[{"id":11529,"web_url":"https://patchwork.libcamera.org/comment/11529/","msgid":"<20200724011749.GI21353@pendragon.ideasonboard.com>","date":"2020-07-24T01:17:49","subject":"Re: [libcamera-devel] [PATCH 3/3] libcamera: qcam: Improve colour\n\tinformation in DNG files","submitter":{"id":2,"url":"https://patchwork.libcamera.org/api/people/2/","name":"Laurent Pinchart","email":"laurent.pinchart@ideasonboard.com"},"content":"Hi David,\n\nThank you for the patch.\n\nOn Sat, Jul 04, 2020 at 10:59:14AM +0100, David Plowman wrote:\n> This patch improves the colour information recorded in DNG files using\n> the ColourCorrectionMatrix metadata for the image. Note that we are\n> not supplying a full calibration using two illuminants, nonetheless\n> the single matrix here appears to be respected by a number of tools.\n> \n> Signed-off-by: David Plowman \n> ---\n> src/qcam/dng_writer.cpp | 93 +++++++++++++++++++++++++++++++++++++++++\n> 1 file changed, 93 insertions(+)\n> \n> diff --git a/src/qcam/dng_writer.cpp b/src/qcam/dng_writer.cpp\n> index 61505d3..222df9f 100644\n> --- a/src/qcam/dng_writer.cpp\n> +++ b/src/qcam/dng_writer.cpp\n> @@ -34,6 +34,63 @@ struct FormatInfo {\n> \t\t\t unsigned int stride);\n> };\n> \n> +struct Matrix {\n\nI'd name this Matrix3d as it's a 3x3 matrix.\n\n> +\tMatrix(float m0, float m1, float m2,\n> +\t float m3, float m4, float m5,\n> +\t float m6, float m7, float m8)\n> +\t{\n> +\t\tm[0] = m0, m[1] = m1, m[2] = m2;\n> +\t\tm[3] = m3, m[4] = m4, m[5] = m5;\n> +\t\tm[6] = m6, m[7] = m7, m[8] = m8;\n> +\t}\n\nPlease add blank lines between functions.\n\n> +\tMatrix(float diag0, float diag1, float diag2)\n> +\t\t: Matrix(diag0, 0, 0, 0, diag1, 0, 0, 0, diag2) {}\n\nAnd we tend to move the brackets to lines of their own after the member\ninitializer list.\n\nI wonder if this should be a static function though,\n\n\tstatic Matrix diag(float diag0, float diag1, float diag2)\n\t{\n\t\treturn { diag0, 0, 0, 0, diag1, 0, 0, 0, diag2 };\n\t}\n\nas using a custom constructor is a bit confusing. When I was reading the\ncode below:\n\n\tMatrix wbGain(1, 1, 1);\n\nI wondered how a 3x3 matrix would be constructed from 3 values only.\n\n\tMatrix wbGain = Matrix::diag(1, 1, 1);\n\nwould be more explicit. Another useful function would be\n\n\tMatrix identity()\n\t{\n\t\treturn { 1, 0, 0, 0, 1, 0, 0, 0, 1 };\n\t}\n\n> +\tMatrix() {}\n\nWe usually put the default constructor first.\n\n> +\tfloat m[9];\n\nWe usually put member data after member functions.\n\n> +\tMatrix transpose() const\n> +\t{\n> +\t\treturn Matrix(m[0], m[3], m[6], m[1], m[4], m[7], m[2], m[5], m[8]);\n\nYou can also write\n\n\t\treturn { m[0], m[3], m[6], m[1], m[4], m[7], m[2], m[5], m[8] };\n\nwhich will cause the compiler to warn in case of loss of precision due\nto implicit casts. Same for cofactors().