Blog Archive — Rully Pratama

Going Back to an Old DSLR in the Smartphone Era: When Technology Makes Us Forget How to Shoot

Mon, 13 Apr 2026 00:00:00 +0000

Last Eid in March 2026, I bought a Canon EOS 50D on a whim. This used to be my dream camera back in the day, and now it’s incredibly cheap. My intention was simple: I wanted to photograph my kid with a “proper” camera.

My photography journey actually started back in 2012 with a Canon EOS 20D. However, since I sold that camera in 2016, I practically stopped using DSLRs.

My complete Canon EOS 20D set before I finally sold it in 2016.

For years after that, I solely relied on smartphones and mirrorless cameras to capture moments. Everything was instant and easy.

But the expectation of reminiscing with a DSLR wasn’t as beautiful as reality. When I looked at the shots from the 50D, I was quite shocked. Most of my photos were either out of focus or the exposure was a mess. It turns out, I’ve been so spoiled by the convenience of modern technology that I “forgot” how to take photos.

The Canon EOS 50D paired with the EF 40mm f/2.8 STM pancake lens.

Here are a few fundamental things that I apparently had to relearn.

1. Forgetting How to Focus (Focus & Recompose)

Back then, with very limited focus points, we were used to the focus and recompose technique. We’d lock the focus on the center point, then slightly shift the camera to adjust the composition.

Now? I’m so used to face AF and eye AF on smartphones and mirrorless cameras that always nail the focus wherever the subject is. My manual focusing instinct is completely gone, which resulted in a lot of blurry photos of my kid.

2. Guessing the Exposure

The ease of seeing exposure in real-time on a smartphone screen actually ruined my lighting instincts. What you see on the screen is the final result.

When using an old DSLR through the optical viewfinder, I just took the numbers from the camera’s built-in light meter as is. I completely forgot to account for lighting challenges like backlight, subjects partially under the shade, or sunrise/sunset conditions.

3. Honest Old Sensors vs Computational Photography

This is where the difference is most noticeable. Modern smartphones use computational photography to produce perfect photos right out of the box. HDR, colors, and sharpness are all adjusted automatically.

On the other hand, the photos from an old DSLR are very “honest”. The resulting files often look flat and need a touch of editing. This hassle directly impacts the next point.

Previewing photos on an old LCD screen with mediocre resolution.

Using a smartphone, photos can be easily shared to social media or WhatsApp right that second. The moment is still fresh.

With an old camera, the workflow is very long. I have to connect the camera to a PC using cables (or a card reader), transfer the files, do some editing, and then send them to my smartphone. Because of how long this process takes, the captured moment has often passed and the photo no longer feels relevant to share.

5. ISO Limitations and Noise Tolerance

Back then, we were very careful about raising the ISO so the photos wouldn’t be filled with noise. Now, smartphones have Night Mode that can turn night into day, and modern mirrorless cameras still look very clean at ISO 6400.

When photographing my kid indoors using the 50D, I just realized how limited this old sensor is. Pushing the ISO above 800 or 1600 produces quite distracting noise. I forgot that back then, shooting indoors often required the use of an external flash.

Conclusion

In the end, I sold the EOS 50D again because I realized I wasn’t quite ready to start using a DSLR again. As a replacement, I bought another dream camera of mine: the Panasonic Lumix GX7. Even though it’s not as advanced as today’s mirrorless cameras, at least I can confirm the focus and exposure right then and there, before pressing the shutter button.

Stay curious! ✨

Installing MikroTik RouterOS on a Headless Mini PC via Serial Console

Mon, 30 Mar 2026 00:00:00 +0000

A few months ago, I decided to switch from pfSense to MikroTik RouterOS. The device I use is a small network appliance based on the Intel Atom CAF 1020 C2358, which I snagged on Tokopedia for around 900k IDR.

The Intel Atom CAF 1020 C2358 mini PC.

My reason for leaving pfSense was quite simple: I really like RouterOS’s lightweight nature, incredibly fast boot time, and the more granular customization offered by WinBox.

This device is very power-efficient, drawing only about 10 watts even under load. The cooling system is passive, so the device’s body will feel slightly warm to the touch. As of writing this, this tiny router has been running for almost a year without any issues at all.

The Headless Installation Challenge

Although the hardware is very satisfying, this appliance doesn’t have any video output ports (VGA/HDMI) at all, meaning it’s completely headless. When I installed pfSense back then, the process went very smoothly because the pfSense installer natively supports booting via a serial session.

Unfortunately, installing MikroTik RouterOS is not that easy. Its default installer won’t show anything on the serial console automatically. So we need a little modification to redirect the installation display to the serial port.

The Solution: Modifying syslinux.cfg

The workaround is actually quite simple. We just need to edit the syslinux.cfg configuration file located inside the installation media (like a USB flash drive).

Open the installation media from another PC, then find and edit the file. Add the SERIAL and CONSOLE commands at the very top, then insert the console=ttyS0,115200n8 parameter on the APPEND line.

More or less, the modified syslinux.cfg will look like this:

$ cat /mnt/syslinux.cfg
SERIAL 0 115200
CONSOLE 0

default system

LABEL system
 KERNEL linux
 APPEND console=ttyS0,115200n8 load_ramdisk=1 -install -hdd

After the file is saved, boot the mini PC again using the modified installation media. Now, the MikroTik RouterOS installation menu will appear perfectly in the serial console application (like PuTTY or Tera Term) connected to your PC.

Happy tinkering! ✨

Why Do the Same RAW Photos Look Different?

Sat, 21 Jun 2025 00:00:00 +0000

The RAW format is certainly no stranger to those of you in the world of photography. RAW has its advantages, including excellent shadow recovery, flexibility in color grading, and details that simply can’t be found in the JPG format. These benefits are because RAW stores the raw data from the photo taken.

The analogy of “raw” for the RAW format and “cooked” for the JPG format seems very fitting. Imagine food: if you have a cooked meal and share it among several people, the taste each person experiences will be relatively the same.

However, if raw ingredients are shared, each person will process them in their own way. The results and flavors can vary widely, depending on taste, cooking experience, and the spices used.

A bit about how cameras capture light

⚠️ Image not found: bayer_filter.avif

The sensor is the part of the camera responsible for capturing light. Unfortunately, the sensor can only distinguish between light intensities (bright/dark), without any color information. To overcome this, a Bayer filter is placed in front of the sensor.

The Bayer filter is shaped like a grid. Half of the grid consists of green filters, while the other half consists of red and blue. This filter only allows light of the same color to pass through to the sensor. With this data, the sensor can eventually capture color information.

But the problem is, each pixel behind the filter only has color information for one color.

Raw data illustration from Bayer filter. Each pixel captures luminosity (light or dark) information without color.

A demosaic process is required to change information that was previously mosaic-like (note the black spots on the pixels in the image) into an image with appropriate color and lighting as it should be.

Demosaicing is just one step of several pipelines a camera must go through to capture light and turn it into digital information. There’s white balance, color correction, gamma correction, and denoising. Each step, if performed differently in method and intensity, will produce a different image.

Camera processing pipeline.

Different apps, different ways to process RAW data

Comparison of software’s raw image processing result. Notice that each software gave their own ’taste'.

Remember, a RAW photo is data from the camera sensor’s reading that is still truly raw. Apps need to apply their own processing pipelines to convert the raw Bayer data into a meaningful photo.

Here I compare 4 images: 1 JPG result from the camera, and 3 from applications capable of processing RAW. The results… are quite varied, to say the least.

The Camera JPG will process according to the profile selected at the time the image was taken, in this case, Canon Standard. Other enhancement features like Auto Lighting Optimizer or noise reduction will affect the resulting JPG.

Adobe Lightroom, as the de facto photo editing app for many, uses the Adobe Color profile for every RAW photo it processes. I like this profile because it aims to produce relatively uniform images even when taken with different camera brands and models.

Windows Photos, which is more often used just to display images without editing, appears to process the RAW photo with lower exposure levels than the others. In my opinion, this app is just basic for displaying RAW photos.

Finally, Apple Aperture produces an excellent image. It’s similar to the image produced by Lightroom with its Adobe Color profile. However, for me, the flower color produced by Lightroom is more true to reality than Aperture’s.

Fun fact: RAW files also store small JPG images as previews

Even though a RAW photo is unprocessed raw data, cameras still embed a small JPG format image inside it. The goal might be as a backup in case an application cannot process the RAW data; it can still display this JPG image.

I tried examining a Canon RAW file using the MediaInfo app, and it shows that besides storing the raw data (TIFF), it also carries a JPG file with a smaller size and resolution than the raw data itself.

Not only containing image raw data, camera raw also bought small JPG data as thumbnail.

Another way to prove this JPG image exists is to try opening the RAW photo using Windows Photos. For the first second or two when the image appears, that’s the JPG image. The app uses this JPG while the RAW data is being processed. Once the result is obtained, the image is replaced with that processing result. This makes it seem like the image is changing.

Embeded JPG helps with initial loading. After the raw data has been processed, the image is swapped, hence the ’transition'.

This also applies in Lightroom. In Library mode, it displays the JPG preview by default (you can see Embedded Preview written in the bottom right corner), while in Develop mode, it displays the image resulting from RAW data processing.

The same behavior can be observed also in Adobe Lightroom.

Thanks for reading and stay curious! ✨

Is the Lumix GX7 ISO Invariant?

Fri, 06 Jun 2025 00:00:00 +0000

My beloved Panasonic Lumix GX7 camera.

Throughout 2012 to 2015 was when I first started getting into photography. The gear I used was a Canon EOS 20D. Since then, my preference for camera brands has always leaned toward Canon.

During that same period, discussions about ISO invariance in camera sensors began to emerge on the internet. I was disappointed because, at that time, test results showed that cameras with Sony-made sensors (like Nikon and Sony itself) had ISO invariant capabilities—marked by excellent shadow recovery—while Canon with its in-house sensors did not.

The defense from Canon users at the time was that Canon was widely used for portraiture, so color science and skin tones were more important than dynamic range. It was a bit of denial, in hindsight, but what can you do? That’s fanboyism for you.

Fast forward to 2025, I finally decided to migrate from a Canon EOS 50D to a Panasonic Lumix GX7. Though it’s an older model, it’s been my dream camera for a long time. Since Panasonic equipped this camera with their own sensor, I was curious whether this sensor is ISO invariant like Sony’s, or not like Canon’s sensors.

To answer this question, I conducted my own experiment.

TL;DR: This camera has an ISO invariant sensor.

It can be assumed that all cameras using Panasonic sensors behave similarly, though this remains to be proven through further testing.

What is the difference between ISO variant and invariant cameras?

Before diving into the testing, it’s worth briefly looking at ISO and the difference between the two.

