[
https://issues.apache.org/jira/browse/PDFBOX-6155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18074267#comment-18074267
]
Tilman Hausherr commented on PDFBOX-6155:
-----------------------------------------
Yes
> AT (Adaptive template pixels) -1 byte skipped
> ---------------------------------------------
>
> Key: PDFBOX-6155
> URL: https://issues.apache.org/jira/browse/PDFBOX-6155
> Project: PDFBox
> Issue Type: Sub-task
> Components: JBIG2
> Affects Versions: 3.0.4 JBIG2
> Reporter: Tilman Hausherr
> Assignee: Tilman Hausherr
> Priority: Major
> Attachments: GenericRefinementRegion215-1776436118615-1.png,
> GenericRefinementRegion267-null-1776436118627.png,
> bitmap-symbol-textrefine-customat.pdf,
> bitmap-symbol-textrefine-negative-delta-width.pdf,
> bitmap-symbol-textrefine.pdf
>
>
> bitmap-symbol-textrefine-customat.pdf has a garbled shape, however
> bitmap-symbol-textrefine.pdf doesn't. The difference is that the problematic
> one has 4 AT bytes.
> [https://github.com/SerenityOS/serenity/blob/master/Tests/LibGfx/test-inputs/jbig2/json/bitmap-symbol-textrefine-customat.json]
> {code:json}
> "refinement_adaptive_template_pixels": [
> 10,
> -1,
> 4,
> 15
> ],
> {code}
> After a lot of reading and searching and trying I was able to trace it to
> {{GenericRefinementRegion.updateOverride()}} which ignores AT bytes if they
> are -1. However the specification says no such thing:
> 6.3.5.3 Fixed templates and adaptive templates
> The AT pixel RA1 can be located anywhere in the field shown in Figure 7, not
> including the current pixel. The AT pixel
> RA2 can be located anywhere in the range (-128, -128) to (127, 127) in the
> reference bitmap.
> See also figure 7 earlier in the specification, which has (-1,-1) values.
> Just removing the two lines fixes the problem but ruins
> bitmap-symbol-textrefine-negative-delta-width.pdf, which has all -1 so I'll
> research that.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
