[
https://issues.apache.org/jira/browse/TIKA-3571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17512730#comment-17512730
]
Tilman Hausherr commented on TIKA-3571:
---------------------------------------
We don't use java font rendering, we have our own. I have observed some weird
temporary slowness in my own multithreaded rendering regression tests but never
investigated this.
PDFBox will always be one of the slowest rendering options available. I watched
a presentation by Dr. Michael Vrhel (artifex) a few months ago who compared the
rendering times, we or PDF.js were the slowest. I'm not really surprised,
because the other ones tested were written in native code.
> Add an interface for rendering engines
> --------------------------------------
>
> Key: TIKA-3571
> URL: https://issues.apache.org/jira/browse/TIKA-3571
> Project: Tika
> Issue Type: Wish
> Reporter: Tim Allison
> Priority: Major
>
> We've now seen a few requests for extracting text _and_ rendering PDFs, and
> certainly it might be useful to have alternatives for rendering files (e.g.
> this [Alfresco
> study|https://hub.alfresco.com/t5/alfresco-content-services-blog/pdf-rendering-engine-performance-and-fidelity-comparison/ba-p/287618]),
> including MSOffice or at least PPTx...
> And there are cases where users don't want the rendered images, but they do
> want OCR to be run against the rendered images.
> I doubt I'll have a chance to work on this for a while, but I wanted to open
> an issue for discussion.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
