Color Accuracy in your Space
Hi Matterport Community,
A few Matterport customers have recently noted some color accuracy issues in some of their scan positions. Although most of you are not affected by this issue, we wanted to bring it to your attention, provide some background, and let you know what we are doing about it. Obviously if you encounter this problem, please contact us. There may be ways we can assist with your particular situation.
Although the issue is greatly amplified by white balance, that is not the only issue when dealing with color accuracy. For anyone not familiar with white balance, it’s the processing engine compensating for the color temperature in the scene. The digital camera in your phone has a processing engine in the phone itself and, of course, you can then edit the color, saturation, and adjust white balance to get results that suit your needs. However, with Matterport the processing engine is only in the cloud and post processing is not an option. In fact, everyone using a Matterport camera, whether it’s the Pro or Pro2, is using the same processing engine.
If you’re experiencing a color related issue with the Pro2, it’s not just the camera since no color management is processed in the camera itself. That said - yes, the Pro and Pro2 do have different RGB sensors and, therefore, the color filter array is a bit different as well, which will result in a difference in color with the final Space.
One problem is with the processing engine, and the Matterport team is tirelessly working on this engine to make the image quality as good as possible for both cameras. The challenge is getting a single processing engine to accurately resolve color for every scene imaginable.
The other problem is calibrating the processing engine to white balance a scene or shift the color slightly to compensate for the color cast by a light source when the scene has more than one type of light source. As you can see in the chart below, incandescent bulbs will vary in color temperature, but for the most part they’re close to the same color. When daylight is introduced into the scene, especially on overcast days, the color cast is blue instead of yellow. In order to compensate, or white balance, for blue, yellow is added to the scene and, inversely, in order to compensate for yellow, blue is added.
Imagine a scan with blue light coming in from the window and yellow light from a floor lamp. If the color shift is made to compensate for the lamp, the window will be even more blue and if the color is shifted to compensate for the window the lamp will be even more yellow. It’s a very complicated scene to calibrate and it only becomes more complicated as the difference in color temperature becomes more extreme.
As I previously mentioned, we really are continuously working on this problem to find the best solution that will work for all situations. In some cases the contrasting color temperatures are too great and what we end up with are less than desirable results. Please bring these to the attention of the support team so they can be documented and used as future samples when further calibrating the processing engine.
Thanks very much for your understanding and I apologize for the inconvenience this issue has caused some of you. Again, please make the support team aware of the Space with too much yellow or too much blue so we can take it to the dev team working on configuring the processing engine.