POV-Ray : Newsgroups : povray.binaries.programming : An updated povr tarball for Unix/Linux. b0bcf6c0 Server Time
19 Jun 2021 21:56:42 EDT (-0400)
  An updated povr tarball for Unix/Linux. b0bcf6c0 (Message 1 to 2 of 2)  
From: William F Pokorny
Subject: An updated povr tarball for Unix/Linux. b0bcf6c0
Date: 26 Apr 2021 07:25:32
Message: <6086a32c@news.povray.org>
The autotools build system has been significantly reworked and the povr 
tarballs are now created with 'make distcheck'. See the files 
INSTALL.txt and README.txt for more detail. The autotools quiet options 
are now available and sometimes defaulted. Use ./configure -q when 
configuring for the full quiet mode for configuration and build.

The default display is now X11 with a Simple Directmedia Layer an option 
by default, if an appropriate development library is found.

The srgb keywords now permit inputs ONLY in the 0 to 1 range for which 
the srgb color space is defined. Previously values outside the range 
generated linear values but ones not making any particular sense.

Fixes for broken dictionary copies. Suspect copies never worked - 
reliably - though perhaps sometimes appeared to work depending upon use.

Additional base v3.8 fixes not github issues documented in the news 
group: http://news.povray.org/povray.beta-test/

Of note .grey,.gray is broken in the v3.8 master branch and fixed in the 
povr branch. While fixing added a configure GREYREC option with several 
variations (111, 240, 601 709 and 2020) and the default was changed to 
the more reasonable sRGB assumed_gamma 1.0 BT709. Note 111 is the older 
color to gray conversion.

Additional changes have been made to the collection of inbuilt 
functions. See include/functions.inc for details. The majority of old 
functions for particular shapes no longer exist. This work is 
stabilizing, but it is not yet complete.

Of note are new functions support 21 bit and 32 bit encoding and 
decoding into and from a double. A few of the inbuilt functions make use 
of the capability to run once an internal vturbulence, say, and pass 
back 2d or 3d vectors.  Plus, f_dec_nbit3c and f_enc_nbit3c which permit 
the encoding of up to 51 bit (normally 48) bit depth values/grays in 
three color channels.

A significant number of self tests added to the core include files - 
especially functions.inc.

Fixed problems with method 3 antialiasing. Extended the older method 1 
and 2 modes so jitter is off by default. Jitter values, when on, can now 
be much larger than 1.0 - which is necessary to avoid grid artifacts and 
poor AA at larger recursion depths. Jitter also made truly random where 
it long has not been despite POV-Ray documentation. The aa values of 0.0 
now support fast paths - and with the big jitter values enable 
interesting effects.  See news groups for more detail pending better 
documentation.

A number of fixes and improvements to parsing and parse messages.

Patterns using internal turbulence moved to new it_ prefixed keywords. 
Where the pattern had an internal frequency or other keywords colliding 
with more general modifier keywords, new ip_ prefixed keywords were 
introduced. The ip_ prefix stands for internal pattern. The it_ prefix 
stands for internal turbulence.

The prefixed keywords added are : ip_frequency, ip_phase, ip_scale, 
it_depth, it_frequency, it_lambda, it_magnitude, it_octaves, it_omega, 
it_translate and it_scale. These keywords make clear intent and 
operation. These changes also fix issues long in POV-Ray where certain 
warps and operations like the frequency keyword were getting used twice 
for a completely incorrect results in some cases (ripples, waves) and in 
otherwise confusing and difficult to control results.

Isosurface evaluate feature pulled and a new report on/off keyword added 
so gradient warnings can be turned off or on by particular isosurface. 
Where functions filtered by other functions there is always a 
discontinuity on the function switch. Where these are not near the 
surface they don't matter to the result making the max gradient report 
noise.

Remaining default color maps removed for all no discrete patterns (for 
all continuous patterns)

The aoi pattern was changed to support the full -1 to +1 range as this 
today happens with perturbed normal patters where the original normal 
inverts with respect to the camera ray and the original surface.

The normal block quilted pattern was fixed for issues dating to its 
creation and extended lo, hi options added. The value based quilted 
pattern was dropped as it was anything, but quilted. New tiled and 
grooved normal block only patterns were added, the latter having too a 
flip keyword.

A new seedfi - "seed function interval" scalar value pattern was added 
with a -1 to 1 range which can be mangled by wave and pattern modifiers 
to create many additional patterns.

Improvements to mesh parsing code especially with respect to some 
particularly misleading error messages.

More generally where flags and keywords not supported in v3.7/v3.8 or 
the povr branch, they've been removed in a way which will cause errors 
during initialization phases and parsing.

Bill P.


Post a reply to this message


Attachments:
Download 'povray-3.8.0-x.povr_b0bcf6c0.tar.gz' (1749 KB)

From: clipka
Subject: Re: An updated povr tarball for Unix/Linux. b0bcf6c0
Date: 29 May 2021 21:08:19
Message: <60b2e583$1@news.povray.org>
Am 26.04.2021 um 13:25 schrieb William F Pokorny:

> The srgb keywords now permit inputs ONLY in the 0 to 1 range for which 
> the srgb color space is defined. Previously values outside the range 
> generated linear values but ones not making any particular sense.

Technically, contrary to Ive's claim (who is usually right in such 
things), the sRGB color space _is_ well-defined(*) for values outside 
the 0..1 range, both positive and negative.

The official sRGB specification (v1.10) refers to limiting the color 
values to the 0..1 regime as "typical", and _in such cases_ mandates 
that simple clipping is to be used for this purpose, but there is no 
provision mandating gamut limiting _per se_, and there is no fundamental 
mathematical problem with the functions outside the range.

(* Note that I'm not saying well-suited. That would be an entirely 
different discussion.)


Post a reply to this message

Copyright 2003-2021 Persistence of Vision Raytracer Pty. Ltd.