\n\n> +\t}\n> +\tMatrix cofactors() const\n> +\t{\n> +\t\treturn Matrix(m[4] * m[8] - m[5] * m[7],\n> +\t\t\t -(m[3] * m[8] - m[5] * m[6]),\n> +\t\t\t m[3] * m[7] - m[4] * m[6],\n> +\t\t\t -(m[1] * m[8] - m[2] * m[7]),\n> +\t\t\t m[0] * m[8] - m[2] * m[6],\n> +\t\t\t -(m[0] * m[7] - m[1] * m[6]),\n> +\t\t\t m[1] * m[5] - m[2] * m[4],\n> +\t\t\t -(m[0] * m[5] - m[2] * m[3]),\n> +\t\t\t m[0] * m[4] - m[1] * m[3]);\n> +\t}\n> +\tMatrix adjugate() const { return cofactors().transpose(); }\n> +\tfloat determinant() const\n> +\t{\n> +\t\treturn (m[0] * (m[4] * m[8] - m[5] * m[7]) -\n> +\t\t\tm[1] * (m[3] * m[8] - m[5] * m[6]) +\n> +\t\t\tm[2] * (m[3] * m[7] - m[4] * m[6]));\n\nThe outer parentheses are not needed.\n\n> +\t}\n> +\tMatrix inverse() const { return adjugate() * (1.0 / determinant()); }\n\nShould we protect against determinant being 0 ?\n\n> +\tMatrix operator*(Matrix const &other) const\n> +\t{\n> +\t\tMatrix result;\n> +\t\tfor (int i = 0; i < 3; i++)\n\nunsigned int i\n\nEven though not strictly necessary, we use curly braces when the inner\nstatement is a compound statement.\n\n> +\t\t\tfor (int j = 0; j < 3; j++)\n\nunsigned int j\n\n> +\t\t\t\tresult.m[i * 3 + j] =\n> +\t\t\t\t\tm[i * 3 + 0] * other.m[0 + j] +\n> +\t\t\t\t\tm[i * 3 + 1] * other.m[3 + j] +\n> +\t\t\t\t\tm[i * 3 + 2] * other.m[6 + j];\n> +\t\treturn result;\n> +\t}\n> +\tMatrix operator*(float const &f) const\n\nI think \"float f\" would be fine, is there a need for a reference ?\n\n> +\t{\n> +\t\tMatrix result;\n> +\t\tfor (int i = 0; i < 9; i++)\n\nunsigned int i\n\n> +\t\t\tresult.m[i] = m[i] * f;\n> +\t\treturn result;\n> +\t}\n> +};\n> +\n> void packScanlineSBGGR10P(void *output, const void *input, unsigned int width)\n> {\n> \tconst uint8_t *in = static_cast(input);\n> @@ -315,6 +372,42 @@ int DNGWriter::write(const char *filename, const Camera *camera,\n> \tTIFFSetField(tif, TIFFTAG_PLANARCONFIG, PLANARCONFIG_CONTIG);\n> \tTIFFSetField(tif, TIFFTAG_SAMPLEFORMAT, SAMPLEFORMAT_UINT);\n> \n> +\t/*\n> +\t * Fill in some reasonable colour information in the DNG. We supply\n> +\t * the \"neutral\" colour values which determine the white balance, and the\n> +\t * \"ColorMatrix1\" which converts XYZ to (un-white-balanced) camera RGB.\n> +\t * Note that this is not a \"proper\" colour calibration for the DNG,\n> +\t * nonetheless, many tools should be able to render the colours better.\n> +\t */\n> +\tfloat neutral[3] = { 1, 1, 1 };\n> +\tMatrix wbGain(1, 1, 1);\n> +\t/* From http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html */\n\nhttps would be better, but it seems that the server doesn't support it\n:-(\n\n> +\tMatrix rgb2xyz(0.4124564, 0.3575761, 0.1804375,\n> +\t\t 0.2126729, 0.7151522, 0.0721750,\n> +\t\t 0.0193339, 0.1191920, 0.9503041);\n\nThis should be const.\n\n> +\tMatrix ccm(1, 1, 1);\n> +\tconst double eps = 1e-2;\n\nOut of curiosity, why 1e-2 ?