Nick Sutrich is from Android Central.
The Pixel 6 and the Pro are stunning phones with a cutting-edge processor from Google that is designed to process artificial intelligence features like nothing else on the market. Magic Eraser, a feature that was teased a year ago and finally delivered in at least some capacity on these phones, was one of the many features that was introduced on the Pixel 6 series.
It is possible to remove things from photos that should not be there, such as a photobomber that ruined the perfect moment, or an object that got in the way of the perfect shot.
There are other people who have been working on this kind of tech. Adobe has been able to do it for a long time, although the new object selection tool in Photoshop is designed to make this function better.
With the release of the S22, it's easier to tone down reflections and shadows with the single click of a button.
Who does it best? The winner of our original tests was the Pixel 6 with its smarts and object identification technology. Does the conclusion of the Magic Eraser vs Object Eraser change with the release of the S22? Let's look at it.
I imported each of the photos into the gallery apps of the two companies, and then used the Magic Eraser or Object Eraser on them to remove people or objects. I included examples from the new version of the program, which is available for $10 a month on Adobe Creative Cloud, and where I highlighted people or areas using the object selection tool.
Pixel Magic EraserSource: Android Central
The idea was that most people wouldn't want to spend a lot of time editing a photo, so comparing automated tools is the most sensible.
I found that the Magic Eraser was the easiest to use. When opening a photo and selecting Magic Eraser, the tool will attempt to identify any objects it thinks need to be erased. A single button can be pressed and a result given in mere seconds if you like the automatic selections.
Manual selections can be made by circling any object you would like to erase, and the Tensor processor of the internet giant takes a moment to process the object and remove it. You can draw a shape with your finger after an object is erased to try and clean it up.
Samsung Object EraserSource: Android Central
There are three different levels of the Object Eraser. There were only two ways to modify photos.
When you open a photo and choose Object Eraser from the menu, you will need to manually select any objects that need to be erased by either tapping or circling them. The object will be tried to be identified and erased from the scene.
You can draw a shape around the area, and Object Eraser will attempt to fill in the area without identifying objects. If you draw around an object or tap it, the method will always try to identify the shapes.
The third is a set of new tools that make it easy to tone down shadows and reflections by selecting either of the two buttons in theUI. I found that the automated method was hit-or-miss, and that it would find nothing to do.
The shadow and reflection tool gives SAMSUNG an edge over other companies, as it could help lighten up photos and remove reflections without being noticed.
Recent years have brought easier-to-use tools to the platform, which has made Adobe Photoshop the professional tool of choice. I use my Windows 11-based PC for this test, and I used the Adobe software for it.
Photoshop object selection toolSource: Android Central
The new object selection tool is a refined version of similar tools from the past. If you choose to use them, you will get more options if you choose to do so. A single click with the tool will initiate a scanning process that can take up to 10 minutes on some computers.
Clicking on other objects will include them in the selection. You can start an automated fill by hitting the backspace key or shift+F5. If you go to the Edit menu and select Content-Aware fill, you will be taken to a more advanced interface where you can make more specific sample selections.
The simplest and most automated result was achieved by using the object selection tool and hitting shift+F5.
I wanted to see what has changed since the release of the S22, so I got some refreshed images. I wanted to highlight the two new tools that can help clean up photos after you take them.
In the two examples above, I used the buttons that are present in the object eraser. The tools worked well in the two examples, making the shadows brighter with a single click. After pressing either button, neither photo has been touched up.
Is it possible to achieve the same results with the Magic Eraser even without the buttons? The answer is a resounding "yes" if we just erase reflections.
The reflection of my living room light ruins a great portrait shot of Krrsantan, one of my favorite new Star Wars characters. Both Magic Eraser and Object Eraser did a great job of removing the reflection, but I like how the software added texture and color to Krrsantan's head.
How about shadows? There is not an easy way to remove shadows in the photos on the Pixel 6. You could bring up shadows across the entire image, but there is no granular option like that. You have to look at the example above to see how shadow erase can improve your photos.
It's time to erase objects in photos. Did the release of the S22 improve this? Not really.
The original photo is followed by the result of Magic Eraser and object eraser. In every instance, the scene looked better without the objects and people I tried to remove.
No matter if it is a large foreground object or a small background object, the Magic Eraser seems to be able to identify the pattern.
The quality of front-facing cameras in phones has led to the rise of selfies. It helps us capture memories of events spent with other people, but it doesn't mean you want to include everyone in the picture. I found a few examples of other people trying to remove my family from their digital existence.
The people in the foreground were found to be distracted by someone in the background. The results are affected by the different methods each company uses to identify objects.
There is a source of the Android Central.
The Pixel Magic Eraser does a better job than the other two. When objects are removed, the Magic Eraser does a better job of giving users the ability to clean up the photo by drawing circles around the area.
In the first photo I took with my wife and me, the performance of both companies was pretty close, but the object selection tool of Adobe ended up creating strange ghost-like remnants of the people in the scene. Further circling of the background helped clean up the floating tent on the shot, and I just got more repeating noise with the tool.
There is a source of the Android Central.
The second photo of my son shows the same pattern as the first, but it is easier to identify something wrong with the photo on the internet. The tool failed again.
There is a source of the Android Central.
