Raspberry Pi 4 with 8GB of RAM...

Released yesterday, there is now a Raspberry Pi 4 with 8Gb of RAM, for $75 in the US!
(See the Raspberry Pi blog).
And it comes along with a beta-64 bit OS, named Raspi OS, that targets Raspberry Pis 3 and higher.

I tried it (on a Raspberry Pi 4, with 4 Gb of RAM), it works fine, and fast!
This beta version does not come with Java installed, but a simple sudo apt-get install default-jdk installs it (JDK 11) in a couple of minutes.
I cloned a repo (https://github.com/OlivierLD/raspberry-coffee.git) and built it without any problem or error.

Now, I might wait a bit longer to get a new Raspberry Pi. As it is now, it should be able to support 16 Gb of RM, I'll wait a bit, and see...

Anyway, that makes yet another good reason NOT to get a Mac.
Steve Jobs vs Eben Upton..., I vote for Eben Upton, biiiiiig time.

Using OpenCV to downgrade an image

This is a bunch of comments on the code visible on github. The goal here is to downgrade a color image to display it on a led matrix, like a small OLED screen here, an SSD1306. We will use OpenCV, in Java. Here are the steps we follow:

• We start from the colored image
• We turn it to gray
• We thresh it
• We resize it (smaller)
• We store it in a file, custom format
• We can then display the image on the led matrix (oled screen here)

Original	Grayed
Threshed	Resized

The level of details of the final display is obtained during the threshold part.
See in OpenCVSwingColor2BW.java:

    // threshold
    Mat threshed = new Mat();
    Imgproc.threshold(gray,
            threshed,
            150, // 127,
            255,
            0);

Tweaking the thresh parameter (150 above) leads to different results.
The final result is stored in a binary file (image.dat).
The matrix used here is 128x64 pixels big. The file will contain 64 lines of 2 longs.
A Java long has 64 bits, 2 longs make 128 bits, that's all we need to encode one line of 128 leds on the screen.
See the code in OpenCVSwingColor2BW.java for details.

Adios Papou