diff mbox

[FFmpeg-devel] avutil/colorspace: add macros for RGB->YUV BT.709

Message ID c0b4aca9-4939-ccf4-541e-6d22147d8533@gyani.pro
State Accepted
Commit 3bef1dab6e881bf593037f831e66ed87c6167e3c
Headers show

Commit Message

Gyan Doshi April 18, 2019, 7:34 p.m. UTC 
On 17-04-2019 11:30 PM, Gyan wrote:
>
>
> On 17-04-2019 11:11 PM, Vittorio Giovara wrote:
>> On Wed, Apr 17, 2019 at 12:26 AM Gyan <ffmpeg@gyani.pro> wrote:
>>
>>>
>>> On 13-04-2019 05:23 PM, Gyan wrote:
>>>> Will be helpful for correct result in filters that paint like
>>>> fillborders/drawbox or those using drawutils.
>>> Ping.
>>>
>> these seem to only work for 8 bit content, is that their only intended
>> usecase?
>> if so, could there be at least a comment mentioning that please?
> Yes, the existing macros use Rec.601 co-efficients, but most data is 
> 709 now.
> Will add a note.

With note.
From 5443848d974e98f05b5601e9f1f0d0fd54aa54b8 Mon Sep 17 00:00:00 2001
From: Gyan Doshi <ffmpeg@gyani.pro>
Date: Sat, 13 Apr 2019 17:01:09 +0530
Subject: [PATCH v2] avutil/colorspace: add macros for RGB->YUV BT.709

---
 libavutil/colorspace.h | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

Comments

Gyan Doshi April 19, 2019, 4:27 a.m. UTC | #1 
On 19-04-2019 01:04 AM, Gyan wrote:
>
>
> On 17-04-2019 11:30 PM, Gyan wrote:
>>
>>
>> On 17-04-2019 11:11 PM, Vittorio Giovara wrote:
>>> On Wed, Apr 17, 2019 at 12:26 AM Gyan <ffmpeg@gyani.pro> wrote:
>>>
>>>>
>>>> On 13-04-2019 05:23 PM, Gyan wrote:
>>>>> Will be helpful for correct result in filters that paint like
>>>>> fillborders/drawbox or those using drawutils.
>>>> Ping.
>>>>
>>> these seem to only work for 8 bit content, is that their only intended
>>> usecase?
>>> if so, could there be at least a comment mentioning that please?
>> Yes, the existing macros use Rec.601 co-efficients, but most data is 
>> 709 now.
>> Will add a note.
>
> With note.

Will apply soon.

Gyan
Gyan Doshi April 19, 2019, 11:42 a.m. UTC | #2 
On 19-04-2019 09:57 AM, Gyan wrote:
>
>
> On 19-04-2019 01:04 AM, Gyan wrote:
>>
>>
>> On 17-04-2019 11:30 PM, Gyan wrote:
>>>
>>>
>>> On 17-04-2019 11:11 PM, Vittorio Giovara wrote:
>>>> On Wed, Apr 17, 2019 at 12:26 AM Gyan <ffmpeg@gyani.pro> wrote:
>>>>
>>>>>
>>>>> On 13-04-2019 05:23 PM, Gyan wrote:
>>>>>> Will be helpful for correct result in filters that paint like
>>>>>> fillborders/drawbox or those using drawutils.
>>>>> Ping.
>>>>>
>>>> these seem to only work for 8 bit content, is that their only intended
>>>> usecase?
>>>> if so, could there be at least a comment mentioning that please?
>>> Yes, the existing macros use Rec.601 co-efficients, but most data is 
>>> 709 now.
>>> Will add a note.
>>
>> With note.
>
> Will apply soon.

Pushed as 3bef1dab6e881bf593037f831e66ed87c6167e3c

Gyan
diff mbox

Patch

diff --git a/libavutil/colorspace.h b/libavutil/colorspace.h
index d0be8cb99a..ef6f6107d6 100644
--- a/libavutil/colorspace.h
+++ b/libavutil/colorspace.h
@@ -119,4 +119,32 @@  static inline int C_JPEG_TO_CCIR(int y) {
 (((FIX(0.50000) * r1 - FIX(0.41869) * g1 - \
    FIX(0.08131) * b1 + (ONE_HALF) - 1) >> (SCALEBITS)) + 128)
 
+// Conversion macros for 8-bit RGB to YUV
+// Derived from ITU-R BT.709-6 (06/2015) Item 3.5
+// https://www.itu.int/rec/R-REC-BT.709-6-201506-I/en
+
+#define RGB_TO_Y_BT709(r, g, b) \
+((FIX(0.21260*219.0/255.0) * (r) + FIX(0.71520*219.0/255.0) * (g) + \
+  FIX(0.07220*219.0/255.0) * (b) + (ONE_HALF + (16 << SCALEBITS))) >> SCALEBITS)
+
+#define RGB_TO_U_BT709(r1, g1, b1, shift)\
+(((- FIX(0.11457*224.0/255.0) * r1 - FIX(0.38543*224.0/255.0) * g1 +         \
+     FIX(0.50000*224.0/255.0) * b1 + (ONE_HALF << shift) - 1) >> (SCALEBITS + shift)) + 128)
+
+#define RGB_TO_V_BT709(r1, g1, b1, shift)\
+(((FIX(0.50000*224.0/255.0) * r1 - FIX(0.45415*224.0/255.0) * g1 -           \
+   FIX(0.04585*224.0/255.0) * b1 + (ONE_HALF << shift) - 1) >> (SCALEBITS + shift)) + 128)
+
+#define RGB_TO_Y_BT709_FULL(r, g, b) \
+(FFMIN((FIX(0.21260) * (r) + FIX(0.71520) * (g) + \
+  FIX(0.07220) * (b) + (ONE_HALF)) >> SCALEBITS, 255))
+
+#define RGB_TO_U_BT709_FULL(r1, g1, b1)\
+(((- FIX(0.11457) * r1 - FIX(0.38543) * g1 + \
+     FIX(0.50000) * b1 + (ONE_HALF) - 1) >> (SCALEBITS)) + 128)
+
+#define RGB_TO_V_BT709_FULL(r1, g1, b1)\
+(((FIX(0.50000) * r1 - FIX(0.45415) * g1 - \
+   FIX(0.04585) * b1 + (ONE_HALF) - 1) >> (SCALEBITS)) + 128)
+
 #endif /* AVUTIL_COLORSPACE_H */