heckflosse
845ebe6906
Faster loading of pixelshift files
2017-03-13 15:44:36 +01:00
Alberto Griggio
7a69ff84a4
use 1-pass method for X-trans demosaic in the fast pipeline
...
As suggested by @heckflosse
2017-03-13 15:12:00 +01:00
Alberto Griggio
6f08233c6f
further adjustments of processing parameters for the fast export pipeline
2017-03-13 14:00:37 +01:00
Alberto Griggio
bdde222609
merge with dev
2017-03-13 09:11:54 +01:00
Hombre
567ed3629c
Automatically selected profiles are now correctly loaded on first use.
2017-03-13 01:30:34 +01:00
Alberto Griggio
d18ce59132
some further tweaks to the fast pipeline
2017-03-12 21:13:54 +01:00
Alberto Griggio
f42710f594
properly rescale the impulseDenoise settings in the fast pipeline
2017-03-12 20:52:32 +01:00
heckflosse
174949e0f0
Fix slowdown for decode of non pixelshift files
2017-03-12 20:39:03 +01:00
Desmis
11940f9121
merge with dev
2017-03-12 16:53:50 +01:00
heckflosse
e2b4623dc5
Merge branch 'dev' into psgtk3
2017-03-11 22:59:08 +01:00
heckflosse
f9a536fd0b
new gtk3 version of pixelshift
2017-03-11 22:30:36 +01:00
Flössie
ebd9e5e132
Make rtengine::Cache build again with hashable keys
...
While STL containers are not guaranteed to build with incomplete types,
`rtengine::Cache` used to build with `std::map` as well as
`std::unordered_map` when I first implemented it. With GCC 6.3 I
recently realized, that this was no longer the case for hashable keys
(i.e. `std::unordered_map` as `Store`).
The recommended workaround is to use a `std::unique_ptr`.
2017-03-11 17:14:11 +01:00
heckflosse
f226934b57
Pixelshift: Added option to equalize brightness of frames and cleaned gui.
2017-03-11 00:25:49 +01:00
Alberto Griggio
359b4136b5
do not use the hard-coded (i.e. not user-selectable) fast export options if the new fast pipeline is selected
2017-03-10 10:53:10 +01:00
Alberto Griggio
64661e1a27
set dirpyrDenoise.median=false in the fast export pipeline
2017-03-09 00:05:27 +01:00
Alberto Griggio
30931359f8
merged with dev
2017-03-08 22:55:32 +01:00
Alberto Griggio
44505ccb08
some enhancements of the Fast Export functionality
...
- remember fast export options across RT sessions
- added option to use a dedicated "Fast Export" pipeline, trading speed for
quality (resizes early instead of at the end)
2017-03-08 22:55:08 +01:00
Flössie
61b913f7f9
Fix rtengine::min() for NaNs ( #3742 )
...
Also fix `LuminanceToneCurve::Apply()`. Kudos to @heckflosse!
2017-03-08 20:23:57 +01:00
Alberto Griggio
93296cff17
continuing with my experiments on an alternative "fast export"
2017-03-08 00:50:50 +01:00
Alberto Griggio
200e6dd882
experimental hacks on simpleprocess.cc
2017-03-07 17:38:32 +01:00
Ingo Weyrich
44ecdf039a
Merge pull request #3734 from Beep6581/shartifact
...
Shadows/Highlights: Use iterated boxblur for radius > 40 when Sharp m…
2017-03-07 00:10:59 +01:00
Ingo Weyrich
5d98668912
Merge pull request #3726 from agriggio/dynamicprofile
...
Dynamic processing profiles
2017-03-06 19:14:32 +01:00
heckflosse
ff5919be4a
Shadows/Highlights: Use iterated boxblur for radius > 40 when Sharp mask is disabled, fixes #3730
2017-03-06 16:52:31 +01:00
Ingo Weyrich
3684ac4053
Merge pull request #3725 from Beep6581/xtranscompressed
...
Support for compressed bayer files from FUJIFILM GFX 50S
2017-03-04 01:34:23 +01:00
Alberto Griggio
153e28deae
merge with dev
2017-03-03 11:23:22 +01:00
Floessie
29a72d1580
Merge pull request #3720 from Beep6581/variadic_min_max
...
