Script 'mail_helper' called by obssrc
Hello community,

here is the log from the commit of package openhtj2k for openSUSE:Factory 
checked in at 2026-11-10 14:44:01
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Comparing /work/SRC/openSUSE:Factory/openhtj2k (Old)
 and      /work/SRC/openSUSE:Factory/.openhtj2k.new.3759290 (New)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Package is "openhtj2k"

Tue Nov 10 14:44:01 2026 rev:15 rq:1396475 version:0.26.0

Changes:
--------
--- /work/SRC/openSUSE:Factory/openhtj2k/openhtj2k.changes      2026-11-09 
16:08:29.915504265 +0100
+++ /work/SRC/openSUSE:Factory/.openhtj2k.new.3759290/openhtj2k.changes 
2026-11-10 14:45:27.971248518 +0100
@@ -1,0 +2,48 @@
+Tue Nov 10 08:49:41 UTC 2026 - Michael Vetter <[email protected]>
+
+- Update to 0.26.0:
+  Added:
+  * Qweights=flat (openhtj2k_encoder::set_qfactor_weights(1)) turns off the
+    visual weights of the Qfactor path, for scientific or other non-visual
+    data. Every subband weight is 1 at every Qfactor, so the steps are
+    MSE-optimal, as the guideline's are from Qfactor=97 up; without the MCT
+    every component is independent data with unit colour gain. The Qfactor
+    still sets the reference step, so a Qfactor gives larger files and smaller
+    errors: on an 8-bit RGB image, 28-54% more bits and 2.6-3.8 dB more PSNR at
+    Qfactor=30 to 90. The default, Qweights=visual, is unchanged.
+    estimate_qfactor --weights flat recovers the Qfactor of such files.
+  Changed:
+  * Qfactor on sub-sampled (4:2:0 or 4:2:2) colour input follows version 2.0 of
+    the JPEG Qfactor guideline, "Controlling JPEG 2000 image quality using a
+    single parameter (Qfactor)" (ISO/IEC JTC 1/SC 29/WG 1 N101553). Its chroma
+    weights include the gain of upsampling the component for display, 2 for
+    4:2:0 and 1.4142 for 4:2:2, on every subband including LL; the encoder used
+    the version 1.0 weights, which leave it out. Below Qfactor=97 the chroma
+    steps of such files are finer by up to 2 (4:2:0) or 1.4142 (4:2:2) times,
+    so the files are larger: a 1024x1024 4:2:0 test image grows by 41% at
+    Qfactor=30, 23% at 75 and 9% at 90. Output for gray, gray + alpha and 4:4:4
+    colour input, and all output without Qfactor, is byte-identical. The
+    experimental analytic weighting (Qcsf=mannos|daly) applies the same gain to
+    sub-sampled components. estimate_qfactor follows the new weights, so on a
+    sub-sampled file written by 0.25.0 or earlier it finds the Qfactor from the
+    luminance component only and reports a mismatch on the chroma components.
+  Fixed:
+  * On the single-tile reuse path (invoke_line_based_stream_reuse() and
+    invoke_line_based_direct() after enable_single_tile_reuse(true)), a frame
+    whose headers differed from the previous frame's could decode with the
+    previous frame's step sizes, packet order or line-decoder sizes. The
+    decoder rebuilt its reused structure only when a few header values changed
+    (SIZ, part of COD, the first QCD step size, and the numbers of COC, QCC and
+    RGN segments); it ignored the rest of the main header, the tile-part
+    headers and the resolution reduction. It now compares a hash of every
+    marker segment that shapes the tile, in the main header and the tile-part
+    headers, together with the reduction. A frame whose tile-part header gives
+    a component another number of decomposition levels, which 0.25.0 refused,
+    now decodes as well. Frames with unchanged headers reuse the structure as
+    fast as before (29.5 ms per 4K frame). (#545)
+  * Every source file includes the standard headers it uses, so the library and
+    tools build with standard libraries that no longer bring one header in
+    through another, such as the libc++ of current Emscripten releases
+    (std::strtoull in jpip_response.cpp).
+
+-------------------------------------------------------------------

Old:
----
  v0.25.0.tar.gz

New:
----
  v0.26.0.tar.gz

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Other differences:
------------------
++++++ openhtj2k.spec ++++++
--- /var/tmp/diff_new_pack.GtUxO0/_old  2026-11-10 14:45:41.447813094 +0100
+++ /var/tmp/diff_new_pack.GtUxO0/_new  2026-11-10 14:45:41.449813178 +0100
@@ -17,7 +17,7 @@
 
 
 Name:           openhtj2k
-Version:        0.25.0
+Version:        0.26.0
 Release:        0
 Summary:        An open source implementation of ITU-T Rec.814 | ISO 15444-15 
(a.k.a. HTJ2K)
 License:        BSD-3-Clause

++++++ v0.25.0.tar.gz -> v0.26.0.tar.gz ++++++
/work/SRC/openSUSE:Factory/openhtj2k/v0.25.0.tar.gz 
/work/SRC/openSUSE:Factory/.openhtj2k.new.3759290/v0.26.0.tar.gz differ: char 
13, line 1

Reply via email to