ISO in film refers to the sensitivity level of the material to light. The higher the ISO value, the more sensitive the film is to capturing light. However, high ISO has the drawback of being grainy. Grain comes from light-sensitive particles. The larger they are, the more sensitive they become, but the grainier they appear.

In digital sensors, ISO refers to the level of amplification of the light signal captured by the sensor, which is later stored as a digital file. Just like in film, high ISO in digital sensors also has drawbacks. A high level of amplification doesn’t just strengthen the light signal of the image we want; it also amplifies the noise.

Here is the path light takes to become a digital file:

Stage	ISO Variant	ISO Invariant
Sensor	Light is captured and produces electrical voltage	Light is captured and produces electrical voltage
Analog Amplification	Amount matches the ISO used	Unaffected by the chosen ISO
Analog to Digital Conv.	Electrical voltage is converted into data	Electrical voltage is converted into data
Digital Amplification	Minimal	Amount matches the ISO used
Output	A digital file	A digital file

From the comparison above, the striking difference between the two is that in ISO variant sensors, amplification is performed during the analog stage (electrical voltage). Therefore, once converted into digital data, shadow recovery will result in a very noisy image.

In ISO invariant sensors, amplification happens at the digital stage. Thus, it’s the same whether the amplification is done in-camera or during post-processing using software like Adobe Lightroom, since both process the same digital data.

Methodology

Two images were taken in RAW format: one with appropriate exposure using a high ISO of 3200. The other was taken at base ISO, in this case 200. The image taken at base ISO will be underexposed, so its exposure level will need to be raised using software, and then we will observe the noise levels in both.

From these results, if both images eventually have identical noise levels—one photo being no more noisy than the other—then it can be concluded that the sensor in this camera is ISO Invariant.

Comparison of testing image & reference image, and it’s respective exposure settings.

Results

After the test image was brightened by about +4.20 EV using Adobe Lightroom, the noise structure remained the same as the reference image. Subjectively, the test image looks slightly more noisy, but not significantly so.

From these observations, it can be concluded that the sensor in the Panasonic Lumix GX7 is ISO invariant.

The testing image after being adjusted to match the reference image’s brightness level. Note that the grain structure remains the same, indicating that Lumix GX7 is ISO invariant.

Knowing that this camera is ISO invariant, users no longer need to fear that their shots aren’t perfectly exposed, as long as they use the RAW format. Exposure can still be adjusted during post-processing using software.

Stay curious! ✨

YouTube Not Saving History? Why?

Sat, 22 Feb 2025 00:00:00 +0000

A while ago, I was following an Age of Empires II tournament on YouTube. Since the tournament lasted for hours, I obviously didn’t watch it all in one sitting. Instead, I would usually continue in my next spare moment.

YouTube usually saves our watch history. This includes when a video isn’t watched until the end; YouTube will start playing from the exact same duration where we stopped in the previous session.

Then something strange happened. YouTube didn’t mark the session! When I tried to rewatch, the video played from the beginning. Even the history playlist on YouTube didn’t record the videos I had watched until the end!

Initially, I thought it was because the YouTube app on my phone was not up-to-date. But this anomaly also occurred on the PC I usually use.

After I searched on Google, it turns out the cause waaaaas…

DNS aka Domain Name Server

Yes, at that time I had just installed AdGuard Home on our home local network. There was one domain that is actually important for YouTube but got blocked:

s.youtube.com

The solution is easy: whitelist or allow the domain s.youtube.com in AdGuard Home, Pi-hole, or NextDNS so that YouTube history works again.

Dual Boot Windows with VHDX

Thu, 13 Feb 2025 00:00:00 +0000

Windows boot menu, showing boot entry from VHDX medium.

Playing around with dual boot installations in Windows is fun. What’s not fun is having to deal with partitions. Not only is it tricky, but it also carries the risk of data loss if not done correctly.

Fortunately, Windows can be installed inside a VHDX file. A VHDX file is a virtual disk format that can be used to store files—whether they be documents, media, or even a full Windows installation. This allows for Windows dual booting without the need to repartition your HDD/SSD.

Assuming your laptop or PC already has Windows installed and you want to add another Windows installation, here are the steps.

Step 1: Prepare the `.vhdx` File

Open Disk Management (Win + X → “Disk Management”).
Click Action → Create VHD.
Save it to your desired location, for example, C:\vhdx\win11.vhdx.
- Choose the VHDX format if you want to install Windows 8 or newer; choose VHD for Windows 7.
- Set the disk size, ensuring it’s sufficient for the Windows installation you want.
- Choose Fixed size for better performance.
Right-click on the newly created .vhdx disk, click Initialize Disk, and choose GPT if using UEFI, or MBR if using BIOS.
Right-click on the initialized disk and create a New Simple Volume. Format it as NTFS and assign your desired drive letter, for example, V:.

Step 2: Mount the Desired Windows Installation ISO

Right-click the ISO file of the Windows operating system you want to install (it can be the same version as the one already installed, or a different one). Then select Mount. Take note of the drive letter assigned, for example, E:.

Step 3: Installation Process

The installation process is performed while the first Windows is running. This is possible thanks to the DISM tool. Here are the steps:

Open Command Prompt as an administrator.
Check the Windows edition you want to install from the .iso file using:

dism /Get-ImageInfo /ImageFile:E:\sources\install.wim

Install Windows to the .vhdx file you created earlier:

dism /Apply-Image /ImageFile:E:\sources\install.wim /Index:1 /ApplyDir:V:\

Replace drive E: with the drive letter of the mounted ISO.
Replace drive V: with the drive letter assigned to the VHDX file.

Step 4: Add to Boot Entry

Create a boot entry pointing to the .vhdx file using bcdedit:

bcdedit /copy {current} /d "Windows 11 VHDX"

Note the GUID (e.g., {xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx}) that appears after running the command above.

Set the device and path to be added to the boot entry:

bcdedit /set {xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx} device vhd=[C:]\vhdx\win11.vhdx
bcdedit /set {xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx} osdevice vhd=[C:]\vhdx\win11.vhdx
bcdedit /set {xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx} detecthal on

Step 5: Test Bootloader

Restart or reboot your system.
On the boot menu screen, select Windows 11 VHDX (or the name you chose in step 4) to boot into the new Windows installation.

Congratulations, your system now has two Windows installations. Dual booting using VHDX is perfect for:

Those who like to experiment with different Windows versions without worrying about partitions, but want to run them on bare metal.
Separating Windows installations for two or more purposes, such as one for personal use and another for work.
Having fun nostalgia with older Windows versions, like Windows 7 or Windows 8.

Good luck! ~ ✨

Energy-Efficient Banana Pi NAS

Fri, 23 Feb 2024 00:00:00 +0000

Banana Pi M1 as NAS device.

This time I will share my experience migrating my NAS device from an Unraid-based build-up PC to a Banana Pi M1 single-board computer (SBC). The motivation for this migration was that I no longer needed a large amount of storage space and wanted to save on my home server’s electricity consumption. This post is related to the energy savings I wrote about previously.

Background

Although I’ve changed NAS hardware several times, the operating system I used was always Unraid. Its main advantage is that it can combine various hard drives of different capacities into a single array. Each hard drive has an XFS partition that can be read on most Linux devices because Unraid writes complete files to a specific hard drive rather than splitting them bit-by-bit like traditional RAID.

This configuration was attractive to me because I could buy a 2 TB hard drive, for example, at a quite affordable price, and then easily add another hard drive of a different size when I needed more space. I primarily used this storage to store media (movies and series).

However, my storage needs changed when I discovered Stremio, a streaming platform that doesn’t require local storage. Now my needs are only for backing up important files, which doesn’t require much space. It felt wasteful to use an HP 8300 SFF build-up PC, which had to run 24/7 as a NAS device.

Why Banana Pi?

I’ve owned this device since 2016. The Banana Pi has an advantage that the Raspberry Pi lacks: a SATA port, so an internal hard drive can be used directly without needing a USB port. Additionally, this board also has a gigabit ethernet port, which complements the SATA port when transferring data to other devices.

I’ll refer to this Banana Pi-based NAS as BananaNAS.

Here are the specifications for both.

Component	BananaNAS	Unraid NAS
CPU	Dual-Core Cortex-A7 ARM @ 1GHz	Quad-Core Intel Core i5-3470 @ 3.5GHz
RAM	1GB DDR3	16GB DDR3
Storage	1 x 2TB 2.5" HDD @ 5400 RPM	2 x 128GB 2.5" SSD & 2 x 2TB 3.5" HDD @ 5400 RPM

Banana Pi M1 after being equiped with supporting accessories, such as HDD and mini PSU.

Test Results

To test a NAS, I wanted to know how fast the transfer speed is over ethernet using iperf3. Meanwhile, the read/write speed on the hard drive via the ethernet network will be tested using NASTester.

iperf3

The first test I performed was a bandwidth test using iperf3. This tool is known to be CPU-bound, meaning the results are influenced by the strength of the CPU used. This test can provide an overview of the difference between Unraid NAS and BananaNAS, which naturally has a weaker CPU than my Unraid NAS.

The command I used was as follows, with the parallel flag (-P) varying from 1 to 4.

iperf3 -c 10.20.30.50 -t 60 -P 1

Network transfer speed test result using iperf3.

The result was predictable: the BananaNAS was 26% slower than the Unraid NAS with its much higher specs. With an average transfer speed of 687.5 Mb/s or about 86 MB/s, it’s still within the range of 5400 RPM 2.5-inch hard drive speeds, which are between 80 to 100 MB/s.

NASTester

Next, I’ll compare the read and write performance from my main PC to each NAS. The connection between the PC and NAS uses CAT 6 UTP cable to ensure a stable gigabit connection.

This test was performed 7 times on each NAS. Each test, including read and write, used different file sizes, ranging from 100MB to 800MB. The goal was to understand the relationship between RAM capacity and transfer speed, as I suspected a large RAM allows files to be cached before being written to the hard drive.

Read and write speed test result using NASTester.

The result: BananaNAS’s transfer speed was 50% slower than the Unraid NAS for both read and write operations. The write speed of BananaNAS for small files appeared slower compared to larger files. Meanwhile, the Unraid NAS appeared more stable across all transferred file sizes.

Power Consumption

Since switching to BananaNAS, the electricity consumption for the NAS device has dropped from 36 watts to only 4 watts! Assuming the NAS device is on 24/7, the monthly electricity cost for the NAS can be reduced by up to Rp50,000.

Banana Pi NAS in its resting place.

Conclusion

Migrating from an Unraid NAS to a BananaNAS sounds like a downgrade. However, considering my needs have changed—from storing large media files to just a backup device—I feel the BananaNAS is quite sufficient.