\n\n> +\n> +\tif (metadata.contains(controls::ColourGains)) {\n> +\t\tSpan colour_gains = metadata.get(controls::ColourGains);\n\ns/colour_gains/colourGains/ or just s/colour_gains/gains/\n\n> +\t\tif (colour_gains[0] > eps && colour_gains[1] > eps) {\n> +\t\t\twbGain = Matrix(colour_gains[0], 1, colour_gains[1]);\n> +\t\t\tneutral[0] = 1.0 / colour_gains[0]; /* red */\n> +\t\t\tneutral[2] = 1.0 / colour_gains[1]; /* blue */\n> +\t\t}\n> +\t}\n> +\tif (metadata.contains(controls::ColourCorrectionMatrix)) {\n> +\t\tSpan m = metadata.get(controls::ColourCorrectionMatrix);\n> +\t\tMatrix tmp = Matrix(m[0], m[1], m[2], m[3], m[4], m[5], m[6], m[7], m[8]);\n\nWould a\n\n\tMatrix(Span m)\n\nconstructor make sense to simplify this line ? Up to you.\n\nWe really try not to use \"tmp\" as a variable name. Possible alternatives\nare \"matrix\", or just \"m\" if you rename the existing \"m\" to \"value\".\n\n> +\t\tif (tmp.determinant() > eps)\n> +\t\t\tccm = tmp;\n> +\t}\n\nPlease add a blank line here.\n\n> +\t/* This is guaranteed to be invertible because all the bits in it are. */\n\nI'd expand this a little:\n\n\t/*\n\t * This is guaranteed to be invertible because all the bits in it are\n\t * (rgb2xyz is hardcoded and invertible, the ccm determinant is checked\n\t * manually above, and wbGain is a diagonal matrix with diagonal\n\t * elements checked to be non-zero).\n\t */\n\n> +\tMatrix colorMatrix1 = (rgb2xyz * ccm * wbGain).inverse();\n> +\n> +\tTIFFSetField(tif, TIFFTAG_COLORMATRIX1, 9, colorMatrix1.m);\n> +\tTIFFSetField(tif, TIFFTAG_ASSHOTNEUTRAL, 3, neutral);\n> +\n> \t/*\n> \t * Reserve space for the SubIFD and ExifIFD tags, pointing to the IFD\n> \t * for the raw image and EXIF data respectively. The real offsets will","headers":{"Return-Path":"","X-Original-To":"parsemail@patchwork.libcamera.org","Delivered-To":"parsemail@patchwork.libcamera.org","Received":["from lancelot.ideasonboard.com (lancelot.ideasonboard.com\n\t[92.243.16.209])\n\tby patchwork.libcamera.org (Postfix) with ESMTPS id EE22EBD878\n\tfor ;\n\tFri, 24 Jul 2020 01:17:57 +0000 (UTC)","from lancelot.ideasonboard.com (localhost [IPv6:::1])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTP id 7840161167;\n\tFri, 24 Jul 2020 03:17:57 +0200 (CEST)","from perceval.ideasonboard.com (perceval.ideasonboard.com\n\t[213.167.242.64])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTPS id F25CF60539\n\tfor ;\n\tFri, 24 Jul 2020 03:17:55 +0200 (CEST)","from pendragon.ideasonboard.com (81-175-216-236.bb.dnainternet.fi\n\t[81.175.216.236])\n\tby perceval.ideasonboard.com (Postfix) with ESMTPSA id 7226E279;\n\tFri, 24 Jul 2020 03:17:55 +0200 (CEST)"],"Authentication-Results":"lancelot.ideasonboard.com;\n\tdkim=fail reason=\"signature verification failed\" (1024-bit key;\n\tunprotected) header.d=ideasonboard.com header.i=@ideasonboard.com\n\theader.b=\"tfGJzCxz\"; dkim-atps=neutral","DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/simple; d=ideasonboard.com;\n\ts=mail; t=1595553475;\n\tbh=GRPy2CzTzlx9uR3gFD7MD0q9ikTr6p2MxF+zdFYPSr0=;\n\th=Date:From:To:Cc:Subject:References:In-Reply-To:From;\n\tb=tfGJzCxzLy7KIS+RcVkjqjTxWRIUwG9vRob7fePd3Zhsj4Ls/0VkSGi58L82tynnf\n\tEdo20VxAlHQf3H5Xgi+YyxrIeaPxITrJUd6179jiuT5qr3eZ6336nIHPnHtzf2krgb\n\tJTeBxh4SEmlgE5iSLF9c6NAdqqAacfe5sMNFZxbI=","Date":"Fri, 24 Jul 2020 04:17:49 +0300","From":"Laurent Pinchart ","To":"David Plowman ","Message-ID":"<20200724011749.GI21353@pendragon.ideasonboard.com>","References":"<20200704095914.17344-1-david.plowman@raspberrypi.com>\n\t<20200704095914.17344-4-david.plowman@raspberrypi.com>","MIME-Version":"1.0","Content-Disposition":"inline","In-Reply-To":"<20200704095914.17344-4-david.plowman@raspberrypi.com>","Subject":"Re: [libcamera-devel] [PATCH 3/3] libcamera: qcam: Improve colour\n\tinformation in DNG files","X-BeenThere":"libcamera-devel@lists.libcamera.org","X-Mailman-Version":"2.1.29","Precedence":"list","List-Id":"","List-Unsubscribe":",\n\t","List-Archive":"","List-Post":"","List-Help":"","List-Subscribe":",\n\t","Cc":"libcamera-devel@lists.libcamera.org","Content-Type":"text/plain; charset=\"us-ascii\"","Content-Transfer-Encoding":"7bit","Errors-To":"libcamera-devel-bounces@lists.libcamera.org","Sender":"\"libcamera-devel\" "}},{"id":11534,"web_url":"https://patchwork.libcamera.org/comment/11534/","msgid":"","date":"2020-07-24T07:24:47","subject":"Re: [libcamera-devel] [PATCH 3/3] libcamera: qcam: Improve colour\n\tinformation in DNG files","submitter":{"id":42,"url":"https://patchwork.libcamera.org/api/people/42/","name":"David Plowman","email":"david.plowman@raspberrypi.com"},"content":"Hi Laurent\n\nThanks for the review of the various patches in this set. Let me roll\nall that up into a v2 and I'll send those out shortly.\n\nBest regards\nDavid\n\nOn Fri, 24 Jul 2020 at 02:17, Laurent Pinchart\n wrote:\n>\n> Hi David,\n>\n> Thank you for the patch.\n>\n> On Sat, Jul 04, 2020 at 10:59:14AM +0100, David Plowman wrote:\n> > This patch improves the colour information recorded in DNG files using\n> > the ColourCorrectionMatrix metadata for the image. Note that we are\n> > not supplying a full calibration using two illuminants, nonetheless\n> > the single matrix here appears to be respected by a number of tools.\n> >\n> > Signed-off-by: David Plowman \n> > ---\n> > src/qcam/dng_writer.cpp | 93 +++++++++++++++++++++++++++++++++++++++++\n> > 1 file changed, 93 insertions(+)\n> >\n> > diff --git a/src/qcam/dng_writer.cpp b/src/qcam/dng_writer.cpp\n> > index 61505d3..222df9f 100644\n> > --- a/src/qcam/dng_writer.cpp\n> > +++ b/src/qcam/dng_writer.cpp\n> > @@ -34,6 +34,63 @@ struct FormatInfo {\n> > unsigned int stride);\n> > };\n> >\n> > +struct Matrix {\n>\n> I'd name this Matrix3d as it's a 3x3 matrix.\n>\n> > + Matrix(float m0, float m1, float m2,\n> > + float m3, float m4, float m5,\n> > + float m6, float m7, float m8)\n> > + {\n> > + m[0] = m0, m[1] = m1, m[2] = m2;\n> > + m[3] = m3, m[4] = m4, m[5] = m5;\n> > + m[6] = m6, m[7] = m7, m[8] = m8;\n> > + }\n>\n> Please add blank lines between functions.\n>\n> > + Matrix(float diag0, float diag1, float diag2)\n> > + : Matrix(diag0, 0, 0, 0, diag1, 0, 0, 0, diag2) {}\n>\n> And we tend to move the brackets to lines of their own after the member\n> initializer list.