RAW conversion settings for .CR2 files

[star gazer]

I my previous thread, I was given advice on setting the DSLR conversation preferences. The advice was to tick the "Create RAW Bayer image" and to set "No black point correction".

When I do this the image becomes very green. I understand that a bayed image has 2G pixels for every 1R and 1B. However doesn't running the de-bayering restore the image to the "correct" colour balance??

After de-bayering, my image is too green especially when the auto screen transfer function is applied. Now I was also given advice to unlock the colour channels before applying the automatic stretch. Fine, no problem, the image then has a better colour balance, but now I cannot tweak the screen transfer function without upsetting the colour, even if i the "re-lock" the channels.

You might say 'don't worry as the screen transfer function as it is only for a temporary view'. Yes that is true, but with the resultant "green" image, life is very much more difficult when it comes controlling colour when I permanently stretch the histogram.

Now for my real question: What is the purpose of the Interpolation options on the Raw Format Preferences dialog? If I ignore the recommended settings (first paragraph above), what am I doing?

I have found that un-ticking everything in the RAW preferences dialog and choosing the Bilinear radio button, it is much easier to control the colour. Again what am I doing here? Am I converting a CR2 file into de-bayered FIT, so further down my workflow I don't need to de-bayer? What are the disadvantages of working this way? What about the other Interpolation options? When would one use them?

Sorry for so many question, but converting RAW files is a bit of a mystery to me.

Thanks...

[georg.viehoever]

If you want to restore the color balance of debayered images, you should have a look at the BackgroundNeutralization and ColorCalibration processes. They are well documented.
If you want to learn about the different debayer methods, just google a bit.
Georg

[pfile]

the main reason to use raw is so that you can calibrate your lights. if you are not going to do that, you might as well just tell DSLR_RAW to use one of the debayering algorithms.

are you shooting through any kind of filter? an aqua-blue cast is normal if you are using a CLS filter. of course that cast should show up whether you debayer using the DSLR_RAW handler or with the Debayer process. have you experimented with the bayer filter settings in the Debayer process to make sure you have the right bayer matrix set? one way to do this is to just take a daylight photo, load as raw and then try debayering it with different settings in the Debayer process until you find the right one. it will be obvious.

if a debayer method is chosen in DSLR_RAW, then yes, you have debayered the file as you've opened it. it's not a FITS file until you save it - it's stored in pixinsight's internal representation of an image.

the interpolation options control how DSLR_RAW converts your bayer matrix raw CR2 to a color RGB image. there are a lot of different algorithms - superpixel, bilinear interpolation, AHD, VNG...

as for undoing any color cast, what i usually do is what you've described with the STF, then transfer the STF settings to the histogram tool and apply. this is done by dragging the STF triangle into the horizontal strip on the histogram tool where it's triangle lives. then the balance is mostly right and you can proceed with ColorCalibration if you like...

[star gazer]

I'm shooting with a Hutech Light Pollution Filter LPS-48.

Here are the same subs processed into a single images and then opened in PI.

The top images was processed entirely in PI. Note the green is after de-bayering. Running background neutralisation does indeed remove the green cast but it just shifts the colour problem elsewhere.

The bottom image was processed in DSS and, while far from perfect is a much better result in my mind. The only problem with DSS is that I always get doughnut stars.

[Nocturnal]

http://www.carpephoton.com/index.php?option=com_content&view=article&id=31:getting-the-right-color-is-easy&catid=10:pixinsight&Itemid=4

Donut stars in DSS are generally caused by the cosmetic options. People can't resist turning them on but don't bother to check the before/after to make sure they have the desired result.

[star gazer]

Hey Nocturnal,

Oouch!!

That's harsh!

I just took the defaults ..for a quick and dirty comparison.

So what are the cosmetic option(s) I should click off?

Thanks for your help

[Nocturnal]

Didn't mean to be harsh but I'm glad I got the point across  Click off -all- cosmetic options. Stack. If you like the result you are done. If you think there are things wrong with your image that cosmetic options can fix, enable them one at a time and see what it does.

The cosmetic options fall into the same category as the drizzle stack option. The larger someone's DSLR is, the more likely they'll try to enable drizzle stacking and the more likely DSS will crash because it can't deal with the 9x 18MP images.

[georg.viehoever]

I'm shooting with a Hutech Light Pollution Filter LPS-48.

Here are the same subs processed into a single images and then opened in PI.

The top images was processed entirely in PI. Note the green is after de-bayering. Running background neutralisation does indeed remove the green cast but it just shifts the colour problem elsewhere.
...

Could you make your PI fits image availlable somewhere? You probably use BackgroundCalibration and ColorCalibration in a wrong fashion.
Georg

[Nocturnal]

I actually think the problem is that the image isn't properly flattened. BN won't work with that type of image. DBE can be used to salvage the situation but proper flattening is the real fix.

[georg.viehoever]

Hi Sander,

you are right, the image is not properly calibrated, both the PI version and the DSS version. You can clearly see the loss of brightness in the corners of both images. I just wanted to try to achieve a reasonable color calibration.

Hi Stargazer,

see http://pixinsight.com/forum/index.php?topic=2570.0 for a complete DSLR calibration workflow. Note that the use of DSS is certainly much easier than the use of PI for calibration and integration, and it only makes sense to proceed to PI to tweak the last bits of signal out of your inage sets. Calibration with flats, biases, darks definitely is a prerequisite for this.

Georg

[star gazer]

I'm shooting with a Hutech Light Pollution Filter LPS-48.

Here are the same subs processed into a single images and then opened in PI.

The top images was processed entirely in PI. Note the green is after de-bayering. Running background neutralisation does indeed remove the green cast but it just shifts the colour problem elsewhere.
...

Could you make your PI fits image availlable somewhere? You probably use BackgroundCalibration and ColorCalibration in a wrong fashion.
Georg

Hi, PI image hasn't had DBE or background neutralisation applied, although if I do apply them colour and contrast are lost at least in comparison to DSS.

[star gazer]

Hi Stargazer,

see http://pixinsight.com/forum/index.php?topic=2570.0 for a complete DSLR calibration workflow. Note that the use of DSS is certainly much easier than the use of PI for calibration and integration, and it only makes sense to proceed to PI to tweak the last bits of signal out of your inage sets. Calibration with flats, biases, darks definitely is a prerequisite for this.

Georg

Hi Georg,

There are many references to that topic in this and other forums. It is very informative especially to a newbie like myself, but I am not sure it is complete, or at least there are steps that are ambiguous. Also many of the switches or options are not explained. Following the steps as best I can, with trial and error, I am not able to get the result I hope for. Generally the result is not as good as DSS not withstanding the doughnuts.

My frustration is I am convinced that PI can produce the post possible result, I just haven't found 'the way' despite many hours of trial and error.

Tim

[Harry page]

Hi

If you want to send us a raw stack , I am sure I could tickle up a quick vid of a basic process 

Harry

[Nocturnal]

First flatten, then worry about post processing. It will be tough to cook something acceptable from an image that has multiplicative defects. In other words one that received more light in the center than on the edges. With DBE you can attempt to create an artificial flat but since there's a target in the center it's hard (if not impossible) to place a sample there.

Flatten your images and DBE will get you 95% to acceptable colors. Then use CC to get the last 5%.