liz/rawTherapee
liz/rawTherapee
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
7,670 Commits 1 Branch 0 Tags
Commit Graph

2281 Commits

Author SHA1 Message Date
rom9
938fc63dd0 Replaced powf calls with the optimized pow_F macro (which also works for the non-SSE2 case). 2019-06-27 23:01:51 +02:00
rom9
a3004bb2b8 Added SSE2 optimization in film negative thumbnail processing. 2019-06-27 22:03:59 +02:00
rom9
157d83d169 Partially reverting my last commit. When using large exponents, huge raw values could cause segfaults in ImProcFunctions:rgbProc (specifically in shadowToneCurve), and in Amaze demosaic. 
Re-added the clipping check on raw values, but with a higher threshold. This way, raw clipping is less frequent in highlight areas (so those can be used as spots for exponents calculation), but the threshold is low enough to avoid the segfaults.
Not a very clean solution, need to find a better one...
 2019-06-26 00:14:03 +02:00
rom9
22f6297a5b Removed clipping check for values above 65535 (reverts commit 9156572). It was just a defensive check to avoid hitting corner cases like this: 1e4f9ac , that is already fixed elsewhere. 
The check is now causing problems with the dual-spot feature, because after clipping, the formula cannot be undone to get back the original raw values.
Since there are no known issues caused by raw values >65k, i'm removing the check until there will be a good reason to re-introduce it.
 2019-06-24 21:48:06 +02:00
rom9
7040378dec Merge remote-tracking branch 'origin/dev' into filmnegative 2019-06-23 23:40:56 +02:00
rom9
2fde6e562a The "dual-spot" exponents calculation feature now honors the master (green) exponent set by the user, and adjusts red and blue based on the ratios between the new, "guessed" exponents. 
This way, if the user has chosen a different master exponent than the default, using the feature just alters the color characteristic of the conversion, while keeping the same contrast level.
 2019-06-23 22:21:24 +02:00
Ingo Weyrich
c6cdaf37eb
Merge pull request #5350 from Beep6581/badpixels-compunit 
Move badpixel code to own compilation unit
 2019-06-22 18:00:13 +02:00
rom9
880a6e125e Added film negative processing in thumbnails, with correct white balance. Calculated channel averages to restore the same initial conditions as RawImageSource, where get_colorsCoeff is called with forceAutoWB = true. Code still needs refinement and cleanup. 2019-06-21 07:12:42 +02:00
rom9
03dc855fc7 Merged cleanup patch proposed by heckflosse, moves exponents negation to a single place. Now it's much easier to read. 2019-06-18 21:08:18 +02:00
rom9
7c0275ca1a Updated copyright notice in new source files 2019-06-18 13:52:58 +02:00
Flössie
ca387e0379 Use terser C++11 resets in ProcParams::setDefaults() 2019-06-18 09:22:45 +02:00
Flössie
b687ca2643 Merge branch 'rom9-filmnegative' into filmnegative-cleanup 2019-06-17 08:57:14 +02:00
Flössie
20fdba77e8 Merge branch 'filmnegative' of https://github.com/rom9/RawTherapee into rom9-filmnegative 2019-06-17 08:07:32 +02:00
Flössie
54cc02eea9 Filmenegative core cleanup 2019-06-17 08:03:46 +02:00
Flössie
80f2b6a002 Replace raw arrays with std::array<> 2019-06-14 08:58:04 +02:00
Flössie
015cffc73a Peripheral cleanups and a fix 
- Whitespace, braces, sorting
- Fixed missing filmnegative in `ProcParams::operator ==(ProcParams)`
 2019-06-13 09:52:13 +02:00
rom9
c819cb63a2 Speed boost in the median vectors fill via loop unrolling 2019-06-12 22:11:34 +02:00
Morgan Hardwood
a686a94a62 Merge branch 'filmnegative' of https://github.com/rom9/RawTherapee into rom9-filmnegative 2019-06-12 09:19:49 +02:00
rom9
8d0755eddc Raised default master exponent from 1.0 to 2.0, to get a reasonable contrast right from the start. 2019-06-11 22:50:36 +02:00
rom9
8fa30d496d Now that medians are known *before* processing, moved multiplication step inside the same loop as exponentiation for further optimization. Patch kindly provided by @heckflosse ;-) 2019-06-11 21:02:16 +02:00
rom9
e1c9197ed5 Moved median calculation block before exponentiation. This way findMinMaxPercentile() can be used again because it works on the original raw file values (which are in a much more reasonable range). Patch kindly provided by @heckflosse ;-) 2019-06-11 20:08:46 +02:00
heckflosse
156f3009d5 badpixels code: further cleanups 2019-06-10 22:15:16 +02:00
rom9
b95bdb1aea Linked red and blue exponent adjuster to the green adjuster, in order to maintain the ratio between exponents if the user moves the green adjuster (master). 
Switched back to vector sort for median calculations: the results of the histogram search function diverge more and more from the simple median calculation as the exponents increase.
At 2.0 the test picture is already impossible to WB as the multipliers are too far off (2.78226e+08 histo vs 9.7927e+11 sort), and the normal WB sliders can't compensate for those huge factors.
 2019-06-10 22:05:54 +02:00
heckflosse
fe16bf7917 Move badpixel code to own compilation unit 2019-06-10 16:40:44 +02:00
rom9
91565728e5 Clamped output values to a max of 65535.f after applying multipliers, to avoid trouble further down the processing pipeline. 2019-06-09 19:25:15 +02:00
heckflosse
df4513f595 Avoid integer overflow when accessing luts with very large values 2019-06-09 18:57:23 +02:00
heckflosse
f0d32c1da7 White dot artifacts caused by CIECAM02, fixes #5342 2019-06-09 14:37:33 +02:00
heckflosse
1e4f9ac248 Avoid integer overflow when accessing luts with very large values 2019-06-09 14:36:38 +02:00
rom9
d2366e633c Enabled bad pixels interpolation for ST_XTRANS after upstream fix c0a033e 2019-06-09 13:09:29 +02:00
heckflosse
b990b89875 interpolateBadPixelsXtrans() : fix oob access 2019-06-09 12:21:42 +02:00
heckflosse
c0a033e717 interpolateBadPixelsXtrans() : fix oob access 2019-06-08 15:13:44 +02:00
rom9
9df8008949 Film negative processing: first usable version. Only supports bayer raw files, thumbnails don't work 
Added performance improvements suggested by heckflosse. Lowered median sampling step from 7 to 5 since calculation is now much faster.