The monthly electricity savings are very attractive, especially since I built the BananaNAS using a device I already owned, so there’s no need to worry about ROI.

Stay curious! ✨

Replacing Old PC with Energy-Efficient Mini PC: Is It Financially Profitable?

Thu, 25 Jan 2024 00:00:00 +0000

A mini PC.

Since the start of the COVID-19 pandemic in 2020 until now, I have been working from home. I use my own electricity and internet connection at home, even though the costs are reimbursed by the company I work for. The PC I use for work usually stays on from 8 AM to 10 PM (including for entertainment after work). This means my PC consumes electricity for 14 hours a day, 5 days a week. It occurred to me: would switching to a mini PC with similar performance but lower power consumption benefit me financially?

To answer that question, let’s calculate the electricity consumption of my old PC compared to a more energy-efficient mini PC.

Item	Old PC	Mini PC
Average power consumption (W)	75	10
Duration turned on (h)	14	14
Daily kWh consumption	1.05	0.14
Monthly kWh consumption	22.05	2.94
Electricity price per kWh (Rp)	1,800	1,800
Monthly electricity cost (Rp)	39,690	5,292

From the table above, it can be seen that the savings from replacing the old PC with a Mini PC is Rp34,398. Quite a significant value, actually. But when we compare it with the purchase price of the Mini PC and how long it takes for the return on investment (ROI) to occur, is it still profitable?

Item	Old PC	Mini PC
Purchase price	–	7,299,000
Selling price	5,000,000	–
Budget deficit	–	2,299,000
Monthly savings	–	34,398
ROI in months	–	66
ROI in years	–	5.5

Assuming my old PC is sold for 5 million rupiah, and the remaining cost for the Mini PC purchase is covered by the monthly savings from reduced electricity usage, I would only reach ROI in 5.5 years! This is a very long duration for a technology device purchase. In my opinion, upgrading technology devices should ideally be done every 3 or 4 years.

Even in a scenario where the Mini PC purchase is intended for productive and profitable purposes, using an existing PC—even if it’s slightly more power-hungry—is still more profitable. Therefore, if I ever buy this Mini PC in the future, it won’t be for the savings, but because of curiosity 😝

Stay curious! ✨

Short Review: TP-LINK TL-AP1750C

Sat, 13 Jan 2024 00:00:00 +0000

Still on the subject of my hobby of buying and using secondhand networking equipment, this time I’m trying out a ceiling access point made by TP-LINK, specifically the TL-AP1750C series.

This access point is actually intended for corporate/office use. Additionally, this device is sold for the Chinese market, as indicated by the firmware only supporting Mandarin. However, on the upside, I managed to get this AP at a very affordable price, which is 185,000 rupiah (~$12 USD) via Tokopedia.

TL-AP1750C, front facing.

Pros and Cons

Here are some advantages that make this access point interesting to buy:

Its price is very affordable compared to its specifications, only 185k rupiah.
Supports 5GHz WiFi AC, has a Gigabit LAN port, and is powered via Power over Ethernet (PoE).
The default firmware has the features I’m looking for, including Band Steering, client isolation, VLANs, etc.

Meanwhile, the disadvantages to consider include:

According to my search results, there is only one seller selling this device on Tokopedia.
The firmware only supports Mandarin, making it a bit difficult to configure.
Since it’s intended for the Chinese domestic market, the WiFi region cannot be changed, so the WiFi AC signal is not detected by some devices, especially iPhones.
There is no power port, so the only way to power this device is via Power over Ethernet (PoE). The PoE voltage required by this device must be exactly 48 volts, so you’ll need to purchase additional equipment like an adapter and a PoE injector.

TL-AP1750C, rear facing.

For those who happen to own this device and need a user manual, you can download it via the following links:

Turn an Old Router into a Managed Switch with OpenWRT

Fri, 01 Dec 2023 00:00:00 +0000

A managed switch is a networking device that is still uncommon in home network topologies. Nevertheless, this type of switch actually holds a wealth of features not found in unmanaged switches, one of which is VLAN configuration. This feature becomes very important if we want to save on UTP cables or the number of ports. With VLANs, a single cable or port can carry networks with multiple subnets simultaneously.

Recently, I migrated my router to Mikrotik RouterOS, which I installed on a mini PC using Proxmox. The mini PC I used has a standard motherboard, so it only has 1 LAN port. In general, a router requires 2 ports: one for WAN and the other for LAN. VLANs can be the solution.

Unfortunately, the cheapest managed switches on the market are still in the range of Rp300,000 (~$20 USD), a price that is still too high for me for a home switch device. Since I enjoy buying secondhand networking equipment, the idea struck me to “marry” an old router with OpenWRT to transform it into a switch.

Secondhand TP-Link TL-WR1043ND and TL-WR741ND as managed switch.

The devices I bought were the TP-Link TL-WR1043ND and TL-WR741ND. I got these devices for Rp65,000 and Rp45,000, respectively. The first device has gigabit ports, making it suitable for local network or LAN VLANs, while the other is capped at 10/100, making it more suitable for WAN VLANs, where ISP internet speeds are typically no more than 100Mbps.

I won’t write the OpenWRT installation procedure as there are already many on the internet. As for VLAN configuration, I might discuss it in a future post. However, the most important thing to remember when using OpenWRT on older router devices is the RAM and flash storage capacity. Devices with 4MB RAM and 32MB flash are no longer supported by the latest versions of OpenWRT, so the solution is to use older releases, such as 18.04 or 19.07. Although these are older versions, they do not diminish functionality at all.

TP-Link TL-WR1043ND dan TL-WR741ND, old yet still fully functional.

I always recommend doing research, research, and more research before buying. Because every router model can behave differently even if made by the same manufacturer. For example, the WAN port on the TP-Link TL-WR1043ND can be used for VLAN configuration (bringing the total usable ports to 5), whereas on the TP-Link TL-WR741ND, the WAN port cannot.

That’s all for this short post. I hope it can inspire friends to reuse electronic devices that are still worth using. Because if not, these devices will become electronic waste that can damage the environment.

Stay curious! ✨

Fan Mod on HBA Card LSI SAS 9210-8i, Making the Chip Much Cooler

Mon, 09 Oct 2023 00:00:00 +0000

Recently, I moved my Network Storage System (NAS) components from a mini ITX case to a refurbished HP 8300 Small Form Factor (SFF) PC. I’m quite happy with this migration because the HP 8300 has more room, allowing me to add more storage drives in the future. Additionally, the airflow in this PC is better, so components like the CPU and HDD are much cooler.

To accommodate more drive bays, I added an LSI 9210-8i HBA Card, which serves to ‘multiply’ storage ports. This HBA card is designed for enterprise use, where servers are kept in very cold rooms with excellent airflow. Consequently, this HBA card is not equipped with a fan, only a passive heatsink.

1 - HBA card, before modification; 2 - after modification, fan power is connected to SATA.

When working, this card tends to get hot. A simple way to check: touch the heatsink with your finger; if it feels too hot to hold, then it likely needs active cooling.

This modification might also be applicable to other series of HBA cards. These are the materials and tools I used to perform this modification:

Small 4 cm 12V fan, two wires - Tokopedia link
Female SATA power extension cable - Tokopedia link
4 sets of M3 X 30mm or M3 X 40mm screws - Tokopedia link
Heat shrink tube, sized according to the fan cable - Tokopedia link
Soldering iron, solder, scissors, and electrical tape

1 - 4 cm fan 12V; 2 - SATA power extension cable.

Fortunately, the holes on this card’s heatsink are perfectly spaced for a 4 cm fan. So, installing the fan is very easy:

Remove the plastic clips holding the heatsink. Press the clip from the top, then cut the retainer at the bottom with scissors. This part will not be used again, so it’s safe to discard.
Cut 2 strips of electrical tape, make a hole the size of the screw shank, then stick it over the clip holes on the bottom of the card. This serves to keep the screw from touching the card components, thus avoiding short circuits.
Position the fan so its holes align with the clip holes you just cleared. Then tighten by installing 2 screws and nuts.
Solder the fan wires to the female SATA power connector. Ensure the fan’s red wire meets the yellow wire, and the black wire meets the black wire. Don’t get them reversed!

Those are the steps for the modification. With the addition of the fan, the chip on this HBA card can be cooler, and hopefully, it will last longer.

Stay tuned! 🚀

Dark Themes for Frequently Used Applications

Mon, 17 Jul 2023 00:00:00 +0000

When different apps are using the same theme color, they seem more coheren.

For me, dark mode is more comfortable and looks better than light mode. But unfortunately, many applications I use have their own separate dark versions. That’s until I found the Catppuccin themes which are available for many applications. As a result, the application interface becomes coherent and looks “unified.”

This post will be updated periodically.

Firefox

Catppuccin theme for Firefox.

Cascade userChrome + Catppuccin theme
The userChrome I use can be downloaded here, with the installation method like this. But please forgive me if my userChrome is messy 🤣

Windows Terminal

Catppuccin theme for Windows Terminal.

Visual Studio Code / VSCodium

Catppuccin theme for VSCode.

Themes can be downloaded via the marketplace, or
Catppuccin for VSCode

Notepad ++

Catppuccin theme for Notepad++.

Catppuccin for Notepad++

Spotify

Catppuccin theme for Spotify.

Use Spicetify to change themes in Spotify
Catppuccin for Spotify

Thunderbird

Catppuccin theme for Thunderbird

Catppuccin for Thunderbird

OBS Studio

Catppuccin theme for OBS Studio

Catppuccin for OBS Studio

Flow Launcher

Catppuccin theme for Flow Launcher

Catppuccin for Flow Launcher

Besides Resolution, Also Pay Attention to Diagonal Size When Buying a Monitor

Tue, 27 Jun 2023 00:00:00 +0000

A while ago, I bought a Lenovo ThinkVision P24q-10 monitor – a 1440p resolution monitor with a 24-inch diagonal size – to replace my broken 1080p monitor. I had researched various things beforehand, ranging from panel type, resolution, and dimensions, to reviews scattered across the internet. However, when the monitor I ordered arrived, I was disappointed because the display space I got was not what a 1440p monitor should be. Why could this happen?

I chose a 1440p monitor because I wanted to upgrade from the 1080p I previously used. With a higher resolution, I hoped to get a wider display space than 1080p. I chose the 24-inch diagonal size because at that time I thought I would like the high pixels per inch (PPI) look, hoping for sharp images.

Windows Display Setting - Scale.

