Skip to content

Sample JPH images, please #113

Closed
Closed

Description

@comstock

Hi. I'm looking for sample JPH images, properly encoded, for testing purposes. Could you supply any?

FYI: The JPH files in the jp2k_test_codestreams repository are all found not valid according to jpylyzer, recently updated to accommodate JPH.

Activity

  1. aous72 commented on Nov 11, 2023

    @aous72
    Owner

    Hi comstock,

    I tested with jpylyzer -- I suspect it has a bug, but more careful check is needed.

    Notes:

    1. Perhaps you already know that, in jpylyzer, you need to specify the format to test, the default is ".jp2." You should use --format jph.
    2. Most of the .jph images in jp2k_test_codestreams/openjph pass with jpylyzer. The tiled ones sometimes fail, but that might be a bug in jpylyzer (We need to check with jpylyzer main developer @bitsgalore); of course, it is possible the problem is in the encoder -- the encoder used to produce the jph images is in a comment in the jph file; this comment is produced by jpylyzer.
    3. The official conformance test images are in 'https://gitlab.com/wg1/htj2k-codestreams/' which has the testfiles_jph and testfiles_jp2 folders. There are files in 'https://github.com/osamu620/OpenHTJ2K/tree/main/conformance_data' which may have the wrong extension to work with jpylyzer.
    4. You can create your own with 'https://kakadusoftware.com/documentation-downloads/downloads/' which I rely on.

    There are resource on 'https://github.com/chafey/HTJ2KResources', but I am not sure if there are any images.

    Let me know how it goes with you. You can also join the HTJ2K discord server 'https://discord.gg/U4Zbq7RGDn' if you like.

    Cheers,
    Aous.

  2. comstock commented on Nov 12, 2023

    @comstock
    Author

    Thank you, so much.

  3. bitsgalore commented on Nov 13, 2023

    @bitsgalore

    I just had a look at this, and I'm able to reproduce the results reported by @aous72. In all cases the offending tests are (this happens with some of the JPHs and some of the codestreams):

        <tests>
            <foundExpectedNumberOfTiles>False</foundExpectedNumberOfTiles>
            <foundExpectedNumberOfTileParts>False</foundExpectedNumberOfTileParts>
        </tests>

    I did a couple of tests, and from what I can see the reason this error is reported, is that the tile-parts aren't implemented correctly.

    See for example this one:

    jpylyzer --format jhc simple_enc_irv97_tall_narrow1.j2c > simple_enc_irv97_tall_narrow1.xml
    

    Based on the image geometry defined in the SIZ marker (values of xsiz, x0siz, xTsiz, ysiz, y0siz, and yTsiz), this image only contains one single tile:

    <xsiz>17</xsiz>
    <ysiz>256</ysiz>
    <xOsiz>1</xOsiz>
    <yOsiz>0</yOsiz>
    <xTsiz>16</xTsiz>
    <yTsiz>256</yTsiz>
    <xTOsiz>0</xTOsiz>
    <yTOsiz>0</yTOsiz>
    <numberOfTiles>1</numberOfTiles>

    It also contains two tile-parts, with the following properties:

    <tileParts>
        <tilePart>
            <sot>
                <lsot>10</lsot>
                <isot>0</isot>
                <psot>739</psot>
                <tpsot>0</tpsot>
                <tnsot>1</tnsot>
            </sot>
            <pltCount>0</pltCount>
            <pptCount>0</pptCount>
        </tilePart>
        <tilePart>
            <sot>
                <lsot>10</lsot>
                <isot>1</isot>
                <psot>96</psot>
                <tpsot>0</tpsot>
                <tnsot>1</tnsot>
            </sot>
            <pltCount>0</pltCount>
            <pptCount>0</pptCount>
        </tilePart>
    </tileParts>

    Note that both tile-parts have different isot values. Since isot (tile index) links each tile-part to its respective tile (see clause A.4.2 of Part 1), for a single-tile image, the expected value here would be 0 for both tile-parts (after all they're both part of the same tile)!

    Also, tnsot represents the number of tile-parts of a tile in the codestream. Its value is either 0 (which means that the number of tile-parts of this tile is not specified in this tile-part), or the correct number of tile-parts for that tile. Since we have a single-tile image with two tile-parts, the expected value here would be either 0 or 2 (not 1!).

    So unless I'm overlooking something obvious, I think this is really a fault of the images.

    BTW Jpylyzer's behaviour is not affected in any way by file extensions. As long as you set the --format option correctly, Jpylyzer will validate against that format, irrespective of the file extension of the file

  4. aous72 commented on Nov 13, 2023

    @aous72
    Owner
  5. bitsgalore commented on Nov 14, 2023

    @bitsgalore

    Turns out this is indeed a Jpylyzer bug, see: openpreserve/jpylyzer#221 (comment)

  6. Repository owner locked and limited conversation to collaborators on Apr 30, 2024
  7. converted this issue into a discussion #138 on Apr 30, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions