Changelog for Version 8

Version	Date	Updates
8.3.0	Feb 22, 2019	• Added PFCImageFile for EXIF and Color Management • Added image resizing capabilities
8.2.0	Oct 4, 2018	• Added preset file parsing for the .NET adapter, Android and iOS platforms • Header file change to add function to read presets from file stream
8.1.1	Sep 26, 2018	• Provide both AVX accelerated and non-AVX libraries by default in all SDK deliveries
8.1.0	Sep 19, 2018	• Added preset file parsing support (changes header file) • Fixed issue with very low Dynamic Range issues that could cause black pixels in the output image • Android: Sample app: fix to save corrected image • Android: proper auto-strength implementation • Fixing issue with very bright images that would result in low contrast output
8.0.3	Sep 6, 2018	• Fixed issue when iVibrancy = 0 on high-key images or photos with while borders • Win: removed dependencies on Visual Studio redistributables • Fixed incorrect version numbers on .NET Adapter .DLL file
8.0.2	Aug 16, 2018	• Better contrast correction on very small images • change iOutputVibrancy to iColorVibrancy to match the Workbench User Interface • Fixed Android sample application crash • Fixed Win Sample 2 crash on Save • Fixed issue that could cause a 1 pixel of bad data at bottom of corrected images

Upgrades Since Version 7

This version requires updating the header file, library, and optional code changes in your applications if you are using AutoCorrect.

Dramatic speed improvement: several options and internal optimizations have been put in place to reduce the Calc, Apply, and AutoCorrect processing times dramatically. On computers and servers, you should expect about a 3x speed improvement when compared to v7.4. The total processing time depends on many factors: image size, correction parameters enabled, and computer characteristics, so exact speed improvements cannot be stated exactly. The following features and improvements account for the speed gains:
1. Face Aware Exposure can now be run on a larger (slower) or smaller (faster) image. This is controlled through the PFCFEATURE enum when using separate Calc and Apply function calls, or through an additional function parameter when calling AutoCorrect.
2. Tint Calculation can optionally be skipped when tint correction is not enabled. Tint detection is performed in the 'Calc' phase, and if you allow the user to enable or disable Tint correction, then it is important to perform Tint detection regardless of the user's choice; this allows Tint to be toggled on or off without needing to 'Calc' a second time. However, if your implementation of Perfectly Clear does not offer Tint correction as a user-facing option, then you can gain a speed improvement by disabling tint detection. Starting in v8.0, Tint detection is disabled when using AutoCorrect. You may choose to disable this using the PFCFEATURE enum when calling Calc and Apply separately.
3. Face detection for the purpose of 'Beautify' corrections is now slightly more accurate. Starting in v8.0, the SDK will default to the same detection rate as our Workbench application.
4. AVX / NEON code optimization. A substantial speed gain was achieved by enabling AVX instruction set support for desktop/server processors and NEON support for ARM processors, which is present in nearly all CPU's produced since 2011 by AMD (Bulldozer and later) and Intel (Sandy Bridge and later). Libraries with and without AVX support are now provided in all SDK deliveries, so be certain to use the proper ones.
V3 Corrections support: Our "V3" products introduced numerous corrections that have not been available in our SDK until now. These corrections include:
1. Finishing & Graduated Filters
2. Image Ambulance (manual exposure correction)
3. Color Vibrancy (not to be confused with Color Restoration - which was previously labelled "Color Vibrancy" in Athentech's products.
4. Foliage & Sky Enhancement
5. Input and Output LOOKS - these are packaged and licensed separately from the rest of the Perfectly Clear SDK
Sharpening Update: In this SDK, we have improved the sharpening algorithm to be less prone to visible artifacts when applied at higher strengths. The maximum sharpening strength has been increased as well. Presets that used the previous sharpening algorithm should not need to be adjusted - previous sharpening levels will produce similar "strength" of sharpening.