Variadic template version of `rtengine::(min|max)()`
2017-03-02 07:40:34 +01:00
Flössie
6e7712831a
Break min() and max() parameter dependencies
...
Also convert most functions in `rt_math.h` to `constexpr` by
implementing `min()` and `max()` natively. `constexpr` indeed has a
positive impact on the generated assembly.
2017-03-01 18:13:16 +01:00
Flössie
e2b8ccd38b
Whitespace correction ( #3719 )
2017-02-28 21:16:10 +01:00
Flössie
d72d931c9e
Variadic template version of rtengine::(min|max)()
...
This change allows for an arbitrary number of arguments to `min()` and
`max()` by using recursion on variadic template functions. The
disassembly of GCC 6.3 was carefully checked for regressions, but
nothing was found other than the flipping of arguments (recursion is
now `(((a,b),c),d)` and was `(d,(c,(a,b)))` before).
I also unified the common type `_Tp` to to the even more common `T`.
2017-02-28 20:57:19 +01:00
heckflosse
5e5ca6eee4
disabled timing code in dcb_demosaic
2017-02-28 20:05:36 +01:00
Ingo Weyrich
37f69bed5b
Merge pull request #3719 from cuniek/dev
...
Improved DCB, less macroblocking on diagonals, much faster code
2017-02-28 19:31:16 +01:00
U-coolermaster2\cuniek
66382743cd
Improved DCB, less macroblicking on diagonals, much faster code
2017-02-27 19:41:23 +01:00
Hombre
6162cbd77d
Adding more profiles to the Output color profile combobox in ICM
...
You can now select output profiles of the "OUTPUT" device class, but
only if they have an RGB color-space.
2017-02-27 01:12:40 +01:00
Desmis
0abadba166
merge with dev
2017-02-24 10:58:12 +01:00
Desmis
08f130ed56
Fixed bug in preview
2017-02-24 10:40:27 +01:00
Ingo Weyrich
bc3003bc0d
Merge pull request #3713 from Beep6581/lcpvignette_speedup
...
Speedup for lcp vignette correction
2017-02-23 21:41:00 +01:00
Desmis
d7088f9261
Fixed bug if spot outside Preview
2017-02-23 16:14:51 +01:00
Desmis
456909192c
Provisory fixed bug spot references - but not outside preview
2017-02-23 10:40:18 +01:00
Alberto Griggio
d3a5a8ee96
started working on support for dynamic processing profiles
2017-02-22 18:47:00 +01:00
heckflosse
7d9e5765ba
Removed unused function and stopwatch
2017-02-22 14:50:10 +01:00
heckflosse
8e205afeed
Revert "Further cleanup and astyled lcp.*"
...
This reverts commit 1348ea06e4
2017-02-22 14:43:41 +01:00
Desmis
44824abd4d
replace M_PI by rtengine::RT_PI for locallab_dev
2017-02-22 07:22:57 +01:00
Desmis
b3cfb6e04c
merge with dev
2017-02-22 07:06:17 +01:00
heckflosse
1348ea06e4
Further cleanup and astyled lcp.*
2017-02-22 01:53:22 +01:00
Flössie
99309aa4ac
Preliminary cleanup for LCPModelCommon
2017-02-21 21:10:33 +01:00
Flössie
d2b4fe5e54
Merge branch 'std_cpp11-dev' into dev ( fixes #3635 )
2017-02-21 20:37:21 +01:00
heckflosse
87a280f8ca
Additional speedup for lcp vignette correction
2017-02-21 19:11:54 +01:00
Desmis
1f7ad78e25
merge with dev - I hope no error
2017-02-21 18:02:48 +01:00
heckflosse
94129861f5
Speedup for lcp vignetting correction
2017-02-19 20:36:45 +01:00
Ingo Weyrich
ac882d2ff2
Merge pull request #3703 from Beep6581/lcpvignette_fix
...
LCP vignetting correction only works with undemosaiced raw files. fix…
2017-02-19 17:39:01 +01:00