My hopes were immediately dashed when Windows automatically selected a Scale of 125% for my monitor. Let’s first calculate the PPI of a 24-inch 1440p monitor with the following formula:

$$PPI = \frac{diagonal~in~pixels}{diagonal~in~inches}$$$$PPI = \frac{\sqrt{width^{2}+height^{2}}}{diagonal~in~inches}$$$$PPI = \frac{\sqrt{2560^{2}+1440^{2}}}{24}$$$$PPI = \frac{\sqrt{8627200}}{24}$$$$PPI = \frac{2937.2}{24}$$$$PPI = 122.4$$

With a scaling of 125%, the perceived PPI will be:

$$Perceived~PPI = PPI + \left ( PPI \times \left ( 1-1.25 \right ) \right )$$$$Perceived~PPI = 122.4 + \left ( 122.4 \times \left ( 1-1.25 \right ) \right )$$$$Perceived~PPI = 91.8$$

A PPI of 91.8 is the PPI for a 1080p resolution monitor with a 24-inch diagonal. Yes, Windows made my 1440p monitor feel like a 1080p monitor! Worse, with this scaling, toolbar icons, images displayed in the browser, and even some text looks blurry. This happens because the scaling performed is not integer scaling.

Why Not Just Return to 100% Scaling?

Based on my observations, the ideal pixels per inch (PPI) for Windows is 96 for low-DPI and 192 for high-DPI, while for macOS it is 110 for non-retina and 220 for retina. It doesn’t have to be exactly those numbers; monitors with PPI close to those values are still comfortable to use. But remember, the smaller the PPI value, the larger the display appears on the screen, and conversely, the larger the PPI, the smaller the display appears on the screen.

Back to my 1440p 24-inch monitor which has a PPI of 122.4, Windows feels that the display on the screen would be very small if using 100% scaling, so it automatically changes it to 125% or a PPI of 91.8 which is more comfortable for the eyes. Here is a comparison of 100% vs. 125% scaling.

Perbandingan display scaling di Windows 11.

As seen in the comparison above, I am presented with choices that are equally unpleasant. Thus, I concluded that the combination of 1440p resolution and a 24-inch diagonal is to be avoided.

Then How to Choose a Monitor Based on Its Resolution and Diagonal?

Besides the case of the monitor I bought, there are several other combinations of resolution and diagonal size that should be avoided. I will summarize the combination between the two in PPI, which I have explained the calculation for above. In the chart below, there are three categories of monitors according to their PPI, including:

Low DPI Good Zone: Good display at 100% scale on Windows or non-retina mode on macOS. 1 pixel on the OS is represented by 1 pixel on the monitor panel (1:1)
Bad Zone: Combinations that MUST be avoided!
High DPI Good Zone: Good display at 200% scale on Windows (integer scaling), or retina mode on macOS. 1 pixel on the OS is represented by 4 pixels on the monitor panel (2:1)

Kombinasi ukuran monitor yang harus diperhatikan.

Conclusion

Choosing the combination of monitor resolution & diagonal size actually comes back to the purpose of use. If used for gaming, then any combination is suitable because most games can handle many resolution options. Scaling and PPI become less relevant because games are built from shaders and textures rendered in real-time.

It’s a different story if used for content creation or office work, which I do more often. Inappropriate PPI can cause text to be very small and hard to see, but it can also make the display large, take up space, and make some elements blurry and uncomfortable to look at.

Looks like I have to hunt for a monitor again 😖

CPU Undervolt: Cooler and More Power Efficient

Sat, 24 Jun 2023 00:00:00 +0000

Apple recently released the ARM-based M1 and M2 series CPUs which are said to be more power efficient than CPUs from Intel and AMD. x86 CPUs made by Intel and AMD were indeed designed more towards performance than efficiency from the start. But there is one way to make x86 CPUs cooler and more power efficient, one of which is by undervolting. This time I will try to find the right undervolt recipe for the Intel Core i5-12500 CPU that I use as my daily driver.

Things to Note

x86 CPUs from Intel and AMD are known to be overvolted out of the factory. This is done to make their CPUs stable. Since more power is supplied to the CPU, its performance is more stable. In addition, the undervolt potential of one chip vs another is different, even from the same manufacturer and series. Use my experiment as a reference, not a fixed rule.

Please note that software or hardware damage that appears might void the warranty. I am not responsible for any damage that may occur. So be careful.

Methodology

Before undervolting the CPU, it’s a good idea to first record the metrics to be measured before the procedure is performed. Metrics measured include maximum Vcore voltage, maximum temperature, maximum CPU package power consumption, and initial performance score.

Maximum Vcore voltage can be taken using the CPU-Z tool when the CPU is at 100% usage. Maximum CPU package temperature and power consumption can be seen using Libre Hardware Monitor. Meanwhile, to find out the initial performance score, I use Cinebench R23.

Change Vcore mode to adaptive and set offset voltage

Undervolting can be done on non-K series Intel CPUs by changing the Vcore mode to adaptive with offset and setting the offset voltage to a negative value little by little until finding the sweet spot. Ryzen CPUs from AMD should be easier, because all AMD Ryzen CPUs are unlocked from the factory. For which settings to change, please check your motherboard’s manual, as each manufacturer has different names.

Every motherboard and CPU basically applies adaptive voltage, where the voltage given to the CPU will be high when the CPU frequency is high, and vice versa when the CPU frequency is low. By enabling adaptive with offset, we can lower the maximum and minimum Vcore values according to the negative offset we set.

I lowered the Vcore offset voltage by -0.010 V (10 mV) then rebooted.

Check measurement metrics, then lower offset voltage as needed

After lowering the offset voltage by -10 mV, I let the PC sit for 10 minutes so the CPU temperature stabilizes. After that, I ran the Cinebench R23 benchmark for 10 minutes, while measuring the maximum CPU package temperature and power consumption. After the benchmark finished, I recorded the metrics I wanted to measure.

Repeat lowering the offset voltage by -10 mV until the PC is felt to be unstable during the benchmark, or when the benchmark score results plunge downwards.

Measurement Results

Max Vcore	Offset Voltage	Cinebench R23 Score	Max Temp	Max Power
1.116	0.000	11,925	83.00	89.00
1.106	-0.010	11,803	83.00	87.80
1.096	-0.020	11,736	82.00	86.80
1.086	-0.030	11,719	80.00	86.00
1.076	-0.040	11,486	79.00	83.60
1.066	-0.050	10,026	75.00	79.10
1.056	-0.060	8,374	71.00	73.80

The interesting thing I found is that even though the offset voltage I lowered was very small, only -10 mV at a time, it had a direct impact on lowering the maximum temperature and power consumption of my CPU package. Even so, CPU performance also decreased. This was confirmed by the Cinebench R23 test results.

The sweet spot I got was when the offset voltage was at -0.40 V (40 mV). At that value, my CPU temperature dropped by 4.82% (from 83 °C to 79 °C), CPU power consumption dropped by 6.07% (from 89 W to 83.6 W), but the decrease in CPU performance was only 3.68%!

Setting the offset voltage lower than -0.40 V caused a very drastic drop in CPU performance, the effect of which was felt when the PC was being used. Therefore, for the CPU I use, a lower offset is not recommended.

Conclusion

Although it cannot match the power consumption efficiency of CPUs with ARM architecture, there is still room for x86 CPUs from Intel and AMD to be even cooler and more efficient.

Assuming the PC is used for work and is on for 9 hours a day, the potential savings in electricity costs in a month is approximately IDR 2,675, while in a year it can save IDR 32,100. It’s not a large amount, but savings that can be obtained for free should still be appreciated, right?

Used Mining GPU Without Display Output, Can It Be Used for Gaming?

Fri, 09 Jun 2023 00:00:00 +0000

GPU EVGA NVIDIA P106-100, GPU specialized for mining without output display ports.

When mining was trending, NVIDIA along with several hardware partners released GPUs specifically for mining. Unlike general GPUs, these GPUs do not have connectors for display output. The mining hype has passed, now leaving piles of used mining-specific GPUs with unclear functions. I got an NVIDIA P106-100 from a marketplace priced at only around IDR 400,000, even though in terms of specifications it is similar to the Geforce GTX 1060 6GB which clearly cannot be obtained at such a low price even in used condition.

Fortunately, there is a trick that can be used to restore the functionality of a mining GPU into a gaming GPU.

Update 2023/06: Some people failed to install the patched driver. Alternatively, you can directly download the following driver and install according to the accompanying instructions. Driver can be downloaded here.

Update 2023/07: Driver patching steps updated according to instructions from the patch made by dartraiden.

Update 2024/01: Driver patching and installation steps are now even easier. Please see the procedure below.

Product Appearance

The P106-100 unit I got was made by the manufacturer EVGA. It’s quite a relief because in my opinion EVGA is one of the manufacturers that produces NVIDIA GPUs with good quality. I got this product for only IDR 480,000. Actually, the seller also sells similar GPUs but with the P104-100 series which is equivalent to the GTX 1070, only a few hundred thousand difference. Tempted at first, but fortunately I got info that the P104-100 series only has a PCI-E Gen 1.1 x4 connection!

Real condition of unit that I bought.

The P106-100 unit I got was quite pitiful: all the plastic clips holding the fan cover were broken, some screws were rusty, and of course the PCB was coated in black dust typical of used mining GPUs. It was a bit hard to clean, especially the heatsink part. But with water and soap, all the dust disappeared instantly. In addition, the remaining thermal paste was dry and had to be replaced immediately.

All these downsides were actually mentioned in the reviews from people on this item’s page, but with such a low price, it feels like money well spent.

Specifications

The P106-100 specifications are very similar to the EVGA GeForce GTX 1060 SC GAMING, with exceptions:

No display output, requires a CPU + Mobo with iGPU or APU
Uses PCI-E bus version 1.1 x16 (PCI-E 3.0 x16 on the gaming version)
Video acceleration not detected with regular drivers

Prerequisites

CPU and motherboard with Intel iGPU or AMD APU support. Recommended Intel 4th Gen or newer or AMD AM4 G-series platform.
Motherboard must have a PCI-E x16 slot. PCI-E version is not important, because the GPU only supports version 1.1, and PCI-E 1.1 x16 has bandwidth equivalent to PCI-E 3.0 x4. If the slot used is only x8 or x4, performance will decrease drastically.
iGPU / APU feature must be activated through BIOS / UEFI
Operating system uses Windows 10 or newer.

Installation

Windows 10 since version 1803 implements functionality similar to NVIDIA Optimus usually used in laptops. The rendering process will be taken over by the discrete GPU while the integrated GPU is responsible for 2D rendering and displaying output to the screen. The dGPU and iGPU communicate with each other through the PCI-E bus. For that, make sure the iGPU driver is installed correctly, according to the CPU used.

Make sure to turn off the internet connection before starting the steps below

Disable driver installation through Windows Update

This step is important because if it’s not turned off, the patched NVIDIA driver we use could one day be overwritten by Windows Update. Save the following script as a .reg file, then run it.

