Why #nofilter always is a lie.

Don’t trust Bayer.

fxp
Photo School

--

Twitter, Facebook—nowadays, you find the #nofilter tag on pretty much any platform that allows photo sharing. Most likely, though, it originated on Instagram, where it’s part of the concept to apply vintage filters to the photos before sharing them.

#nofilter seems to be a hipster tag which is supposed to convey that you see the moment shared in a very real, pure form. There are even Websites out there who claim to find pictures which are not really #nofilter, but indeed #yesfilter.

The problem: Such as thing as #nofilter does not exist in digital photography. Ever.

Bayer what?

What is a filter? Basically, it’s anything that modifies a signal, in our case: anything that modifies the “picture” as it comes from the camera sensor.

CC BY-SA 3.0
Cburnett

Now let’s take a standard digital camera. The actual photo sensor is unfortunately color blind, it only detects different brightness levels. To capture color photos, the manufacturer puts an array of color filters on top of it (as shown on the left), the so-called Bayer Matrix—the first filter that your photo passes.

After taking a photo, the camera still receives a black and white photo (remember, the actual sensor is color-blind), but now it knows that a bright pixel behind a green part of the filter means that it’s very green in that region of the photo etc. A comple algorithm called de-mosaicing brings all this information together and generates a (first) color image.

That’s not where the image processing ends. This first picture looks dull and ugly, nothing like the accurate, crisp photos you’re used to see on your display. It needs to be white-balanced, color-corrected, sharpened etc. Some even automatically add tonemapping (a.k.a. “HDR”) processing. HP shows a nice overview of the whole process.

All this “magic” happens inside your cell phone, PC, or camera (and is one of the reason why cameras need powerful computing units) before you ever get to see the photo.

Does that mean that…?

Yes. There is no such thing as #nofilter because there is no defined neutral processing. Every digital photo has been processed and filtered. Even when you use the default setting of your camera (maybe labeled filter: none or off), your photo has gone through a multitude of filtering and transformation operations, in that case the ones that your camera/ software manufacturer considers to be “neutral”. It doesn’t make the photo a “pure” or “unbiased” view of the world.

Often the opposite is acutally true, and processing brings the image closer to what we perceive as reality. White balance is an example: candle light is very red, and light in the shadow more blueish, but would you perceive a photo with a strong red or blue tint to be a natural representation of a scene? Or take exposure compensation: you see the clouds in the sky as well as the dog in the shade on a bright sunny day because your eye adjusts for the strong contrast in the scene. Without any processing you’ll never have a photo which represents that.

In short, it would make more sense to tag #nofilter photos with #defaultfilteronmycamera—but it doesn’t change the fact that there no closer or farther from reality than many other photos.

What about analog? Foveon?

The Foveon sensor doesn’t have a Bayer matrix but still color filters above the sensor panes. It also needs the other post-processing described.

To add the #nofilter tag, you need a digital image. ;) Similar processing as described above is done when scanning a negative, slide or print.

Maybe it would be fair to label (physical, un-scanned) Polaroids or slides as nofilter.

--

--

fxp
Photo School

Director of Data Analytics at vroom.com (buy and sell your car online and get it shipped home). I love all things data, bread, and photography.