Human perception of color tone per bit

• Download ColorBitDiffTest.zip - 9.8 KB

Introduction

Simple app where the user can interact and show that it is impossible to see the difference between 2 colors if we only change one bit on one channel at a time on any standard monitor. Today's standard monitor are ~all RGB with 8 bits per channel.

Background

There is a debate on the net actually about digital camera sensor where some manufacturer start to produce 10 bits per channel instead of 8 bits per channel. Many peoples think that a sensor with 10 bits will produce better image. Yes it will, but who can see the difference?

I do not want to argue about the necessity of it because I'm not an expert of the domain. But trying my app would, I hope, put a doubt in any mind that 10 bits should be unnecessary. It appears to me to be powder to the eyes. I suspect that it is a marketing strategy more than anything else.

Using the code

Move slider to see the triangle appears. Can you see the triangle if you only move channel offset from 0 to 1?

You screen is surely an 8 bits per channel. If you can't see the image with a difference of 1 bit with 8 bits per channel, then would you want 10 bits per channel? Why would you want more tone?

Points of Interest

Update a few hours after my initial post of my tip...

The current tips you are reading started from that post: DPRreview Blog. I wanted to enforce my point with an example. After discussion, I realized that I was wrong. Although 8 bits is sufficient to display any image on a screen or paper, like showed here, I realized the need for higher bit count per pixel channel is not for direct display. It is to keep a maximum of information for post processing. Having more bit per pixel per channel when editing will translate by a minimum lost of information. That should translate by a better transformed image or video. There is also another advantage when shooting in compressed format (jpg or mpg) directly where the internal camera processor will also be able to keep higher resolution (lower loss) when processing the raw image data. 

History

2018-03-09, Version 1 of the little application