Liner & none linear - what does it mean ??

[Skipper Billy]

I am trying really hard to get to grips with Pixinsight and many of the tutorials talk about linear and none linear without explaining what it means.

Anyone able to explain in language suitable for a numpty ???

This is making my head hurt !!!!!!

[ronin]

In what context is it being used?

[kalasinman]

I've heard it used in regards TIFF vs FITS files. I don't get it either.

[Skipper Billy]

Used in terms of workflow process eg - as part of the the linear process you can do XXX and as part of the non-linear process you should do YYY or ZZZ.......

[StuartJPP]

Linear data/format means that the data is in either floating point notation or some other large number format (64bit integer for instance). It is done like this so that any computation/processing on it at this stage won't lose data or resolution. It prevents faint data being lost and bright areas from being clipped.

Nonlinear format means it has been scaled so that it can be displayed on a computer monitor. Being scaled means that data is lost but at the final stage it doesn't matter as much.

So you do most things in linear format first (calibration, stacking etc.) then nonlinear last to produce a final image or an image for using in Photoshop or the like.

PixInsight's Screen Transfer Function scales the linear data temporarily so that you can see what the data looks like on your monitor but the underlying data is undisturbed.

Hope that doesn't confuse matters more

[Skipper Billy]

Great - thanks very much!

[frugal]

The linear data will be where the value in the file will be directly proportional to the number of photons captured. I.e. if you need 10 photons to get a value of 1, then you need 20 photons to get a value of 2 and 100 photons to get a value of 10. So Linear data just refers to the values received by the camera due to the number of photons captured. It is linear because if you plot the number of photons against the stored value it is a straight line (hence linear).

However the human eye does not work on a linear basis. The human eye works logorithmically, so higher values are exponentially brighter than lower values. 

Stretching the data from linear to non-linear just involves applying this logarithmic scale to the input data, this makes it more useful to the human eye.

When your camera converts the data to JPEG it applies this stretch itself. When Photoshop opens a RAW file, it applies a stretch when it does so. This way you never realise that the data is being converted from one data model to another.

With astrophotography you get to chose the stretching factors to make the most of the image.

[NickK]

Linear is best thinking of a straight line, where the relationship between  input and output pixel values remain the same before and after processing.

Non-linear - think of a curve between input and output pixel values of processing an image.

An image is made up of pixel values - between 0 and max pixel value (a range). So when you capture, the minimum and maximum values for the entire sub image may be a small portion of the theoretical full range (i.e. between 0 and max value).

If you use linear - then you would not alter this range in relationship the 0 to max pixel range.

If you use non-linear - then then you change this relationship. so for example you could scale up the entire image pixel minimum and maximum values  to fit the full maximum range. Just like histogram stretching where you see a curve ( i.e. non linear).