The last photo of my wife and me at the pool was photobombed with a shirtless dude in the background. No thanks. He was completely erased from the scene in a convincing way, only leaving a small artifact around my wife's shoulder.
The Object Eraser did a good job of wiping him out, but it left a small distraction. Drawing around it made it worse. The method was simply not good. The tool did not leave a ghost-like halo around where the person used to be.
This round, it was clear that Google won.
There are other things that you might want to remove from an image. Powerlines can get in the way. Sometimes, it's a car that drove in front of the camera at the wrong time, or an obnoxious red sign. These tools can help erase silly mistakes.
There is a source of the Android Central.
The first example was captured on my phone while the car was moving and it was impossible to just take it back. I could either toss the photo or attempt to remove the car. The result of the Pixel Magic Eraser was used to post this one, but only because of the default aspect ratio on the site.
In this case, neither of the two companies did a good job of removing the car and making the result look believable. It was knocked out of the park by the use of the software. I drew a shape around the trees and used an automated fill to fix them after I erased the car. Three clicks and it looks perfect.
There is a source of the Android Central.
The previous examples were cut-and-dry. On the day I took this, the sunset was bright and beautiful, but the phone lines ruin the shot. The power lines were identified and removed with the help of the Magic Eraser. Any artifacts left behind look like clouds.
I had to draw shapes around the power lines to erase them because no one could identify the lines. That is not an efficient way to do things. In the end, the result of the program is worse than that of the program, especially where the power lines intersect with the trees.
There is a source of the Android Central.
On my trip to New York City, I loved the way the tall buildings reflected the setting and the architecture below was gorgeous. I didn't pay much attention to the red parking sign that sullied the view.
I am not sure if I would give any of these technologies much of a winning award. I focused on the red sign because none of them could clear out the silver pole. Too many of the surrounding texture were repeated by the program and ended up being more distracting.
The area was mostly darkened by the two companies, making it look like a shadow in the scene. In this case, the winner is the internet giant, and the least distractive of the three, although I was not happy with any of the results.
There is a source of the Android Central.
The last one is the least modified of all the photos, and it was to test each technology's ability to deal with harsh shadows in low-light environments. The loudspeaker on the side of the building looked out of place to me, given the rest of the scene.
I prefer how the pseudo shadow was handled in place of the loudspeaker. It looks natural if you know where to look. The runner-up and third place were created by the artifacts of the Object Eraser.
Nicholas Sutrich is a source.
Jerry Hildenbrand is an expert in machine vision and I had a chat with him to understand how our phones can suddenly identify objects in pictures.
There are many reasons why modern phones have multiple cameras. While some of them are to provide a wider angle for photos or to grant better zoom detail, these sensors can also be used to approximate depth data since they provide a stereoscopic image, much like our eyes work.
Unlike the human brain, a computer doesn't really understand what it sees without a lot of training. In the case of Magic Eraser, Object Eraser, or Content-Aware Fill, none of these technologies are able to use this data and therefore, have a perceptive disadvantage for precise object detection.
Even though this is the case, the Tensor does an excellent job of automatically identifying objects in a scene it thinks shouldn't be there. Since introducing portrait mode on the Pixel 3, which only had a single rear camera, Google has been mastering edge detection, which is part of this.
The best way to detect an object in a photo is to use a camera lens that has a shallow depth, and the best way to make it appear like it was taken with a camera is to use a software program. There is little doubt that the same Magic Eraser algorithm is being used by Google.
Edge detection works by identifying contrast between objects.
The second is machine learning itself, which is what parts of Tensor are built for. In this case, the processing chain inside your phone is the one that recognizes different objects correctly, because millions of objects are run through an algorithm.
Machine learning standards can be different than what humans think of when they identify objects. In the machine learning world, correct can be wrong, yet, so long as it is consistent, it is still correct.
The last piece of the puzzle is able to trick the human eye because it identifies an object correctly.
Jerry says thatdemosaicing is how colors are created by tricking the eye. If you balance the right amount of red, blue, green, and black you force the eye to see every color, including the ones that cannot be reproduced 100%.
Jerry says that the same data can be used to create an object, and all that is left is to identify nearly patterns and colors.
The results will only be as good as the algorithm itself, which was trained via machine learning. When Magic Eraser runs up against the parts of building a photo that our eyes can discern best, it gets funky.
Many of the difficulties are related to the fact that Magic Eraser is only intended for wiping out background objects rather than foreground objects. I would say that it checks out given the results of our test here.
We have determined that the best-automated method between all three of these is the one by Google, but what about the rest of the powerful tools? If you want to recreate a scene as perfect as can be, then you need to use a program like Photoshop.
If you take enough time to learn the other tools, they will make up the difference.
Powerful tools like the healing tool in Photoshop can enable better results than automated tools if you are able to manually cut and paste. I really didn't spend much time on either of the photos for the manual comparisons, that should be no surprise to anyone.
Almost no one will want to take the time to learn these tools well enough to do a good job. Paying for a program is not even considered to do the work. Machine learning can be used in real-world situations that are actually useful, thanks to the amazing tools that are provided by both Google and Samsung.
Magic Eraser does an incredible job of helping users clean up photos without much effort at all. While it doesn't do as good of a job as the one done by the one done by the one done by the one done by the one done by the one done by the one done by the one done by the one done by the one done by the one done by the one
All Rights Reserved © 2024