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows\WindowsUpdate]
"ExcludeWUDriversInQualityUpdate"=dword:00000001

Download and patch NVIDIA Driver

Download and extract NVIDIA driver version 551.23 (must be this version) (using 7-Zip or WinRAR). Then download the patched file and extract all files and folders into the same folder where the setup.exe file is located.

Before installing the driver, make sure there has never been another NVIDIA driver installed on the PC. If the PC is clean, then you can proceed to the next step. If not, you need to completely remove the installed NVIDIA driver using the DDU (Display Driver Uninstaller) tool.

Driver installation can be done by right-clicking the install.bat file, then selecting Run as administrator. The installation window will appear, then follow the process until finished.

Edit registry to activate hybrid mode

Open regedit.exe then go to the address HKEY_LOCAL_MACHINE \ SYSTEM \ CurrentControlSet \ Control \ Class \ {4d36e968-e325-11ce-bfc1-08002be10318}. There will be subkeys named 0000, 0001, etc.. Open them one by one and look for the one related to the newly installed NVIDIA GPU. In that subkey, set the key EnableMsHybrid to the value 1 (if the key does not exist, it can be created as a 32-bit DWORD type).

Restart the PC then reconnect to the internet.

NVIDIA P106-100 installed with GeForce GTX 1060 6GB driver.

Performance

The GPU unit I got runs cool and quiet. Power consumption at idle is only around 5 watts. Temperature at idle averages at 33 °C, while under load the temperature is stable at 70 °C and never higher. When used for gaming, the fan sound is almost inaudible.

Unfortunately, I don’t have a Geforce GTX 1060 6GB GPU to be able to compare directly with this P106-100 GPU. But the good news is, Linus Tech Tips has done a performance comparison using the following devices:

OS: Windows 10
Game: Shadow of the Tomb Raider
CPU: Intel Core i9-9900K
RAM: 16GB
GPU 1: NVIDIA Geforce GTX 1060 6GB
GPU 2: NVIDIA P106-100

The result is very promising, the P106-100 gets an average frame rate of 64 FPS, while the GTX 1060 is at 68 FPS. With a gap of only 4 FPS, of course this result is very good, considering the price of the P106-100 which is very affordable in the market.

However, keep in mind that the results obtained will vary according to the game being run, as well as the CPU & motherboard combination used. The figures above are just a benchmark.

Conclusion

With a cheap price, is this NVIDIA P106-100 worth buying? The answer depends.

If you already have a CPU and motherboard combo that supports iGPU / APU and want to upgrade your GPU but with a tight budget, then this GPU is worth trying. With a GPU priced at around 400 thousand, you can already experience the performance of a GPU priced at 1 to 2 million.

Fixing Strange Windows Keyboard Behavior in VMware Fusion on macOS

Sat, 13 May 2023 00:00:00 +0000

Recently, I installed macOS on my daily driver PC, aka Hackintosh. However, some work-related applications only run on Microsoft Windows. One example is Microsoft Power BI. So I inevitably installed a virtual machine using VMware Fusion to run those applications. But I experienced some strangeness related to the keyboard.

Ctrl + Arrow Causes macOS to Switch Desktops

Change Ctrl + arrow key combination settings in macOS.

The combination of the Ctrl key plus an arrow, Ctrl + → for example, is very often used for navigating between cells in Microsoft Excel. Unfortunately, when I use this key combination, instead of my cursor moving, my host operating system responds by switching from one desktop to another.

This is because the Ctrl + ← or Ctrl + → key combinations are used by macOS for the function mentioned above. The solution is, I changed the key combination to Control + Command + → so it doesn’t collide with Windows. This setting can be found in Settings → Keyboard → Shortcuts → Mission Control → Move left a space / Move right a space.

Ctrl + Left Click Becomes Right Click

Change secondary click settings in VMware on macOS.

One of the key combinations in macOS that I find ‘strange’ is Ctrl + Left Click being translated as Right Click. This behavior is also carried over to Windows via VMware. But the good news is this function can be disabled as follows.

Through the VMware menu bar, go to Virtual Machine → Settings → Keyboard & Mouse, select the active profile, then click the Edit profile… button. Then go to Keyboard & Mouse → Mouse Shortcuts. There, turn off the Secondary Button option. That way the Ctrl + Left Click combination will work as it should in Windows.

Hope this helps! ✨

Intel Core i5-12500 Hackintosh with Gigabyte B660M DS3H DDR4

Mon, 08 May 2023 00:00:00 +0000

Hackintosh with Intel Core i5-12500.

Bored with the same routine, and since I rarely play games anymore, I finally turned my daily driver PC into a Hackintosh PC. The processor I use is an Intel Core i5-12500, paired with a Gigabyte B660M DS3H DDR4 motherboard.

Since Intel 10th Gen is the last generation where the iGPU can work in macOS, I had no choice but to install a discrete GPU. For now, I’m using an AMD Radeon HD 7750, and planning to upgrade to an RX Vega 56 or RX 6600. The GPU I initially used was too ‘old’ to be paired with my motherboard. But with this tool, that’s no longer an issue.

Hardware Used

Category	Item
Processor	Intel 12th Gen Core i5-12500
Motherboard	Gigabyte B660M DS3H DDR4
GPU	AMD HD 7750 2GB
RAM 1	2 X 8GB Kingston Fury Beast 3200MHz DDR4
RAM 2	2 X 8GB Kllisre 2400MHz DDR4
Boot Drive	Kingston NV2 500GB NVME Gen4
Audio	Realtek ALC897 (ALC Layout ID 12)
Ethernet	Realtek RTL 8125 2.5GbE Controller
Bios Version	F22

What Works

Feature	Status
CPU Power Management	🟢 Working
Sleep / Wake	🟡 Partial
GPU Acceleration (OpenCL & Metal)	🟢 Working
Ethernet	🟢 Working
Audio	🟢 Working
Headset and Mic	🟢 Working
Ethernet	🟢 Working
iMessage/Facetime and App Store	🟢 Working
AirDrop/Handoff	🔴 Not Working

BIOS Settings

Make sure to reset BIOS settings to default first.

Tweaker

Advanced CPU Settings
- Hyper-Threading Technology → Enabled

Settings

Platform Power
- Platform Power Management → Disabled
- Power Loading → Disabled
IO Ports
- Initial Display Output → PCIe 1 Slot
- Internal Graphic → Disabled
- Above 4G Decoding → Enabled
- Above 4GB MMIO BIOS assigment → Enabled
- Re-Size BAR Support → Disabled
- IOAPIC 24-119 Entries → Enabled
- Super IO Configuration
  - Serial Port → Disabled
  - Parallel Port → Disabled
- USB Configuration
  - Port 60/64 Emulation → Disabled

Boot

Fast Boot → Disabled
Windows 10 Features → Other OS
CSM Support → Disabled
Secure Boot → Disabled

USB Mapping

Since OS X 10.11 El Capitan, Apple has imposed a limit that the operating system can only use 15 USB ports, and this applies up to the latest version of macOS. Previously this could be overcome with the USBInjectAll kext, but unfortunately it can no longer be used on the last few versions of macOS.

The solution is to map which USB ports will be used and save them as a kext. Using USBToolBox, this mapping process can be done through Windows or Linux.

Test Results Using Geekbench 6

Geekbench 6 Score Comparison.

Using Geekbench 6 testing, it can be seen that the performance of the Hackintosh CPU I use is roughly equivalent to Apple devices using the M1 chip. Single core performance on x86 processors will look low, the explanation is here.

For GPU compute performance, my Hackintosh is far lower than devices with the M1 chip. But remember that I’m using a GPU released in 2012! If we include an AMD RX 6600 XT GPU in the comparison, then the M1 chip will easily lag behind.

However, keep in mind that although Hackintosh has the potential for greater raw performance than devices from Apple, real Apple devices will be more efficient and more power-saving, thanks to specialized accelerators like the T2 chip, neural engine, media engine, etc.

Example EFI Folder

OpenCore EFI File List.

Using someone else’s EFI folder is usually not recommended even if both have identical hardware configurations. However, taking an EFI folder from a similar source as an example will simplify the Hackintosh installation process. Some files can be copied directly (e.g. drivers, SSDT, kext), but settings like config.plist must be adjusted according to each person’s hardware configuration and needs.

Here are some useful links:

Update 2023/05/13:

I found the OpenCore Auxiliary Tool which has the same function as ProperTree (editing the config.plist file) but with a very easy-to-use GUI. In addition, this tool can also be used on Windows, Linux, and macOS.

Good luck! ✨

Tool: GOPupd

Sat, 29 Apr 2023 00:00:00 +0000

GPU-Z view before and after tweak using this tool.

A while ago, I bought an AMD Radeon HD 7750 GPU from Facebook for 350,000 rupiah. I deliberately bought this old GPU for Hackintosh purposes. AMD GPUs are known for having excellent compatibility with macOS compared to Nvidia. When I paired this GPU with an Intel 12th Gen platform, the display only appeared on the monitor once the booting process reached the Windows login screen. Initially, I thought the GPU I bought had a defect, but I was wrong.

After I researched various internet forums, the reason why a black screen appeared during the booting process was that the GPU I installed was too old to support booting in UEFI-only mode (non-CSM). Older GPUs usually do not have UEFI firmware in their Video BIOS. Whether a GPU has UEFI firmware or not can be seen with the GPU-Z tool.

GPU without UEFI firmware, as seen on GPU-Z the UEFI checkbox isn’t checked.

The GOPupd tool functions to insert GOP information into the Video BIOS, making older GPUs UEFI-compatible. Not just AMD GPUs, this tool can also be used for Nvidia GPUs. However, not all older GPUs can be ’tricked’. Here is a list of compatible GPUs:

AMD: HD 6xxx, 7xxx series, or newer.
Nvidia: Series with GT21x, GF10x, GF119, GK1xx/GK2xx, GM1xx, GM2xx chips, or newer.

The steps to be taken are:

Extract the original Video BIOS (VBIOS) from the GPU you’re using (easiest way is using GPU-Z, then click the share button under the manufacturer logo).
Run the GOPupd tool then drop the original VBIOS file we obtained from step 1.
Re-flash the VBIOS to the GPU (For AMD use ATIFlash, for Nvidia use NVFlash).

Keep in mind that any risk of damage is your own responsibility ✌️

Here are some useful links (including download links):

Get Windows 11 Fluent Emoji for Windows 10 and Windows 11

Mon, 25 Oct 2021 00:00:00 +0000

Windows 11 fluent emoji compared to Windows 10 counterpart.

There are many new design aspects in Windows 11 compared to previous Windows versions. One of them is the emojis. Unfortunately, these new emojis were not included in the retail version of Windows 11 released on October 10, 2021. Fortunately, the new emojis from the Insider Preview version can be installed on the retail version of Windows 11, and even on Windows 10.

