From: "Niklas Söderlund" <niklas.soderlund@ragnatech.se>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Sakari Ailus <sakari.ailus@linux.intel.com>, linux-media@vger.kernel.org
Subject: Re: [raw2rgbpnm,PATCH] Add support for all RAW memory layouts
Date: Thu, 22 Jan 2026 10:28:34 +0100 [thread overview]
Message-ID: <20260122092834.GC2316109@ragnatech.se> (raw)
In-Reply-To: <20260122092004.GA239742@killaraus>
Hi Laurent,
On 2026-01-22 11:20:04 +0200, Laurent Pinchart wrote:
> Hi Niklas,
>
> I know this has been merged already, but it only came to my attention
> now.
>
> On Sun, May 11, 2025 at 05:46:59PM +0200, Niklas Söderlund wrote:
> > Convert all supported RAW input images to GRBG memory layout before
> > feeding it to the RAW to RGB conversion algorithms. This way all layouts
> > can produced good colors in the output image.
>
> I don't think the result will be accurate. It does produce better
> colours, but all liens and columns in the output image end up being
> swapped in groups of two. I would have kept printing a warning message.
Thinking about it now, yes having kept or updated the warning message to
make the user aware of the result not being pixel perfect would likely
have been a good idea.
>
> > Limit the conversion to input images that have an even width and height
> > to make the conversion easier.
> >
> > Images with and uneven width or hight can sill be processed, but will
> > be processed without first being converted to GRBG layout. Such input
> > images will result, just as before this change, in output images with
> > bad colors. The warning message about this have been preserved.
> >
> > Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
> > ---
> > raw2rgbpnm.c | 98 ++++++++++++++++++++++++++++++++++++++++++++++++++--
> > 1 file changed, 95 insertions(+), 3 deletions(-)
> >
> > diff --git a/raw2rgbpnm.c b/raw2rgbpnm.c
> > index 5838bc7347e9..b834add4ea6c 100644
> > --- a/raw2rgbpnm.c
> > +++ b/raw2rgbpnm.c
> > @@ -186,6 +186,97 @@ static unsigned char *read_raw_data(char *filename, int width, int height,
> > return b;
> > }
> >
> > +static int raw_layout_to_grbg(const struct format_info *info, unsigned char *src,
> > + int src_width, int src_height, unsigned int src_stride)
> > +{
> > + int swap_line = 0;
> > + int swap_gbrg = 0;
> > + int dst_x, dst_y;
> > +
> > + switch (info->fmt) {
> > + case V4L2_PIX_FMT_SBGGR16:
> > + case V4L2_PIX_FMT_SBGGR14:
> > + case V4L2_PIX_FMT_SBGGR12:
> > + case V4L2_PIX_FMT_SBGGR10:
> > + case V4L2_PIX_FMT_SBGGR8:
> > + swap_line = 1; /* BGGR -> GBRG */
> > + swap_gbrg = 1; /* GBRG -> GRBG */
> > + break;
> > + case V4L2_PIX_FMT_SGBRG16:
> > + case V4L2_PIX_FMT_SGBRG14:
> > + case V4L2_PIX_FMT_SGBRG12:
> > + case V4L2_PIX_FMT_SGBRG10:
> > + case V4L2_PIX_FMT_SGBRG8:
> > + swap_gbrg = 1; /* GBRG -> GRBG */
> > + break;
> > + case V4L2_PIX_FMT_SRGGB16:
> > + case V4L2_PIX_FMT_SRGGB14:
> > + case V4L2_PIX_FMT_SRGGB12:
> > + case V4L2_PIX_FMT_SRGGB10:
> > + case V4L2_PIX_FMT_SRGGB8:
> > + swap_line = 1; /* RGGB -> GRBG */
> > + break;
> > + case V4L2_PIX_FMT_SGRBG16:
> > + case V4L2_PIX_FMT_SGRBG14:
> > + case V4L2_PIX_FMT_SGRBG12:
> > + case V4L2_PIX_FMT_SGRBG10:
> > + case V4L2_PIX_FMT_SGRBG8:
> > + /* No conversion needed. */
> > + break;
> > + default:
> > + return -1;
> > + }
> > +
> > + if (src_width % 2 != 0 || src_height % 2 != 0) {
> > + printf("WARNING: bayer layout conversion not supported for uneven sized images -> expect bad colors\n");
