Some smartphones use intense in-camera automatic image filtering, well beyond what is necessary to capture the scene’s color and details. The real issue with aggressive filtering is that it can irremediably destroy data on the only copy of the photo you’ll ever have.
You can think of it as built-in Instagram-style filters that are designed for convenience because manufacturers sometimes believe that consumers want images that “pop”, without spending an additional few seconds of tweaking.
Although some users may like these styles very much, others might dislike them just as much. In our experience, it is unlikely that in-camera heavy filtering will please you in every shot.
From a camera review perspective, we cannot objectively rate if such a filter is good or bad, beautiful or ugly. Just like fashion, everything goes, as long as it makes you happy. We are interested in the camera’s ability to capture good image data, which is what you ultimately pay for.
"THE ABILITY TO CAPTURE GREAT IMAGE DATA: WHAT YOU PAY FOR"
First, let’s clarify some terminology we’ll be using:
- “image processing”: software work that improves the image data quality
- “image filtering”: software work that changes the style (aesthetic) of the photo.
- “context photo”: a great approximation of what we see
- Including how dark the scene actually is
- Only to provide the context of the shot.
- Not a quality benchmark
Example #1 Daylight Photo
When heavy filtering is applied, the Image Quality can objectively be affected, and we will demonstrate this in the example below. We are looking at a scene through two photos. The context photo shows something very close to what the scene actually looks like. The second is straight out of a smartphone camera (Handset A) and has been heavily processed by the default camera app. The user has no choice.
Even without context, some people might notice see that the photo has been heavily filtered. If the filtering bothers you, it’s still difficult to pinpoint “why” you find it unpleasant. By removing some color layers and look at individual color channels, everything becomes clear as day.
Below you can see that heavy filtering is removing a LOT of details (specific color components represented as gray levels here). That is why many people think over-filtered photos do look weird, if not downright fake. You can see the massive difference with a lot of important details missing.
The filtered data affects how the lighting looks because some shadow clues have been modified. Now it seems there’s a skylight directly above the plants, but it’s not the case at all. Even the color and texture of the hardwood floors have completely changed. You can use the slider below to compare the two images.
"A LOT OF DETAILS HAVE BEEN DESTROYED, FOREVER"
If excess filtering bothers you, it’s because your brain thinks “something is wrong”, and for good reason: it doesn’t look like anything you have seen in the real word: “something is off”. This is very similar to the Uncanny Valley in computer graphics and robotics, where anything less than perfectly natural-looking can trigger a rejection reaction.
If you like the heavily filtered look, the good news is that it takes a few seconds to apply a filter that will do exactly that! However, if you don’t like that look, it’s better to pick a camera that captures a natural/realistic style because the filtered image can never go back to look like the natural one.
That’s why some phones give you the option to save a RAW (unprocessed) that contains less processed image data. However, it takes a lot more space, so we can’t expect most users to turn that on at all times.
Example #2 Night Photo
The same thing also applies for night shots. As you may already know, when it comes to low-light photography, brighter is not always better. And now, we can show you how mindful you need to be with heavy filtering. Consider the scene below, a night scene that you could see near your home, as shown in the context photo below.
Below is a photo that was taken in auto-mode with Handset A, which has heavy automatic filtering. As you can see, the scene is substantially modified.
But whether you “like” the style of Handset A’s photo or not, let’s look at how the data is impacted. Again, we zoom in and peel off some color layers to make things comprehensible. Note: we increased the contrast on these monochrome shots to make it easier to see the nuances on PCs and mobiles.
Again, this is just an example of how image filtering can affect the underlying data of a photo. As data is being exaggerated, amplified or erased by filters, it reduces your options of styles and further modification drastically. Don’t like this look? Well… tough luck. What’s done is done.
In case you wonder, the context photo can easily be filtered to look like Handset A’s photo, since the data is excellent. The opposite isn’t true, the photo from Handset A can never look like the context photo, and would be very difficult to filter into a different style because too much data is already lost.
Conclusion: heavy filtering is better done on a copy, not the original
There’s nothing wrong with filtering or altering photos, but when the built-in camera attempts to do so too aggressively, it robs users of creative opportunities. The more the image is filtered, and the more you are boxed into the style that was chosen by the phone.
If that was your one shot at recording a moment, the camera app cannot possibly know if your intent is to filter it, and how. Research (from Yahoo/Flickr) has shown that many consumers want to record certain memory as close as they see it."THE MORE THE IMAGE IS FILTERED THE MORE YOU ARE BOXED INTO THE STYLE THAT WAS CHOSEN BY THE PHONE"
As of today, even the most advanced AI cannot know the photographer’s intent. Heck, maybe you had no intent at that time but might have an idea later.
If you want to keep your creative options as open as possible, heavy filters would ideally NOT be done in-camera, but on a copy of the photo inside various apps, such as the built-in photo gallery, photo editors or social media sharing apps. It’s a pity to destroy data with in-camera heavy filtering, in the name of convenience. The best compromise in our view is to have the camera app do light filtering, without going too far.
If you have time, read one of the other articles in the same series: Night Photos: Brighter Is Not Always Better.