Problem with RAW Conversion Parameters with Canon 550D raw files

[xatamec]

Hi,

I have a problem with the raw images from a Canon 550D when I try to do all the callibration/integration process with the darks,bias, flats and light images.
First of all I set the DSLR_RAW parameters as shown in this image:


Once done, when I open a RAW file (CR2) or try to convert it to fit with BatchFormatConversion, the value of the green and blue chanels of all the pixels of the image is 0, it happens with all the raws I've tried to open or convert to fit (bias, darks, flats and lights).

If I uncheck the "Create Raw Image" and choose AHD Interpolation, then the image is correctly debayered and both green and blue channels do have data.

I've done the whole callibration/stacking process with Pixinsight several times with images from other Canon DSLR cameras (350D, 1000D and 50D) and have never had this problem before.

For your reference I'm using Pixinsight 1.7 with all the updates installed on a laptop with Windows Vista and 4Gb RAM.

Any ideas of what's happening? I can upload images if this helps.

Thanks,
Sergi

[pfile]

it could be that the version of DCRAW that PI is using does not know how to parse the 550D CR2 files. they are all a bit different and updates have to be made to dcraw each time a new camera comes out.

however, i thought the 550D came out some time ago and so perhaps it should already be supported? i'm not sure how to discover the version of dcraw that PI is using since i believe it's been compiled into a pixinsight library rather than being a standalone binary.

[xatamec]

Thank you pfile. After reading your post I've done a Google search and have found that Dave Coffin released the dcraw version 9.0 on 28 May 2010, this is the first dcraw version supporting the Canon 550D raw files.
Which dcraw version is used in Pixinsight 1.7?

Sergi

[pfile]

the pixinsight dcraw version has got to be at least at that level.

can you take a daylight picture and post the CR2 somewhere? i can try to open it at least.

[Juan Conejero]

Hi Sergi,

The current DSLR_RAW module uses dcraw revision 1.443, released May 2011, so AFAIK there should be no problems with the 550D (you can check the version by selecting DSLR_RAW on Format Explorer and hovering (or double-clicking) the 'implementation' item).

As Rob says, a CR2 of a daylight picture taken with your camera would help me a lot to identify the problem, since otherwise I have no way to get a 550D for testing (mine is a 450D and works fine all the time).

That said, I'll try to keep the DSLR_RAW module in sync with the latest dcraw version ASAP... May 2011 is getting a bit 'vintage' in dcraw terms

[xatamec]

Thank you Juan and pfile!

It's been a mistake  sorry!

At some stage of the callibration process which I don't remember now I got an error message saying something like: the values in the image are 0 or near to 0 (don't know yet why this error was produced).
So I took a raw flat and moved the cursor over the image. Wherever I put the cursor the readout was always 0 for the B and G pixels, look at the attached picture. The raw bayer images are very dark, so I believed the G and B chanels were really 0 on the image because the same happened with all the raws I opened afterwards.
However yesterday I stretched one of the raws with an STF and could see clearly the bayer matrix when zooming it.
It seems that for some reason when the mouse is moved over the image at the size it is opened by default by PI, the cursor is always over a red pixel of the matrix and that's why the readout of the image is always 0 for the B and G channels, despite there are twice as many green pixels than red pixels (that's why the streetched image looks greenish) and that lead me to confusion.



Thank you for your help!

Sergi