\n>\n> I wonder if this should be a static function though,\n>\n> static Matrix diag(float diag0, float diag1, float diag2)\n> {\n> return { diag0, 0, 0, 0, diag1, 0, 0, 0, diag2 };\n> }\n>\n> as using a custom constructor is a bit confusing. When I was reading the\n> code below:\n>\n> Matrix wbGain(1, 1, 1);\n>\n> I wondered how a 3x3 matrix would be constructed from 3 values only.\n>\n> Matrix wbGain = Matrix::diag(1, 1, 1);\n>\n> would be more explicit. Another useful function would be\n>\n> Matrix identity()\n> {\n> return { 1, 0, 0, 0, 1, 0, 0, 0, 1 };\n> }\n>\n> > + Matrix() {}\n>\n> We usually put the default constructor first.\n>\n> > + float m[9];\n>\n> We usually put member data after member functions.\n>\n> > + Matrix transpose() const\n> > + {\n> > + return Matrix(m[0], m[3], m[6], m[1], m[4], m[7], m[2], m[5], m[8]);\n>\n> You can also write\n>\n> return { m[0], m[3], m[6], m[1], m[4], m[7], m[2], m[5], m[8] };\n>\n> which will cause the compiler to warn in case of loss of precision due\n> to implicit casts. Same for cofactors().\n>\n> > + }\n> > + Matrix cofactors() const\n> > + {\n> > + return Matrix(m[4] * m[8] - m[5] * m[7],\n> > + -(m[3] * m[8] - m[5] * m[6]),\n> > + m[3] * m[7] - m[4] * m[6],\n> > + -(m[1] * m[8] - m[2] * m[7]),\n> > + m[0] * m[8] - m[2] * m[6],\n> > + -(m[0] * m[7] - m[1] * m[6]),\n> > + m[1] * m[5] - m[2] * m[4],\n> > + -(m[0] * m[5] - m[2] * m[3]),\n> > + m[0] * m[4] - m[1] * m[3]);\n> > + }\n> > + Matrix adjugate() const { return cofactors().transpose(); }\n> > + float determinant() const\n> > + {\n> > + return (m[0] * (m[4] * m[8] - m[5] * m[7]) -\n> > + m[1] * (m[3] * m[8] - m[5] * m[6]) +\n> > + m[2] * (m[3] * m[7] - m[4] * m[6]));\n>\n> The outer parentheses are not needed.\n>\n> > + }\n> > + Matrix inverse() const { return adjugate() * (1.0 / determinant()); }\n>\n> Should we protect against determinant being 0 ?\n>\n> > + Matrix operator*(Matrix const &other) const\n> > + {\n> > + Matrix result;\n> > + for (int i = 0; i < 3; i++)\n>\n> unsigned int i\n>\n> Even though not strictly necessary, we use curly braces when the inner\n> statement is a compound statement.\n>\n> > + for (int j = 0; j < 3; j++)\n>\n> unsigned int j\n>\n> > + result.m[i * 3 + j] =\n> > + m[i * 3 + 0] * other.m[0 + j] +\n> > + m[i * 3 + 1] * other.m[3 + j] +\n> > + m[i * 3 + 2] * other.m[6 + j];\n> > + return result;\n> > + }\n> > + Matrix operator*(float const &f) const\n>\n> I think \"float f\" would be fine, is there a need for a reference ?\n>\n> > + {\n> > + Matrix result;\n> > + for (int i = 0; i < 9; i++)\n>\n> unsigned int i\n>\n> > + result.m[i] = m[i] * f;\n> > + return result;\n> > + }\n> > +};\n> > +\n> > void packScanlineSBGGR10P(void *output, const void *input, unsigned int width)\n> > {\n> > const uint8_t *in = static_cast(input);\n> > @@ -315,6 +372,42 @@ int DNGWriter::write(const char *filename, const Camera *camera,\n> > TIFFSetField(tif, TIFFTAG_PLANARCONFIG, PLANARCONFIG_CONTIG);\n> > TIFFSetField(tif, TIFFTAG_SAMPLEFORMAT, SAMPLEFORMAT_UINT);\n> >\n> > + /*\n> > + * Fill in some reasonable colour information in the DNG. We supply\n> > + * the \"neutral\" colour values which determine the white balance, and the\n> > + * \"ColorMatrix1\" which converts XYZ to (un-white-balanced) camera RGB.\n> > + * Note that this is not a \"proper\" colour calibration for the DNG,\n> > + * nonetheless, many tools should be able to render the colours better.\n> > + */\n> > + float neutral[3] = { 1, 1, 1 };\n> > + Matrix wbGain(1, 1, 1);\n> > + /* From http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html */\n>\n> https would be better, but it seems that the server doesn't support it\n> :-(\n>\n> > + Matrix rgb2xyz(0.4124564, 0.3575761, 0.1804375,\n> > + 0.2126729, 0.7151522, 0.0721750,\n> > + 0.0193339, 0.1191920, 0.9503041);\n>\n> This should be const.\n>\n> > + Matrix ccm(1, 1, 1);\n> > + const double eps = 1e-2;\n>\n> Out of curiosity, why 1e-2 ?\n>\n> > +\n> > + if (metadata.contains(controls::ColourGains)) {\n> > + Span colour_gains = metadata.get(controls::ColourGains);\n>\n> s/colour_gains/colourGains/ or just s/colour_gains/gains/\n>\n> > + if (colour_gains[0] > eps && colour_gains[1] > eps) {\n> > + wbGain = Matrix(colour_gains[0], 1, colour_gains[1]);\n> > + neutral[0] = 1.0 / colour_gains[0]; /* red */\n> > + neutral[2] = 1.0 / colour_gains[1]; /* blue */\n> > + }\n> > + }\n> > + if (metadata.contains(controls::ColourCorrectionMatrix)) {\n> > + Span m = metadata.get(controls::ColourCorrectionMatrix);\n> > + Matrix tmp = Matrix(m[0], m[1], m[2], m[3], m[4], m[5], m[6], m[7], m[8]);\n>\n> Would a\n>\n> Matrix(Span m)\n>\n> constructor make sense to simplify this line ? Up to you.\n>\n> We really try not to use \"tmp\" as a variable name. Possible alternatives\n> are \"matrix\", or just \"m\" if you rename the existing \"m\" to \"value\".\n>\n> > + if (tmp.determinant() > eps)\n> > + ccm = tmp;\n> > + }\n>\n> Please add a blank line here.\n>\n> > + /* This is guaranteed to be invertible because all the bits in it are. */\n>\n> I'd expand this a little:\n>\n> /*\n> * This is guaranteed to be invertible because all the bits in it are\n> * (rgb2xyz is hardcoded and invertible, the ccm determinant is checked\n> * manually above, and wbGain is a diagonal matrix with diagonal\n> * elements checked to be non-zero).\n> */\n>\n> > + Matrix colorMatrix1 = (rgb2xyz * ccm * wbGain).inverse();\n> > +\n> > + TIFFSetField(tif, TIFFTAG_COLORMATRIX1, 9, colorMatrix1.m);\n> > + TIFFSetField(tif, TIFFTAG_ASSHOTNEUTRAL, 3, neutral);\n> > +\n> > /*\n> > * Reserve space for the SubIFD and ExifIFD tags, pointing to the IFD\n> > * for the raw image and EXIF data respectively. The real offsets will\n>\n> --\n> Regards,\n>\n> Laurent Pinchart","headers":{"Return-Path":"","X-Original-To":"parsemail@patchwork.libcamera.org","Delivered-To":"parsemail@patchwork.libcamera.org","Received":["from lancelot.ideasonboard.com (lancelot.ideasonboard.com\n\t[92.243.16.209])\n\tby patchwork.libcamera.org (Postfix) with ESMTPS id 473D2BD86F\n\tfor ;\n\tFri, 24 Jul 2020 07:25:03 +0000 (UTC)","from lancelot.ideasonboard.com (localhost [IPv6:::1])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTP id C3EEF61167;\n\tFri, 24 Jul 2020 09:25:02 +0200 (CEST)","from