Corrupted data after compression

[RobinFrcd]

Hi,
I'm trying to compress an uncompressed dicom: example.zip

The code I'm using to do it:

import pydicom
from deid.dicom import replace_identifiers
from PIL import Image
from io import BytesIO
from pydicom.encaps import encapsulate

dcm = pydicom.read_file('example.dcm')

with BytesIO() as bio:
    Image.fromarray(dcm.pixel_array).save(
        bio,
        optimize=True,
        quality=95,
        format="JPEG",
        lossless=True,
    )
    dcm.PixelData = encapsulate([bio.getvalue()])
    dcm.file_meta.TransferSyntaxUID = pydicom.uid.JPEGBaseline
    
dcm.is_little_endian = True
dcm.is_implicit_VR = False
dcm[0x7fe0, 0x0010].VR = "OB"
dcm.save_as('dcm_comp.dcm')

But when I try to convert the dicom to JPG with dcm2pnm, it says can't change to unencapsulated representation for pixel data.

$ dcm2pnm -d -v /tmp/dcm/dcm_comp.dcm /tmp/dcm/dcm_comp.jpg
I: reading DICOM file: /tmp/dcm/dcm_comp.dcm
I: preparing pixel data
E: can't change to unencapsulated representation for pixel data
E: can't determine 'PhotometricInterpretation' of decompressed image
E: mandatory attribute 'PhotometricInterpretation' is missing or can't be determined
F: Missing attribute

module       | version
------       | -------
platform     | Linux-5.19.0-76051900-generic-x86_64-with-glibc2.35
Python       | 3.10.6 (main, Aug 17 2022, 14:01:22) [GCC 11.2.0]
pydicom      | 2.3.0
gdcm         | 3.0.19
jpeg_ls      | _module not found_
numpy        | 1.22.4
PIL          | 9.2.0
pylibjpeg    | 1.4.0
openjpeg     | _module not found_
libjpeg      | _module not found_

dcms.zip

[darcymason]

E: mandatory attribute 'PhotometricInterpretation' is missing or can't be determined

I think this is clear - you need to correctly set the PhotometricInterpretation.

I'm not expert on compression, maybe others can help with some of the details. The original file was RGB, but I suspect PIL's jpeg is one of the YBR types.

In any case, no evidence that it is a pydicom issue at this point, so converting to discussion...

[RobinFrcd]

PhotometricInterpretation is already set to RGB. I've already tried to change it to other YBR_ values but didn't help.