Clipping.... What is it?

[Aidan]

Can someone explain in relativity simple terms to u defat and what is meant when clipping is referred to in images?

I read being mentioned quite frequently, but even after being in the AP game for a couple of years, I don’t fully understand what it refers to!

Rgds

Aidan

[vlaiv]

Clipping is process of cutting off certain values due to various factors.

It is usually consequence of limited range of numbers that we work with. For example one can have "clipped stars" and bright parts of the image. This happens if target is bright enough (like some stars tend to be) and exposure is long enough. Maybe best analogy would be trying to pour water in buckets. All buckets can hold 5 liters of water and you try to pour 3, 6 and 8 liters in respective buckets. You will end up with 3L of water in first, but 5L and 5L in second and third - as they can only hold so much water. Values of 6 and 8 liters were cut off because of bucket size and you ended up with 5L in each.

Because clipping cuts off values to same value - you end up with "white" surface instead of features in that place in the image. You can also clip black levels in histogram manipulation - it will create uniform dark surface without details - just like you selected area and did "bucket fill" tool in PS.

Whenever you have any sort of cut off - you are changing information in image and loosing details / features. That is why this is bad (but sometimes can be used to good effect - like minimizing background noise and smoothing background out).

HTH

[ollypenrice]

Do you understand the histogram of an image? Start with this. It isn't hard to understand but it's the starting point for the conversation. Below is a widefield image of M42 and its histogram generated in Photoshop by opening Levels. The graph you see is a bar chart of pixel brightnesses over the whole image. On the x axis (left to right) we have dark pixels on the left and bright pixels on the right. The vertical axis displays the number of pixels at each point on the brightness axis, left to right. We see that the most popular brightness in this image lies a bit less than a quarter of the way from darkest to brightest. There are very few pixels at the very darkest point on the left and very few at the brightest point on the right. The darkest pixels on the left will be the darkest bits of background sky. The brightest bits on the right will be the cores of stars saturating the sensor. This is a fairly healthy histogram. Now let's ruin it by clipping!

 

Black clipping: I can, in levels, move the black point on the left well to the right and cut out the darkest pixel values, like this:

Many beginners will see this as a way of removing sky gradients and adding impact and contrast. In fact all it does is cut out precious faint signal from subtle low brightness parts of the data. You are throwing away much of your precious capture. This is black clipping and is to be avoided like the plague.

Now we can look at white clipping but this doesn't really happen in the same way. It happens either when exposures are too long to stop bright parts of the target saturating the chip or when a stretch is too aggressive. Here I have taken the same image and stretched it so hard that I have run out of range on the screen and all the brightest parts are at the same value, 'burned to white' or 'saturated.'

 

The histogram of this overstretched or saturated image looks like this:

 

What we see is that the healthy image's histogram has a flat line on the left and a flat line on the right showing that it has not been significantly black clipped (on the left) or white clipped (on the right.)

It is perfectly possible to combine two stretches of the same data to cover a wider range of brightnesses but that is another story.

Olly

 

Edited May 27, 2019 by ollypenrice

[adyj1]

Anyone in France? Pop round to @ollypenrice and check he's OK, would you? I'm starting to worry he got abducted half way through that last post... 😉 

Phew - he's back. Stand down Les Astronomeurs! 😂 

Edited May 26, 2019 by adyj1

[ollypenrice]

27 minutes ago, adyj1 said:

Anyone in France? Pop round to @ollypenrice and check he's OK, would you? I'm starting to worry he got abducted half way through that last post... 😉 

Rural France = no light pollution but also = rural French internet!!!

lly

[Aidan]

A picture paints a thousand words.   Thanks Olly.  Thats pretty much cleared things up for me.  I was on the right track in the way i was interpreting clipping.  Makes much more sense with you descrption and images.  Thank you