From 860a89dcd261f97a172e8ae8a0075c5d71b15336 Mon Sep 17 00:00:00 2001
From: Anton Firszov <anton.firszov@sapagroup.com>
Date: Tue, 29 Aug 2017 14:26:21 +0200
Subject: [PATCH] comments on rounding logic

---
 .../GolangPort/Components/Decoder/JpegBlockPostProcessor.cs    | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/src/ImageSharp/Formats/Jpeg/GolangPort/Components/Decoder/JpegBlockPostProcessor.cs b/src/ImageSharp/Formats/Jpeg/GolangPort/Components/Decoder/JpegBlockPostProcessor.cs
index be03b5dd6..08cf1eb60 100644
--- a/src/ImageSharp/Formats/Jpeg/GolangPort/Components/Decoder/JpegBlockPostProcessor.cs
+++ b/src/ImageSharp/Formats/Jpeg/GolangPort/Components/Decoder/JpegBlockPostProcessor.cs
@@ -78,6 +78,9 @@ namespace SixLabors.ImageSharp.Formats.Jpeg.GolangPort.Components.Decoder
             this.data.ResultBlock.NormalizeColorsInplace();
             Size divs = component.SubSamplingDivisors;
 
+            // To conform better to libjpeg we actually NEED TO loose precision here.
+            // This is because they store blocks as Int16 between all the operations.
+            // Unfortunately, we need to emulate this to be "more accurate" :(
             this.data.ResultBlock.RoundInplace();
 
             this.data.ResultBlock.CopyTo(destArea, divs.Width, divs.Height);