> > + return 0;
> > + }
> > +
> > + if (swap_line) {
> > + /* Swap pixel pairs on each line AB -> BA. */
> > + for (dst_y=0; dst_y<src_height; dst_y++) {
> > + for (dst_x=0; dst_x<src_width; dst_x += 2) {
> > + if (info->bpp == 8) {
> > + unsigned char *p1 = &(src[src_stride*dst_y+(dst_x+0)]);
> > + unsigned char *p2 = &(src[src_stride*dst_y+(dst_x+1)]);
> > + unsigned char tmp = *p1;
> > + *p1 = *p2;
> > + *p2 = tmp;
> > + } else {
> > + unsigned short *p1 = (unsigned short *)&(src[src_stride*dst_y+(dst_x+0)*2]);
> > + unsigned short *p2 = (unsigned short *)&(src[src_stride*dst_y+(dst_x+1)*2]);
> > + unsigned short tmp = *p1;
> > + *p1 = *p2;
> > + *p2 = tmp;
> > + }
> > + }
> > + }
> > + }
> > +
> > + if (swap_gbrg) {
> > + /* Swap R and B components, format is always GBRG at this point. */
> > + for (dst_y=0; dst_y<src_height; dst_y += 2) {
> > + for (dst_x=0; dst_x<src_width; dst_x += 2) {
> > + if (info->bpp == 8) {
> > + unsigned char *p1 = &(src[src_stride*(dst_y+0)+(dst_x+1)]);
> > + unsigned char *p2 = &(src[src_stride*(dst_y+1)+(dst_x+0)]);
> > + unsigned char tmp = *p1;
> > + *p1 = *p2;
> > + *p2 = tmp;
> > + } else {
> > + unsigned short *p1 = (unsigned short *)&(src[src_stride*(dst_y+0)+(dst_x+1)*2]);
> > + unsigned short *p2 = (unsigned short *)&(src[src_stride*(dst_y+1)+(dst_x+0)*2]);
> > + unsigned short tmp = *p1;
> > + *p1 = *p2;
> > + *p2 = tmp;
> > + }
> > + }
> > + }
> > + }
> > +
> > + return 0;
> > +}
> > +
> > static void raw_to_rgb(const struct format_info *info,
> > unsigned char *src, int src_width, int src_height, unsigned char *rgb)
> > {
> > @@ -480,7 +571,8 @@ static void raw_to_rgb(const struct format_info *info,
> > case V4L2_PIX_FMT_SBGGR10:
> > case V4L2_PIX_FMT_SGBRG10:
> > case V4L2_PIX_FMT_SRGGB10:
> > - printf("WARNING: bayer phase not supported -> expect bad colors\n");
> > + if (raw_layout_to_grbg(info, src, src_width, src_height, src_stride))
> > + error("Can't convert RAW layout to GRBG");
> > /* fallthrough */
> > case V4L2_PIX_FMT_SGRBG16:
> > case V4L2_PIX_FMT_SGRBG14:
> > @@ -523,10 +615,10 @@ static void raw_to_rgb(const struct format_info *info,
> > case V4L2_PIX_FMT_SBGGR8:
> > case V4L2_PIX_FMT_SGBRG8:
> > case V4L2_PIX_FMT_SRGGB8:
> > - printf("WARNING: bayer phase not supported -> expect bad colors\n");
> > + if (raw_layout_to_grbg(info, src, src_width, src_height, src_stride))
> > + error("Can't convert RAW layout to GRBG");
> > /* fallthrough */
> > case V4L2_PIX_FMT_SGRBG8:
> > - /* FIXME: only SGRBG8 handled properly: color phase is ignored. */
> > buf = malloc(src_width * src_height * 3);
> > if (buf==NULL) error("out of memory");
> > qc_imag_bay2rgb8(src, src_stride, buf, src_width*3, src_width, src_height, 3);
>
> --
> Regards,
>
> Laurent Pinchart
--
Kind Regards,
Niklas Söderlund
next prev parent reply other threads:[~2026-01-22 9:28 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-05-11 15:46 [raw2rgbpnm,PATCH] Add support for all RAW memory layouts Niklas Söderlund
2026-01-22 9:20 ` Laurent Pinchart
2026-01-22 9:28 ` Niklas Söderlund [this message]
2026-01-22 12:36 ` Sakari Ailus
2026-01-24 20:18 ` Niklas Söderlund
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260122092834.GC2316109@ragnatech.se \
--to=niklas.soderlund@ragnatech.se \
--cc=laurent.pinchart@ideasonboard.com \
--cc=linux-media@vger.kernel.org \
--cc=sakari.ailus@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox