Be careful when removing power lines and dust spots in Luminar NEO

Thursday 25 August 2022, 15:30 by Nando Harmsen | 3,614 views | 3 comments

This article was translated from Dutch with the help of AI. Read the original.

With Luminar NEO, we gained the ability to automatically remove power lines and dust spots. That is very handy, of course, and Luminar is keen to promote it. Especially because other software does not have this feature yet. But it also has disadvantages. It can sometimes be rather time-consuming.

It was naturally one of the selling points: removing power lines and dust spots from a photo with the press of a button. The so-called artificial intelligence recognises lines or dust spots in the photo and automatically removes them for you.

Forget the populist term artificial intelligence for a moment. There is nothing intelligent about it. It is nothing more than image recognition, a piece of software that recognises certain pre-programmed elements in an image.

Once it recognises them, a clever, but not intelligent, edit is applied so that they appear to simply disappear.

However easy and appealing it may be, this image recognition is far from intelligent, so mistakes are made. Some elements in the photo are incorrectly recognised as a power line, or are not recognised when they should be. The same applies to dust spots.

If you let Luminar NEO loose on a photo with the power line and dust spot removal tools, part of the work will be done for you. Power lines disappear and dust spots on the sensor are neatly removed. Handy, isn't it? That saves work.

Removing dust spots, and sometimes other things too

If your sensor is not completely free of dust spots, which is perfectly normal, you can see them in your photo. They become visible especially with very small apertures, but also when you apply very strong and aggressive contrast.

Luminar NEO can automatically filter these out using image recognition. And Luminar does this reasonably well to well. They are filtered out quite effectively, especially in the sky, where they are often easiest to see.

It may be difficult to see in this small image, but there are several dust spots in the centre. The entire sky is covered in dust spots, which you would be able to see if you zoomed in.
However, if you look closely at a photo, there may still be dust spots present. Luminar does not always recognise them, especially in areas with lots of detail. Keep that in mind.

After the automatic removal of dust spots, there are still a number that Luminar NEO has missed. That is not a problem, because you can still do this manually.
Elements in your image can also be incorrectly identified as dust spots. You notice this especially when photographing a starry sky, where many faint stars are filtered out.

Even though some dust spots may be missed, it naturally saves a lot of work if 90% has already been done. Looking through the photo afterwards is the finishing touch.

Removing power lines

Things become more difficult for Luminar NEO when 'remove power lines' is used. Lines against the sky are easy. Luminar NEO handles those almost perfectly.

It is different when lines run against a busy background, or pass in front of or behind objects. In that case, lines are only partly removed, or too much is removed, causing your object to become distorted or even disappear completely.

The result after automatically removing power lines contains quite a few errors. You would expect an AI to understand the difference between the edge of the roof and a line.
Luminar NEO cannot recognise all of them either, and sometimes parts are identified as power lines when they are not. This makes the result seem rather random. The result is a photo with many small, inconspicuous errors that you do not notice immediately.

Check your photo after automatic removal

If you use the automatic power line or dust spot removal, it is advisable to look at the photo carefully and closely afterwards. Find the areas where things went wrong and correct them.

Do not assume that there are not many errors. Here, an entire lamp has been removed by the automatic power line removal, while parts of the power line have been left behind. So you still need to check the entire photo carefully for errors.
With dust spots, it is a matter of removing any spots that were missed and repairing any elements that were incorrectly identified as dust spots.

With power lines, it is more difficult, because you also need to look for areas where parts that were not power lines have been removed. By stepping back one step at a time in the editing panel and looking at 'before' and 'after', you will find them.

It is also possible to mask everything with good preparation and draw the power lines into the mask.

Time-saving or time-consuming?

Are these automated features actually time-saving, or not? If the software had genuine artificial intelligence, it would save time. In that case, the software would recognise the difficult parts of the photo without any problems.

After automatically removing power lines, you must check not only that everything is gone, but also that Luminar NEO has not removed too much.
Because there is nothing intelligent in the programme and it only uses image recognition, it is far from perfect. Checking the photo for errors, especially after removing power lines, can sometimes take considerably more time than simply doing it manually.

A simple line in the sky is usually not a problem for Luminar NEO, but it becomes problematic against a busy background. The same applies to dust spots, although to a lesser extent.

You might also like

• 

  The imgmi app, Luminar for your smartphone

  by Nando Harmsen

• 

  Converting a Luminar AI catalogue to Luminar NEO

  by Nando Harmsen

• 

  Luminar NEO gets a panorama feature

  by Nando Harmsen

• 

  Luminar NEO gets AI-generated content with GenSwap

  by Nando Harmsen

• 

  Luminar is changing in 2024

  by Nando Harmsen

• 

  What if AI spot removal doesn't see all the spots?

  by Nando Harmsen

3 comments

Have your say

Please note: Articles older than a week can only be commented on by signed-in visitors.
Would you still like to comment? sign in or create a free account.