mail-oi1-x242.google.com (mail-oi1-x242.google.com\n\t[IPv6:2607:f8b0:4864:20::242])\n\tby lancelot.ideasonboard.com (Postfix) with ESMTPS id 7DDC56093B\n\tfor ;\n\tFri, 24 Jul 2020 09:25:01 +0200 (CEST)","by mail-oi1-x242.google.com with SMTP id k6so7259591oij.11\n\tfor ;\n\tFri, 24 Jul 2020 00:25:01 -0700 (PDT)"],"Authentication-Results":"lancelot.ideasonboard.com;\n\tdkim=fail reason=\"signature verification failed\" (2048-bit key;\n\tunprotected) header.d=raspberrypi.com header.i=@raspberrypi.com\n\theader.b=\"mH23ek1e\"; dkim-atps=neutral","DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/relaxed;\n\td=raspberrypi.com; s=google;\n\th=mime-version:references:in-reply-to:from:date:message-id:subject:to\n\t:cc; bh=DE+twLUc3YC2+OEfdychoOagDwx0xisjwp/jbzq9xoo=;\n\tb=mH23ek1evndGXZU6QjunLT8jIF6aMCyYHe516XP5ikeSto2c7oNGSOdd3YqEiP3Chx\n\trkcaBZaAiXtVxKw7OBReqFuGmdb0okYA1rurMXtlPgcZNyyuFuJrcOVw7wwV3lPxZ6Lw\n\tmDJPqOqeJIQNwjGZ+UvVsGTkV6BEBbbnv5Zoj37aCDT2DHFo79hk1It5jPKlzwcEGbO+\n\tpISL6QlJET0QMkiimHObuQHZ3OVMhkeaakf40mFUCOA4LSj3JlzeL4QQG0MwEgnTr1q/\n\tsCqdFn6UnDjhUqcPL97Bp2CB+zxLjyQw6VzK5Zn7GIG7ZC3wOw6zD3HBpeZ9pBlW1/7g\n\t/atQ==","X-Google-DKIM-Signature":"v=1; a=rsa-sha256; c=relaxed/relaxed;\n\td=1e100.net; s=20161025;\n\th=x-gm-message-state:mime-version:references:in-reply-to:from:date\n\t:message-id:subject:to:cc;\n\tbh=DE+twLUc3YC2+OEfdychoOagDwx0xisjwp/jbzq9xoo=;\n\tb=APfBH6qNYqUMdxyEDHMXmp1TzwauhL+bnE1nmjx3LBYgTFomrTE0BIo+YJ+aWy6NGh\n\t4Po6xg1Xz4U+ruSCl52J7ucs0126D9PR9OaVAZ29bFePvUUFdHfd7TrVGPEXLRnp0z7Z\n\tO/VCmqiDFXWbN1epKGPvEV1afesHFWIj3bunQYcBrjkfclVE4WSK+3CaHvIiJLAOCDZC\n\tTa/mfiD/gNnGjtOomQIqq//5aOapw+A7dIUC4+zgTLi5lRB4299k0g3FD5rRswR/wEy8\n\tlG518+Un5Q9bfjVEbNeKBhdYLbKC3v9dmjqwEHmcIpwkNjgXzUr1+X0fH3bo+cC5Q1b0\n\t9Byg==","X-Gm-Message-State":"AOAM5311TJlZZ5dKeKGXz2PHHEyDNr/wsnwjDAVnNFQudlw2aWj2Y0S+\n\tNjbsvJUMUG6QeXTb0J1xw4lNAVC7wmRvYqZoD8UmwlBYz8k=","X-Google-Smtp-Source":"ABdhPJyDVSumq2NJQDEMkT49mUGw8BpelFDcPo+oRvQXs9apQuPUdV01msNNPwjQWG83qI+OChzbEs9PDQIBqiDPPk8=","X-Received":"by 2002:aca:ac01:: with SMTP id v1mr6750610oie.22.1595575499941; \n\tFri, 24 Jul 2020 00:24:59 -0700 (PDT)","MIME-Version":"1.0","References":"<20200704095914.17344-1-david.plowman@raspberrypi.com>\n\t<20200704095914.17344-4-david.plowman@raspberrypi.com>\n\t<20200724011749.GI21353@pendragon.ideasonboard.com>","In-Reply-To":"<20200724011749.GI21353@pendragon.ideasonboard.com>","From":"David Plowman ","Date":"Fri, 24 Jul 2020 08:24:47 +0100","Message-ID":"","To":"Laurent Pinchart ","Subject":"Re: [libcamera-devel] [PATCH 3/3] libcamera: qcam: Improve colour\n\tinformation in DNG files","X-BeenThere":"libcamera-devel@lists.libcamera.org","X-Mailman-Version":"2.1.29","Precedence":"list","List-Id":"","List-Unsubscribe":",\n\t","List-Archive":"","List-Post":"","List-Help":"","List-Subscribe":",\n\t","Cc":"libcamera-devel@lists.libcamera.org","Content-Type":"text/plain; charset=\"us-ascii\"","Content-Transfer-Encoding":"7bit","Errors-To":"libcamera-devel-bounces@lists.libcamera.org","Sender":"\"libcamera-devel\" "}}]