+    except Exception as err:
+        logger.error("Unable to parse PONI-file: %s", options.geometry)
+        raise err
+        return 1
+    f2d = poni.getFit2D()
+    
+    if len(options.input)==0:
+        logger.error("No input HDF5 file provided. Aborting")
+        return 1
+    frames = [fabio.open(i) for i in options.input]
+    if os.path.exists(options.output):
+        if options.force:
+            mode = "w"
+        else:
+            logger.error("Output file exist, not overwriting it. Aborting")
+            return 1
+    else:
+        mode = "w"
+
+    dest_dir = os.path.dirname(os.path.abspath(options.output))
+    with Nexus(options.output, mode) as nxs:
+        entry = nxs.new_entry(entry="entry", program_name=application_name, force_name=True)
+        instrument = nxs.new_instrument(entry=entry, instrument_name="instrument")
+        if poni.wavelength:
+            beam = nxs.new_class(instrument, "beam", "NXbeam")
+            beam.create_dataset("incident_wavelength", data=poni.wavelength*1e10).attrs["unit"] = "A"
+        detector = nxs.new_class(instrument, "detector", "NXdetector")
+        detector.create_dataset("x_pixel_size", data=float(poni.pixel2)).attrs["unit"] = "m"
+        detector.create_dataset("y_pixel_size", data=float(poni.pixel1)).attrs["unit"] = "m"
+        detector.create_dataset("beam_center_x", data=float(f2d["centerX"])).attrs["unit"] = "pixel"
+        detector.create_dataset("beam_center_y", data=float(f2d["centerY"])).attrs["unit"] = "pixel"
+        detector.create_dataset("detector_distance", data=f2d["directDist"]*1e-3).attrs["unit"] = "m"
+        detectorSpecific = nxs.new_class(detector, "detectorSpecific", "NXcollection")
+        detectorSpecific.create_dataset("nimages", data=sum(i.nframes for i in frames))
+        detectorSpecific.create_dataset("ntrigger", data=1)
+        mask = poni.detector.mask
+        if mask is None:
+            mask = numpy.zeros(poni.detector.shape, dtype="uint32")
+        else:
+            mask = mask.astype("uint32")
+        detectorSpecific.create_dataset("pixel_mask", data=mask)
+        data = nxs.new_class(entry, "data", "NXdata")
+        cnt = 0
+        for fimg in frames:
+            if isinstance(fimg.dataset, h5py.Dataset):
+                cnt += 1
+                if options.copy:
+                    data.create_dataset(f"data_{cnt:06d}", data=fimg.dataset[()],
+                                        chunks=(1,)+fimg.shape,
+                                        **hdf5plugin.Bitshuffle(nelems=0, cname='lz4'))
+                else:
+                    data[f"data_{cnt:06d}"] = h5py.ExternalLink(os.path.relpath(fimg.dataset.file.filename, dest_dir), fimg.dataset.name)
+            elif isinstance(fimg.dataset, numpy.ndarray):
+                cnt += 1
+                if fimg.dataset.ndim < 3:
+                    dataset = numpy.atleast_3d(fimg.dataset)
+                    dataset.shape = (1,)*(3-fimg.dataset.ndim)+fimg.dataset.shape
+                else:
+                    dataset = fimg.dataset
+                data.create_dataset(f"data_{cnt:06d}", data=dataset,
+                                    chunks=(1,)+fimg.shape,
+                                    **hdf5plugin.Bitshuffle(nelems=0, cname='lz4'))
+            else: # assume it is a list
+                for item in fimg.dataset:
+                    # each item can be either a dataset or a numpy array
+                    if isinstance(item, h5py.Dataset):
+                        cnt += 1
+                        if options.copy:
+                            data.create_dataset(f"data_{cnt:06d}", data=item[()],
+                                                chunks=(1,)+fimg.shape,
+                                                **hdf5plugin.Bitshuffle(nelems=0, cname='lz4'))
+                        else:
+                            data[f"data_{cnt:06d}"] = h5py.ExternalLink(os.path.relpath(item.file.filename, dest_dir), item.name)
+
+                    elif isinstance(fimg.dataset, numpy.ndarray):
+                        cnt += 1
+                        if fimg.dataset.ndim < 3:
+                            dataset = numpy.atleast_3d(item)
+                            dataset.shape = (1,)*(3-item.ndim)+item.shape
+                        else:
+                            dataset = item
+                        data.create_dataset(f"data_{cnt:06d}", data=dataset,
+                                            chunks=(1,)+fimg.shape,
+                                            **hdf5plugin.Bitshuffle(nelems=0, cname='lz4'))
+                    else:
+                        logger.warning("Don't know how to handle %s, skipping", item)
+    return 0
+
+
+def main():
+    options = parse(sys.argv[1:])
+    return process(options)
+
+
+if __name__=="__main__":
+    sys.exit(main())
diff --git a/src/fabio/app/meson.build b/src/fabio/app/meson.build
index d7b5429d..a4ad94fd 100644
--- a/src/fabio/app/meson.build
+++ b/src/fabio/app/meson.build
@@ -8,6 +8,7 @@ py.install_sources(
   'viewer.py',
   'densify.py',
   'eiger2crysalis.py',
+  'hdf2neggia.py'
   ]
 ,
   pure: false,   # Will be installed next to binaries