DSLR Linearity

[dundizzle]

Hey all, I'm still very new to this whole AP deal and I'm starting to play with StarTools using others' raw data for the time being while I get the last of my equipment together.

I have a quick question regarding opening files from an unmodded DSLR (Canon Rebel T3), which were stacked in DSS. Obviously the data has been Bayered coming from a DSLR, but how do I know if the data is "Linear, Bayered, Not white balanced" vs. "modified and not linear"? Are there settings on the DSLR white balance I can use to control this?

I've also heard that using DSS causes some kind of stretching or auto-balancing to the data. Are there any tips/tricks to minimize this, or avoid it all together?

Any info is much appreciated, cheers!

[admin]

Hi apologies for the delay - it's a busy time of the year...

DSS unfortunately applies the white balance of the camera, so the data will be white balanced. As long as you don't apply any sort of modifications (stretches, sharpening, etc.) and just save the (very dark) data straight out of DSS, then your data should be linear. It's *very* important that StarTools gets its hands on the data when it is as 'virgin' as possible. Any sort of modification (which includes color balancing) will reduce StarTools' effectiveness in virtually all areas of processing, for the simple fact that correct noise propagation is difficult/impossible to Track for data that is already noisier than it needs to be.

Modified and not linear is for images that have already been processed - it's a last resort measure for StarTools to try and work with such data.

Hope this helps!

[Chunkles]

Hi Ivo,

This has been a question of mine as well. I also am using data from a T3 that has been stacked in DSS. I perform no other manipulation of the image after stacking; I just save it as a FTS file.

So, my image would be: Bayered AND white balanced (since you mentioned that DSS is already applying the white balance from the camera).

But the two linear options in StarTools are:
1. Linear, not Bayered, is whitebalanced
2. Linear, is Bayered, not whitebalanced

Which one should be used in this case since both options are only 50% accurate for this scenario?

Thanks!

[admin]

Hi Ivo,

This has been a question of mine as well. I also am using data from a T3 that has been stacked in DSS. I perform no other manipulation of the image after stacking; I just save it as a FTS file.

So, my image would be: Bayered AND white balanced (since you mentioned that DSS is already applying the white balance from the camera).

But the two linear options in StarTools are:
1. Linear, not Bayered, is whitebalanced
2. Linear, is Bayered, not whitebalanced

Which one should be used in this case since both options are only 50% accurate for this scenario?

Thanks!

Aha, I understand where the confusion stems from now.
If you read carefully, it actually says;

Code: Select all

Linear, was not bayered or is whitebalanced

and

Code: Select all

Linear, was Bayered, is not whitebalanced

The option first means;

1. This data is linear
2. This data was not bayered (e.g. this data does not originate from a source with a bayer matrix, such as a mono CCD that uses colour filters, or some other type of color filter array other than a Bayer pattern)
3. OR This data was bayered (e.g. this data does originate from a source with a bayer matrix, such as an OSC, DSLR, etc), but it has been white balanced already by some other software.

The second option means;

1. This data is linear
2. This data was bayered (e.g. this data originates from a source with a bayer matrix, such as an OSC, DSLR, etc.)
3. AND this data has not been white balanced.

If you're using DSS with DSLR data, then choose option 1, since the data has been white balanced (condition #3).

The whole reason behind letting StarTools know all this, is that a Bayer matrix allocates twice as many pixels to the green channel than it does to the blue and red channels (translating in a better, less noisy signal in the green channel). Therefore, StarTools can take this into account when it tracks noise propagation (the 'Tracking' feature that drives most modules in StarTools).

[Chunkles]

Thanks for the explanation! That explains some of my issues since I have always chosen option #2 when I should have been choosing #1!

I didn't actually realize that DSS was white balancing until I read this thread, and instead was focusing on the "not Bayered" statement in #1.

This does, however, raise a question I've always had about DSS: Should I not be checking the "Align RGB channels in Final Image" option when stacking? Does that further process the image (which is what we don't want)?

Thanks!!

[admin]

Thanks for the explanation! That explains some of my issues since I have always chosen option #2 when I should have been choosing #1!

I didn't actually realize that DSS was white balancing until I read this thread, and instead was focusing on the "not Bayered" statement in #1.

This does, however, raise a question I've always had about DSS: Should I not be checking the "Align RGB channels in Final Image" option when stacking? Does that further process the image (which is what we don't want)?

Thanks!!

Don't check Align RGB channels, nor Per Channel Calibration.

[dundizzle]

Thanks for the explanation! That explains some of my issues since I have always chosen option #2 when I should have been choosing #1!

I didn't actually realize that DSS was white balancing until I read this thread, and instead was focusing on the "not Bayered" statement in #1.

This does, however, raise a question I've always had about DSS: Should I not be checking the "Align RGB channels in Final Image" option when stacking? Does that further process the image (which is what we don't want)?

Thanks!!

Don't check Align RGB channels, nor Per Channel Calibration.

Thanks Ivo, this makes so much more sense now!

Are there any other DSS tips/step-by-steps to keep the data as virgin as possible? I can find plenty of walkthroughs with using DSS, but I have a strong feeling that they all manipulate the data.

Edit: Also, if using dcraw to pre-process the images before stacking, would you still use Option 1 even though the whitebalance isnt applied?

[MadMaxwellSmart]

To all who posted in this thread, a salute and thank you. I have been using the wrong one, it always pays to read carefully...Thanks Ivo, I went back and processed one of the images and had better results....Thank you

[admin]

if using dcraw to pre-process the images before stacking, would you still use Option 1 even though the whitebalance isnt applied?

In that case you can use option #2.

[e.ventura]

I know this is an old(ish) thread, but doesn't DSS have a couple of checkboxes on the "RAW/FITS DDP configuratioon" window for NOT applying any whitebalance? Does it nevertheless apply whitebalance and we should always choose the "whitebalanced" option in StarTools?