In Windows, emojis are stored in the Segoe UI Emoji font file with a .ttf extension. As such, font files from one version of Windows can be installed on another version of Windows. Below, I’ve provided the emoji font from Windows 11 Insider Preview 22478.1000.

Download the emoji font in a .zip file through the following link.
Extract the downloaded .zip file.
Select all .ttf files, right-click, and select Install for all users.
If a confirmation dialog box appears to replace already installed fonts, click Yes.
Restart your computer for the changes to take effect.

Install Windows 11 on PC and Laptop Without TPM and Secure Boot

Sun, 24 Oct 2021 00:00:00 +0000

Ilustrasi Windows 11.

Many people want to try Windows 11, which was just released in October this year. Unfortunately, many are disappointed because Microsoft requires Windows 11 to be installed on devices that have TPM and Secure Boot. I will share a way to fresh install Windows 11 without TPM and Secure Boot.

Here I assume the reader already has experience installing Windows on a laptop or PC.

For those who frequently install Windows or other operating systems on PC devices, you’re likely familiar with this tool. Shortly after Windows 11 was released, Rufus added support for creating Windows 11 USB installers that can be used on devices without TPM or Secure Boot.

Here is what you’ll need:

Windows 11 ISO file which can be downloaded from the official website
Latest Rufus which can be downloaded here
USB flash disk with at least 8 GB capacity

Rufus Windows 11.

Here are the steps:

Make sure the USB flash disk is plugged into the PC.
Open Rufus, select the flash disk you want to use, then select the Windows 11 ISO file.
Then in the Image Option section, select Extended Windows 11 Installation (no TPM/no Secure Boot/8GB- RAM).
Click Start and wait until the process is complete.
Install Windows 11 on your PC or laptop as usual.

How to Enable Encrypted DNS on iOS and macOS Devices

Tue, 31 Aug 2021 00:00:00 +0000

Encrypted DNS on iOS iPhone.

Actually, since iOS 14 and macOS Big Sur, both operating systems have supported encrypted DNS or DNS-over-HTTPS / TLS. Unlike Android, encrypted DNS settings are not directly accessible to users. Let’s find out how to activate this feature.

DNS and Its Function

Every device connected to the internet has a unique identity. One of these identities is an IP address. For one device to communicate with another, the device’s IP address must be known first.

An IP address is a combination of numbers that is not easily remembered by humans. For example, 103.102.166.224 or 142.251.10.139. To enable humans to access the internet using a specific address, such as google.com or wikipedia.org, there must be a machine that translates the web address into an IP address. This is why DNS (Domain Name System) servers were created.

Encrypted DNS illustrated.

Unfortunately, the web address we aim for can be easily read by other parties, such as internet service providers, Wi-Fi administrators, or even malicious actors. This is because the web addresses we query from a DNS server are unencrypted. To avoid this, it is highly recommended to use encrypted DNS on various devices connected to the internet.

Solution 1: Using Third-Party Applications

I mentioned earlier that the settings for encrypted DNS on iOS and macOS are not easily accessible to users. Therefore, many applications and service providers offer their services. One of them is 1.1.1.1 from Cloudflare. VPN apps like ProtonVPN or NordVPN also enable encrypted DNS when using their services.

While easy to use, I don’t particularly like this solution. Having additional apps on a phone can reduce storage space, make the device heavier, etc. Furthermore, some services, especially VPNs, also ask consumers to pay for more complete features.

Solution 2: Using Native iOS and macOS Features

To use encrypted DNS on iOS / iPhone and macOS / MacBook, users must download a configuration profile that tells the operating system which DNS to use.

Step One

Click the following links to download the configuration profile of the DNS you want to use (if confused, you can use 1.1.1.1 Cloudflare or Google DNS):

Step Two

After downloading one of the profiles above, a notification will appear stating that this website is requesting permission to install a profile on the device. Click Allow. Then go to Settings -> Profile Downloaded. Finally, click install and enter your device passcode.

Congratulations, your device is now using encrypted DNS, making the websites you visit harder to monitor or block by internet service providers! Now your device should be able to open sites blocked by ISPs, such as reddit.com.

However, Keep in Mind…

Although using encrypted DNS can prevent users from being tracked by third parties regarding the sites they visit, it is not impossible for the DNS server we use to store our internet history. Therefore, please choose a DNS server provider you trust most.

Sources

A Deeper Look at Double Clicking Cells in Microsoft Excel and Google Sheets

Mon, 28 Jun 2021 00:00:00 +0000

Double click cell on Microsoft Excel illustrated.

Every spreadsheet document consists of thousands of boxes, which we usually call cells. A cell is the intersection of a row and a column. Columns are denoted by letters (A, B, C), while rows use numbers (1, 2, 3). Therefore, cells have addresses with a combination of letters and numbers, e.g., C30, B15, etc.

Users can interact with cells. From writing values inside, editing, deleting, copying, to pasting. Additionally, interactions with cells can be in the form of dragging or double-clicking. Interestingly, every double-click applied to a cell with different values will yield different results.

Creating Sequences

Double click cell to create sequence.

Double-clicking an active cell can generate a sequence in the empty cells below if the double-clicked cell meets the following criteria:

Two or more cells containing consecutive numbers (e.g., 1, 2, etc.)
One cell containing text with a number at the end
One cell containing a date

Filling Cells with the Same Content

Double click cell to fill the same content.

Double-clicking one or more active cells can fill the empty cells below with the same content as the active cell, provided the double-clicked cell meets these criteria:

One cell containing text
Two or more cells containing the same number or date (not sequential)

However, Double-Clicking a Cell Does Not Fill Already Populated Cells

Double click cell does nothing when the cell below it isn’t empty.

Double-clicking one or two active cells will NOT populate other cells if they already contain content. To replace the content of other cells, you must drag the active cell downwards or sideways.

So, those are some of the things that can be achieved through double-clicking cells in Microsoft Excel or Google Sheets. Hopefully, this trick can increase your productivity. Stay curious! ✨

Fixing Clock Issues in Windows and Linux Dual Boot

Sat, 05 Jun 2021 00:00:00 +0000

Ubuntu Linux and Windows dual boot illustration.

Playing around with Linux is indeed fun. Trying a system that is completely new to your daily routine is a thrill in itself. However, problems often arise, especially when Linux is paired with Windows on a single computer device via dual boot. One of them is the issue regarding the clock.

It’s a trivial matter, but very annoying if ignored. I first observed this problem when using Ubuntu Linux as a second operating system on my laptop, after Windows, of course. Every time I switched from Ubuntu to Windows, my system clock was always messed up. For example, if it was 4:00 PM, the laptop clock showed 9:00 AM, and so on.

At first, I didn’t understand this anomaly. But after reading here and there, it turns out there is one fundamental difference that causes the mess-up in time.

Fundamental Difference between Windows and Linux

Every computer device we use, whether it’s a PC or Laptop, whether made by Lenovo or Apple, definitely has a device called the Real-time clock (RTC). The function of this device is to store the time so that the computer’s clock is always accurate. This device is usually equipped with a battery, so if no power is supplied to the computer, the time will still be stored accurately.

Any operating system used on a computer device will read the time stored in the RTC and display it in the system. If a user changes the clock in Windows, for example, Windows will save that time value back to the RTC. However, each operating system has its own way of reading and saving time to the RTC.

Linux, macOS, BSD, and other operating systems rooted in Unix will save the time to the RTC in the Coordinated Universal Time or UTC time zone. This means if a user sets the time in Ubuntu Linux to 7:00 PM WIB (UTC+7), Ubuntu will save a time value of 1:00 PM UTC to the RTC device. When Ubuntu is turned on, it reads the time from the RTC in UTC and converts it to local time, in this case, WIB / UTC+7. Additionally, most Linux distributions will fetch the time from the Internet whenever the computer is turned on.

This is where the problem starts. Windows—which has its roots in MS-DOS—by default saves time both in the system and to the RTC in the local time zone. This means if you set the clock in Windows to 7:00 PM, what’s stored in the RTC is also 7:00 PM. Thus, when a user moves from Linux to Windows, Windows reads the RTC written by Linux in UTC but treats it as local time. This is what causes the Windows clock to always be messed up when used for dual boot with Linux.

Although Windows has a time synchronization feature via the internet, unfortunately, it’s not done immediately every time Windows starts up. It takes a few moments for this synchronization to occur, as scheduled by Windows.

Solution: Make Windows Read and Write Time in the UTC Time Zone

The available solutions can actually be applied to either Linux or Windows.

Linux can be modified to save and read time from the RTC in the local time zone. Conversely, Windows can also be set to save and read time from the RTC in the UTC time zone. However, the second solution is the one that should be taken. RTC in the UTC time zone is, in my opinion, more stable and future-proof.

It’s easy. Open regedit, then add a new DWORD and give it a hexadecimal value of 1 to the following registry:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\TimeZoneInformation\RealTimeIsUniversal

You can also copy the following command and paste it into the Command Prompt with Administrator access:

reg add "HKEY_LOCAL_MACHINE\System\CurrentControlSet\Control\TimeZoneInformation" /v RealTimeIsUniversal /d 1 /t REG_DWORD /f

Or even easier, you can download the following .reg file and run it by double clicking. Administrator access may be required. Download the file here.

Notes:

The method I mentioned above might not work on older versions of Windows due to existing bugs in those versions. These bugs include:

For 64-bit Windows 7 and older versions of Windows 10, there is a bug that requires the use of QWORD with a hexadecimal value of 1 instead of DWORD. This bug has been fixed in newer versions of Windows 10.
For Windows Vista before SP2, there is a bug that reverts the RTC time to the local time zone shortly after the computer wakes up from sleep or hibernate mode.
For Windows XP, there is a bug that makes daylight saving time (DST) inaccurate. However, this does not affect Indonesians since they don’t use DST.
For even older versions of Windows, the following article is worth reading: https://www.cl.cam.ac.uk/~mgk25/mswish/ut-rtc.html.

Stay curious! ✨

FreeBSD as a Daily Driver Operating System in 2021

Mon, 24 May 2021 00:00:00 +0000

FreeBSD as daily driver desktop OS.

When talking about operating systems (OS) commonly used in daily life, I’m sure many people are used to and familiar with Windows. Some also use macOS, but unfortunately, macOS is essentially limited to Apple devices. For tech enthusiasts, GNU/Linux is certainly familiar. But what about FreeBSD?

Ancestrally, macOS, Linux, and BSD come from the same family tree. The difference is that macOS succeeded in the desktop realm, while Linux and BSD are mostly used on servers. Even so, many people use Linux and BSD as their daily desktop operating systems.