Added support for Fuji X-Trans raw files.

Applied SSE2 patch provided by @heckflosse, improves performance in main processing loop.

Moved film negative processing stuff in its own compilation unit.

Code cleanup: removed redundant omp directives.

Added check for dead pixels, going above threshold after inversion. ST_BAYER only for now.

Reverted leftover hack in cropwindow.cc
 2019-06-08 01:40:29 +02:00
heckflosse
3b19b9f55b Rawtherapee 5.6 crashes with.dng from PixelShift2DNG-0.9.8.67, fixes #5348 2019-06-07 16:37:59 +02:00
heckflosse
6486c491f8 Vibrance causes RT to freeze, fixes #5346 2019-06-05 12:55:08 +02:00
heckflosse
3c0b0ffd33 Small speedup for loading unpacked 16 bit raw files 2019-06-04 17:01:05 +02:00
heckflosse
a391e256bc Fix possibly uninitialized variable 2019-06-04 16:30:54 +02:00
heckflosse
f18724e016 Silence some warnings 2019-06-04 16:30:22 +02:00
heckflosse
d46a043add basic support for Fujifilm GFX 100 2019-06-04 14:17:38 +02:00
luz.paz
cc60ea2248 Fix source comment and misc. typos 
Found via `codespell -q 3 -I ../rawtherapy-whitelist.txt -S ./rtdata/languages -L hist,fo,reall,bloc,alph`
 2019-06-03 17:38:50 -04:00
Desmis
18d977502d Improve Tone-mapping iterates 2019-05-19 18:12:46 +02:00
heckflosse
6958d46e70 Don't preprocess and demosaic again when using coarse rotation or flip, fixes #5327 2019-05-19 16:06:54 +02:00
Flössie
351f8f4e12 Fix indentation 2019-05-16 10:33:07 +02:00
Flössie
00b007f47a Fix compiler warning for non-SSE2 builds 2019-05-16 10:28:13 +02:00
heckflosse
141e9f632f Scan clut folder: Don't check for unused profile 2019-05-13 23:12:40 +02:00
Morgan Hardwood
a2e2ace1c8 Added colormatrix for samsung SM-N960F #5311 2019-04-28 17:15:00 +02:00
Morgan Hardwood
26f29a5e81 Added ColorMatrix for HUAWEI DLI-L22 2019-04-23 18:53:04 +02:00
Ingo Weyrich
73b336a3f1
Update FTblockDN.cc 
Disable StopWatch
 2019-04-18 20:43:40 +02:00
Ingo Weyrich
5d0b0d3d93
Merge pull request #5225 from Beep6581/denoise_speedup 
small speedup for denoise
 2019-04-17 22:46:18 +02:00
Hombre57
cb72769634 Solving issue #5267 : "ProcEvent int() operator should be const" 2019-04-09 08:45:21 +02:00
Morgan Hardwood
198fc8d8dd Added matrix for Canon PowerShot SX150 IS 2019-04-08 01:02:35 +02:00