[
https://issues.apache.org/jira/browse/PDFBOX-4832?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17105635#comment-17105635
]
Tilman Hausherr commented on PDFBOX-4832:
-----------------------------------------
I assume you mean {{isScaledUp}}. For some reason, that one takes only the CTM
into account and not the scaling of the graphics device.
> Binary CCITT image with nearest neighbor interpolation instead gets upscaled
> with bicubic, PDFBox thinks it's downscaling
> -------------------------------------------------------------------------------------------------------------------------
>
> Key: PDFBOX-4832
> URL: https://issues.apache.org/jira/browse/PDFBOX-4832
> Project: PDFBox
> Issue Type: Bug
> Components: Rendering
> Affects Versions: 2.0.19
> Reporter: Gábor Stefanik
> Priority: Major
> Attachments: pdf_scanned-eng-bw.pdf
>
>
> The attached PDF contains a CCITT-encoded black-and-white image of a roughly
> US Letter-sized page (611x783 PDF points page size), scanned at 200DPI. The
> image is slightly undersized, at 1696 pixels wide, instead of the expected
> 1697. The PDF calls for this image to be upscaled using nearest-neighbor
> interpolation (downscaling is always smoothly interpolated, to avoid
> artifacts - see PDFBOX-2364).
> When I try to render this page to an RGB image in either 200 or 300 DPI, the
> image unexpectedly ends up containing shades of gray. Upon closer inspection,
> it is clear that the upscaling was performed using bicubic interpolation
> instead of nearest-neighbor. Stepping through the PDF rendering code reveals
> that, in both cases, PDFBox somehow thinks the image is being _down_scaled,
> and applies the behavior from PDFBOX-2364. Additionally, PDFBOX-4831 may also
> be at play here, as the computed target image dimensions aren't round
> integers, but very slightly undersized (e.g. 2174.999858106px instead of
> 2175px).
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