In this article, we will discuss the possibility of using FreeBSD as a daily desktop operating system. Note that this article is not a step-by-step tutorial, but rather a proof of concept. If you want to try it yourself, Google and YouTube provide plenty of tutorials you can follow.

Let’s get straight to the main point. We’ll use FreeBSD 13.0. An ideal desktop operating system in 2021 should be capable of:

GUI and built-in system accessories
Internet connectivity and social media access
Multimedia consumption (video playback, YouTube, Netflix, Spotify, etc.)
Document and spreadsheet manipulation
Simple photo and/or video editing
Online meetings

Let’s break them down one by one!

GUI and Built-in Accessories

By default, FreeBSD does not offer a GUI out of the box. Users must download and install it themselves. Options include GNOME, KDE, Xfce, or MATE, similar to most Linux distributions. Since I prefer a minimalist and lightweight look, I chose MATE.

To make the process easier and simpler, I used an installation script by Felix Caffier, which you can access on this GitHub link.

Accessories on MATE Desktop.

The MATE desktop offers fairly complete features for a daily driver, such as a file manager, calendar, text editor, image viewer, calculator, etc. For me, these are more than enough. Missing features like a video player can be easily installed via the pkg command in the terminal, which I’ll discuss in the next section.

Internet Connectivity

FreeBSD has good support for internet connectivity. Popular web browsers like Firefox and Chromium are always updated to the latest versions, just like on Windows, macOS, and Linux. Thus, it’s almost certain that all websites will open correctly.

Internet browsing on FreeBSD. There are various web browser available to use.

I tried Firefox, Chromium, and Epiphany — the default GNOME browser — to browse Facebook, Twitter, and Instagram. I also used them to post on Facebook and create new tweets. Everything ran smoothly, just like using any other operating system.

Interestingly, when I tried logging into Instagram using the Epiphany browser, Instagram automatically sent a notification to my email about the login attempt. The operating system I was using was correctly detected as FreeBSD. However, the notification stated I was using Safari!

Suprisingly, Epiphany browser was detected as Safari by Instagram.

This is explained by the fact that Epiphany uses WebKit as its engine, just like Safari and several other browsers.

Multimedia and Entertainment

A daily operating system isn’t complete without entertainment. Music and videos are often used as breaks from work. How does FreeBSD meet these needs?

The FreeBSD repository provides music and video player options comparable to Linux distributions. I easily installed Rhythmbox for music, and VLC and mpv for video, using the pkg command. With these, playing local media files is not an issue.

YouTube consumption is also smooth, thanks to support for up-to-date web browsers. In my tests, YouTube videos could be played up to 1080p without much frame drop using Firefox and Chromium. Of course, this also depends on the computer specifications.

Spotify cannot open due to lack of DRM on FreeBSD.

Issues begin to arise when FreeBSD is used to play DRM-protected media like Netflix and Spotify. First, there are no native applications for either platform on FreeBSD, meaning no official support. On other operating systems, even without a specific app, Netflix and Spotify can still be accessed via the browser. Unfortunately, not on FreeBSD.

This leads to the second problem: the absence of a DRM module in web browsers available on FreeBSD. Thus, sites like Spotify and Netflix refuse to work, displaying error messages like the one above.

So far, there is no solution for Netflix. Spotify can still be used using a command-line application in the terminal. While not an elegant solution, for those who don’t mind being called a “freak,” it’s a fun alternative to try.

Alternatively, Spotify is available as CLI / TUI apps.

Two applications make this possible: spotify-tui and spotifyd. Both are available on various operating systems. Installation instructions for FreeBSD can be found here and here.

Document and Spreadsheet Manipulation

Regarding documents and spreadsheets, the de facto standard is Microsoft Office. Regardless of how one gets it, Microsoft Office is widely used. Like Linux, there is no native Microsoft Office application for FreeBSD, leaving two alternatives.

Native Office Suite on FreeBSD

Several open-source alternatives are available, including LibreOffice and Apache OpenOffice. Both have very similar features since they originate from the same root. However, in terms of UI and compatibility with MS Office files, LibreOffice is better.

LibreOffice on FreeBSD.

Lighter alternatives are available if full features and MS Office compatibility aren’t critical, such as Abiword for word processing and Gnumeric for spreadsheets.

Web-based Office Suite

Working online has become the norm. Collaborating on the same platform to create documents is now easier, using web-based office suites like Google Docs and Microsoft Office Online.

The advantage here is that no installation is required, only a capable web browser. Fortunately, as discussed, web browser options on FreeBSD are diverse and always updated.

Web based Office Suite on FreeBSD.

However, remember that most web-based office platforms require users to store files in their online storage. I don’t think this is an issue, as almost everyone has a Google or Outlook account nowadays.

Simple Photo and Video Editing

FreeBSD has never been widely known as a desktop OS, let alone a photo and video editing workstation. That title usually belongs to macOS and Windows, where hardware acceleration is broadly accessible and commercial editing apps are primarily available.

Darktable as Adobe Lightroom’s alternative is having rather complete features and good performance.

Here are some free or open-source editing apps available on FreeBSD that can be installed via pkg:

GIMP — Adobe Photoshop alternative
Inkscape — Adobe Illustrator and CorelDRAW alternative
Scribus — Adobe InDesign alternative
Darktable and RawTherapee — Adobe Lightroom and Affinity Photo alternative
Kdenlive — Adobe Premiere Pro, Final Cut Pro, and DaVinci Resolve alternative

Online Meetings

Due to the current pandemic, many jobs are handled online, including meetings. As a daily operating system, it wouldn’t be complete if it couldn’t handle online meetings.

In this area, FreeBSD clearly looks like a “stepchild” compared to Linux. Several platforms that have Linux apps — like Zoom, Microsoft Teams, Slack, and Discord — are not available on FreeBSD. Therefore, users must settle for using a web browser.

Recently, a zoom-video-conferencing-client package became available via pkg. This Zoom app uses a Linux compatibility layer, which comes with limitations: audio doesn’t work, and users must log in every time the app opens.

Thus, the online meeting experience on FreeBSD is less than ideal. Read David Schlachter’s article on the same topic here.

Final Words…

It’s been an enjoyable experience using FreeBSD as a desktop for the last few weeks.

FreeBSD is a robust, secure, and stable operating system. Major services like Yahoo!, WhatsApp, and Netflix use or have used FreeBSD as their backend servers. But as a desktop OS? It’s rarely heard of.

From my tests, I conclude that FreeBSD can be used for daily tasks. However, that doesn’t mean I recommend it. Again, this article is a proof of concept, so feel free to use whatever operating system you think is best for you.

Stay curious! ✨

Changing Hyperlink Color in Microsoft Word, Excel, and PowerPoint

Sat, 08 May 2021 00:00:00 +0000

Changing hyperlink color in Microsoft Office.

The link or hyperlink feature is a function that serves to make text, objects, or images into active links. When clicked, the text or object given a hyperlink will take the user to the destination link, whether it’s a website page or an email address. Unfortunately, text turns blue when made into a link and its color cannot be changed manually. Fortunately, there is an effective way to work around this.

Create a Shape Over the Link Text

The first step is to place a shape over the text that will be made into a hyperlink. Any shape will do, as long as it can cover each part of the text. For example, I’m using a rectangle shape.

Create a shape over the link text.

Add the Link Address to the Created Shape

The purpose of the shape we created is to serve as the clickable point for the link. The advantage of using a shape as a link instead of text is that the shape’s appearance doesn’t change when assigned a link. This makes it ideal for disguising links so the original text formatting remains unchanged.

Add link address to the shape.

Disguise the Shape So the Text is Visible Again

As mentioned, the shape is only meant to be the link’s click point. To make the text visible again, the shape on top of the text must be made transparent. To do this, set Shape Fill to No Fill and Shape Outline to No Outline. This way, the text underneath will be visible once more.

Disguise the shape so the text is visible.

Well, it’s quite simple, isn’t it? If you’re still having trouble, check out this video tutorial now! ❤️

Trying Google Camera on Non-Google Pixel Android Phones

Sat, 28 Dec 2019 00:00:00 +0000

Result from Google Camera.

For Android phone enthusiasts, the name Google Pixel is certainly familiar. This Google-made phone is famous for having amazing photo results. Even so, its camera specifications are actually quite ordinary. From Pixel 1 to Pixel 3, there is only one camera, and even that has a resolution of only 12 megapixels. So how can it produce such great photos?

The key is in the software. Google, as a software company, believes that good images can be obtained from processing. This image processing power lies in the Google Camera app. This camera app was developed by Google itself. Unfortunately, this application can officially only be used on Pixel phones.

However, don’t worry—many developers out there have successfully created ports or modifications of this app so it can be used on non-Pixel phones. But before we get there, let’s look at some of the advantages of Google Camera compared to most built-in phone camera apps.

Attention: As this article is being updated (May 2021), Apple starting from iPhone 11 has introduced a similar feature called Night Mode. Additionally, almost all Android phone manufacturers have embedded similar features in their latest phones. Thus, the results obtained are not much different.

HDR+ Feature

This feature delivers super clear images while evening out the brightness levels. This feature has actually been in the Google Camera app since before Pixel phones were made. However, it had to be turned on manually because phone processor power at that time was not as strong as it is now.

Example of photo results using HDR+.

Generally, HDR is obtained by combining several photos at once. But in the Google Camera app, taking photos with HDR+ feels just like taking a regular photo.

This is where the cleverness lies. The app continuously saves photos even before you press the shutter button. That way, when the shutter button is pressed, the app combines several of the latest photos to create a photo with just the right lighting.

Taking many images to be combined.

Because current phone processing power is adequate, HDR+ is automatically applied to every photo you take using the Google Camera app.

Portrait Mode Feature

Smartphones in 2019 wouldn’t be complete without this feature. Portrait mode allows taking pictures with clear separation between the subject and the background with the help of blur or bokeh.

Technically, this feature requires a phone with at least two cameras: one to take the picture and another to map the depth of field. Google Pixel 1 through Pixel 3 only have one camera. How can Portrait Mode be done?

Portrait Mode Feature.

Google equips the Google Camera app with artificial intelligence (AI) that can distinguish the subject from its background. Of course, the results can vary and are not as accurate as using two cameras. But when I tried it, portrait mode worked very well—considering my Xiaomi phone only has one camera!

Example of Portrait Mode result on a Xiaomi phone using Google Camera. Left: a normal photo taken with HDR+ mode. Right: the blur/bokeh effect is obtained through AI processing on a phone with a single camera. The result is very satisfying.

The GCam app will save two photos: one without the blur effect and another with blur or bokeh. If you use Google Photos as your gallery app, you can even readjust the blur intensity you want.

Night Sight Feature

The most relied-upon feature of Google Camera is Night Sight. This mode allows bright photos to be taken in low-light situations. Before the Google Pixel, Huawei was indeed the first to bring this feature to Android phones. However, in my opinion, the results obtained from the Pixel are better.

iPhone photo result compared with Google Pixel 3 using Google Camera.

The way it works is similar to HDR+: combining several photos to produce a clear photo. The difference is that this process takes about 2 to 5 seconds, depending on lighting conditions. During the process, there should be no movement from either the photographer or the subject. Therefore, this mode is more suitable for landscapes or stationary objects.

Here are some examples of Night Sight results taken from the GCam app on my Xiaomi Redmi 5 Plus.

Example of Night Sight result using Google Camera on Xiaomi Redmi 5 Plus. Left: Photo taken using GCam HDR+. Right: Photo taken using Night Sight. The left result is more similar to the actual situation when seen directly by the eye, but the right one is brighter and clearer.

Top: photo taken with HDR+ mode. Similar to what the eye sees. Bottom: Night Sight photo result. Taken handheld / without a tripod. The Night Sight mode result exceeds human vision.

Left: Photo taken using GCam HDR+. Right: Photo taken using Night Sight. The results are not too different, but the Night Sight result is clearer and noise-free.

Automatically Shutdown Your Computer Without Additional Apps

Sun, 15 Dec 2019 00:00:00 +0000

Laptop Illustration.

Sometimes you’re faced with a situation like this: the computer you’re using is processing something. Maybe it’s rendering your ex’s wedding video or downloading a movie. The process takes a while, but you’re in a hurry to meet someone.

On the other hand, you don’t want your laptop or PC to stay on for hours unused until night. What should you do? In this article, we’ll learn how to make your computer shutdown automatically.

Every operating system installed on our computers, whether it’s Windows, macOS, or Linux, has a feature that can automatically turn off the computer after a certain period. Not many people know about this feature because it exists in the form of a command line or instruction line.

Don’t worry, even though it’s in the form of a command line, this feature is very easy to use and doesn’t require any additional applications. Here’s how to do it:

Windows

Windows Run command.

Run the Command Prompt by clicking Start, then type Command Prompt. Even easier, you can access it through the Run menu: press Win+R, type cmd, and press Enter.

Then type the following command line and press Enter:

shutdown -s -t 3600

Here’s an explanation for the options that can be used:

Command	Explanation
shutdown	command to turn off the computer
-s	command to shutdown. Can be replaced with -l for logout, -r for restart, and -h for hibernate
-f	add this option to force close stubborn programs that might prevent shutdown
-a	this option can be used to cancel a previously issued shutdown command
-t	timer option for shutdown. If not filled, the computer will turn off in 20 seconds. If filled with 0, the command will be executed immediately. It can also be filled with a number in seconds (e.g., 60 seconds x 60 minutes = 3600 seconds for 1 hour)

Shutdown command in Windows.

If used frequently, this command can also be made into a shortcut! This shortcut can then be placed on the desktop for easy access.

Right-click on an empty area of the desktop. Select New > Shortcut.
“Type the location of the item:” will appear. Fill it with the same command line as the example above. You can also adjust the time as desired. Then click Next.
After that, enter a name for the newly created shortcut. To make it easy to remember, I’ll name it Turn off 1 hour. Click Finish.

Auto shutdown shortcut for 1 hour.

At this stage, the shortcut has been successfully created. However, it doesn’t have an icon yet. Changing the icon is easy. Right-click the shortcut, then select Properties.

On the Shortcut tab, click the Change Icon… button. Select the desired icon, then click OK. Your newly created shortcut should now have its icon changed.

macOS / Linux

macOS and Linux also have a shutdown command accessible via the terminal. But since both OSes are descendants of UNIX, the use of this function is slightly different than in Windows.

On macOS, to access the Terminal, press Command + Space on the keyboard to bring up Spotlight Search. Type “terminal” and press Enter.

Meanwhile, on Linux, accessing the terminal varies from one distribution to another. But generally, the terminal is in the Utilities section of the application list.

Run Terminal in macOS.

Once the Terminal app is open, enter the following command:

sudo shutdown -h -t 3600

On macOS and Linux, only the root user can call the shutdown command. Therefore, we use sudo. Furthermore, here are the options that can be used:

Command	Explanation
sudo	command to run functions as root user
shutdown	command to turn off the computer
-h	command to shut down (halt) the computer. Can also use -P which will also turn off the computer. To restart use -r
-a	this option can be used to cancel a previously issued shutdown command
-t	timer option for shutdown. If not filled, the computer will turn off in 20 seconds. If filled with 0, the command will be executed immediately. It can also be filled with a number in seconds (e.g., 60 seconds x 60 minutes = 3600 seconds for 1 hour)
time	besides in seconds, absolute time can also be entered here in hh:mm format e.g. 23:30

Shutdown command in macOS.

After the command is entered, press Enter, and you’ll be asked to enter your root account password. Press Enter once more after entering the password, and the computer will turn off according to the specified time.

So, that’s an easy way to automatically shutdown your computer without additional apps. Hope it helps! ✨

City Pop Music: From Mariya Takeuchi to Fariz RM

Sun, 01 Dec 2019 00:00:00 +0000

City Pop Illustration

Mariya Takeuchi certainly didn’t expect that City Pop would be back in demand. The song “Plastic Love,” which she performed 35 years ago, is now considered one of the best city pop songs ever. This Japanese music genre brings a fresh, bold, and optimistic musical vibe. Yet, not many people know that there are also musicians and music with a similar style in Indonesia.

What is City Pop music?

The City Pop genre carries its own philosophy. City Pop emerged as a reflection of optimism towards Japan’s economic progress in the 1980s. At that time, car stereo systems were booming. People enjoyed this music while driving past the shimmering neon lights of Tokyo. That’s why this music is called City Pop.

If you listen closely, you will find a blend of Pop, Jazz, and Funk. Sometimes, synthesizer sounds also grace the melodies. All of this was thanks to the rapid growth of new technological inventions in Japan.

There are many versions of how this genre was born. However, musician and composer Tatsuro Yamashita, with his song “Ride on Time,” is said to have boosted the popularity of City Pop. Mariya Takeuchi, Tatsuro Yamashita’s wife, performed “Plastic Love” in 1984, which for me perfectly illustrates the City Pop genre itself.

City Pop music reached the peak of its popularity in the 80s. After that, it wasn’t talked about much anymore. Then, in 2010, City Pop got a second wind. The online Vaporwave community started talking about it again. This kind of music was listened to again. YouTube, through its algorithm, displayed “Plastic Love” as a recommendation on many users’ homepages. Unfortunately, these songs cannot be found on online music services like Spotify.

How about City Pop in Indonesia?

In Indonesia, City Pop doesn’t have its own specific category. Instead, it is generally considered pop music. It doesn’t have its own dedicated fans like Rock, Jazz, etc. Nevertheless, it cannot be said that City Pop is absent from the Indonesian music scene.

Economic progress in the 80s, along with the glamour of Jakarta, also led to the growth of City Pop-style music. For me, Fariz RM and Chandra Darusman, in some of their songs, bring the same vibe of Pop, Jazz, and Funk: City Pop!

Try listening to a collection of Indonesian City Pop in this Spotify playlist. It’s great to listen to, especially when driving through city streets.

The song “Selangkah ke Seberang” by Fariz RM has a very strong City Pop feel to me. Although this song has a fairly long intro, it’s very pleasant to hear. The melody is filled with 80s-style electronic instrument sounds (synthesizers), bolstered by a prominent bass line.

Candra Darusman has a different approach. His music feels more lively and energetic thanks to the use of brass instruments. Listen to the song titled “Kekagumanku” and feel the energy.

Unfortunately, the revival of Japanese City Pop couldn’t quite bring Indonesian City Pop back to life. Most people would react with surprise at how old-fashioned my music taste is when I listen to those songs. But that’s okay, those songs are clear evidence that Indonesian musicians also love to experiment with many styles and genres 🎵

Blog Archive — Rully Pratama

Going Back to an Old DSLR in the Smartphone Era: When Technology Makes Us Forget How to Shoot

1. Forgetting How to Focus (Focus & Recompose)

2. Guessing the Exposure

3. Honest Old Sensors vs Computational Photography

4. The Hassle of Sharing Moments

5. ISO Limitations and Noise Tolerance

Conclusion

Installing MikroTik RouterOS on a Headless Mini PC via Serial Console

The Headless Installation Challenge

The Solution: Modifying syslinux.cfg

Why Do the Same RAW Photos Look Different?

A bit about how cameras capture light

Different apps, different ways to process RAW data

Fun fact: RAW files also store small JPG images as previews

Is the Lumix GX7 ISO Invariant?

What is the difference between ISO variant and invariant cameras?

Methodology

Results

YouTube Not Saving History? Why?

DNS aka Domain Name Server

Dual Boot Windows with VHDX

Step 1: Prepare the .vhdx File

Step 2: Mount the Desired Windows Installation ISO

Step 3: Installation Process

Step 4: Add to Boot Entry

Step 5: Test Bootloader

Energy-Efficient Banana Pi NAS

Background

Why Banana Pi?

Test Results

iperf3

NASTester

Power Consumption

Conclusion

Replacing Old PC with Energy-Efficient Mini PC: Is It Financially Profitable?

Short Review: TP-LINK TL-AP1750C

Pros and Cons

Turn an Old Router into a Managed Switch with OpenWRT

Fan Mod on HBA Card LSI SAS 9210-8i, Making the Chip Much Cooler

Dark Themes for Frequently Used Applications

Firefox

Windows Terminal

Visual Studio Code / VSCodium

Notepad ++

Spotify

Thunderbird

OBS Studio

Flow Launcher

Besides Resolution, Also Pay Attention to Diagonal Size When Buying a Monitor

Why Not Just Return to 100% Scaling?

Then How to Choose a Monitor Based on Its Resolution and Diagonal?

Conclusion

CPU Undervolt: Cooler and More Power Efficient

Things to Note

Methodology

Change Vcore mode to adaptive and set offset voltage

Check measurement metrics, then lower offset voltage as needed

Measurement Results

Conclusion

Used Mining GPU Without Display Output, Can It Be Used for Gaming?

Product Appearance

Specifications

Prerequisites

Installation

Disable driver installation through Windows Update

Download and patch NVIDIA Driver

Edit registry to activate hybrid mode

Performance

Conclusion

Fixing Strange Windows Keyboard Behavior in VMware Fusion on macOS

Ctrl + Arrow Causes macOS to Switch Desktops

Ctrl + Left Click Becomes Right Click

Intel Core i5-12500 Hackintosh with Gigabyte B660M DS3H DDR4

Hardware Used

What Works

BIOS Settings

Tweaker

Settings

Boot

Step 1: Prepare the `.vhdx` File