I've some issues, passing Raw Datas to an OpenCV Mat.
At first i used dcraw, but there is too much processing for me, my output images is blurred and pinked (yeah my picture has too much red in it i don't know why). So i tried to change dcraw processing params, but it was worse.
So i'm trying to used the unprocessed data and to fill a cv::Mat with it ! I read a lot of topics in this forum, and tried a lot of fixes, but i couldn't manage to do what i needed.
If I understand the documentation of user_flip correctly, not setting it (keeping it's default value of -1) should "auto rotate" the image according to what the accelerometers of the camera have detected when the image was shot.
Are there any known issues with Sony's arw files? It does not seem to work, while xnView (which claims to use dcRaw) rotates the image...
You've already explained that clipping in raw data was difficult to detect (has different forms), and that clipping values may even depend on the color channel.
Imagine a theoretical picture, with, after pre_mul or cam_mul applied, has:
red clipping to 60000.
green clipping to 40000 (I consider g and g2 to have the same behavior).
blue not clipping, climbing up to 50000.
How does dcraw_process() -no_auto_bright=1- stretch this result to 65535?
I guess it can't stretch each channel independently, since this would mess the white balance.
Sorry if this has been posted before! I would like to know if there is an API call to retrieve what is the Bayer CFA pattern ('RGGB', 'BGGR', 'GRBG', 'GBRG'. I ) on a given image. Dcraw can be called with -v -i and will output the pattern, RG/GB for example here:
Apologies if I missed something but is there an API-function to determine the file-type of the raw-file opened (and unpacked) with LibRaw?
Specifically, I would like to know if the file represented by the unpacked C++ LibRaw-class is a DNG-file. I can obviously check the filename extension but I'm assuming LibRaw does some internal matching of file types. Is that exposed through the API somewhere?
I'm trying to do my own processing: as far as my tests are correct, calling dcraw_process() do the whole job and produces a 8bit image.
I want to get the full bits, so I guess I have to stay with raw2image().
It seems that calling raw2image() gives us a non demozaised image, r, b, and twice g, and that just after a call to raw2image(), the second g cannot be ignored.
I own an old Nikon D100 camera and I use both Ufraw and Luminance HDR in order to get simple or HDR photos depending on content. Since Ufraw uses the old dcraw library, it produces a better result while dealing with bright areas. That is funny as they say libraw should be better. I used Darktable as a control app and it produces the same result.
Recent comments