A simpler solution was to use my mobile phone as a webcam. A quick search uncovered a handful of candidates so I set about with a quick unscientific test in a normal boring office with unflattering light (definitely not Insta perfect).

There are handful of mobile apps that used WiFi to ‘broadcast’ the video from my mobile device to my laptop. At my laptop, this mobile camera showed up as a normal webcam which I could easily select for input with Zoom or any WebRTC application.

The Software

Epocam is an iPhone only app (with a matching driver/program for Windows or Mac) and works via USB or Wifi. Epocam is only $7.99 and sold by Elgato, who has a wide variety of video/audio components for the serious web broadcaster. The app had a few nits but otherwise worked well. You open your video application SELECT the Epocam as your “camera” and then open the Epocam application on your iPhone. After a few moments, they would magically connect.

Epocam is my recommendation based upon ease of use, however, it doesn’t work on Android devices. Elgato sells other products for your home studio. I thought this was cool “Stream Deck” which is a little macro-control physical keypad that allows you to single press to perform several functions (think switching your camera source back and forth). This YouTube video explains in more detail.

iVCam works on both iPhone and Android however it only seems to only support Windows computers. iVCam is from e2eSoft and appears to be a Chinese company which I found to be the Shanghai Yitu Information Technology Co. The website is in English, basic but there was none of the typical company information which was suspect. Pricing was not on the website and you can purchase from the PC client or via the app using Google Pay. It’s an annual subscription of $7.99/$9.99 for basic or advanced or a single $24.99 one-time fee.

Despite lack of Western marketing capabilities. iVCam worked as promised and had a similar performance as the EpocCam application.

DroidCam was the clunkiest of them and supports both iPhone and Android and but only supports Windows/Linux (no Mac). It appeared to be a techie project of some sort. Whereas the other applications have a ‘self-discovery’ function to match the client IP with the host IP. With DroidCam you had to type in the IP address of the host PC/computer. Despite that, it worked fairly well and the pro version is a miserly $5.99.

Finally, there is from Camo from Reincubate this iphone ONLY product offers a lot of video tuning elements (you can easily switch between various input devices on the fly). Its matching desktop software works on Macs and PCs. The downside, it only supports a wired USB connection. Worth checking out but for $39.99 it’s expensive if you using it only as a virtual webcam.

My discovery – the Apple iPhone 11Pro Max ($1,100) camera was beaten to the ground by a new Samsung Galaxy A32 5G ($250)! Even a cheapie $79 BLU Vivo X5 phone looked better IMHO.

The Testing

I started the test with the bias that my iPhone was likely the best virtual webcam. I quickly found the iPhone delivered the ‘worse’ quality as a virtual webcam.

To create a reference, I took a native pic with the front-facing cameras of the iPhone, Samsung, and Blu. Note: The Samsung was sharp and the iPhone was a bit yellowish to my liking. The Blu while sharp was washed out. I would deem all of them acceptable.

iphone

Samsung

BLU

Below are sample screenshots from live Zoom calls using a variety of phones and virtual webcam software. For reference, I included a Logitech C920 Pro ($79) shot. Note: my office has a full window providing natural lighting and I attempted to take all shots with same lighting..

IVCam Samsung	IVCam Iphone	IVCam Blu
Droid Samsung	Droid iPhone	EpocCam Iphone
	Logitech C920

As an additional comparison, here are two photos (from the far end) using the Zoom mobile app. In this case, the iPhone seemed warmer and not as washed out as the Samsung.

iPhone

Samsung

As the video is going from your mobile to your laptop before being sent out again there was some lag. However, the lag was not super noticeable. In a side-by-side comparison, you could visually denote the lag though. The virtual webcam apps also support using the phone for the microphone input (but I didn’t test that).

Key Takeaway

For the money a virtual camera is a handy tool to consider. The Zoom mobile client might includes as a ‘feature’ to facilitate multiple cameras). Disappointing though was how consistently bad the iPhone looked even against a $79 BLU Android mobile (when operating as a virtual webcam).

Photo by Jeroen den Otter on Unsplash

The post Using your mobile camera as a webcam? appeared first on Chris Kranky.

WebRTC using a wideband codec offered the promise of high fidelity. But if your input source is bad (bad microphone or acoustics), then no amount of fancy technology further downstream will make you sound better. I argued, go out and by yourself a better microphone (article here).

In the pre-TikTok era, I argued you should invest in a decent lighting set-up, it matters and I wrote an article on inexpensive lighting solutions. This was again pre the days of dozens of ring lamps being available on Amazon.

But I want to get back to video quality because it matters too. Years ago I called a major webcam provider to ask what they were doing for higher optical quality & faster frame rates – they thought I was crazy, it was all about styling. I also contacted Taiwanese OEM manufacturers of webcam assemblies (which typically cost about $1.25). Is anyone asking for high quality/faster frame rates? Nope. I suspect in the last year they all may have changed their tune. 3rd party webcams have gotten better and this is a super article on webcam comparisons.

I have become annoyed with the perpetual ‘soft’ focus that ALL online video conferences seem to have (clearly good). Is this the best we can do? For this article, I wanted to do a simple comparison of Zoom vs WebRTC. Is there a clear winner? Yes, perhaps.

Jitis did a nice comparative YouTube video segment in 2018 however, it’s 3 years later and Zoom likely hasn’t been asleep at the switch. Where are we today? I set out to find out in my highly unscientific but typical office worker set-up.

• I downloaded a classic eye chart and printed it on our office laser printer (you can see this below)
• I taped the eye chart to the wall and set up a recent vintage Lenovo T470P running Windows 10 using Chrome Version 90.0.4430.212 (Official Build) (64-bit) and Zoom April 30, 2021 version 5.6.5 (823)
• The laptop was set up ~ 26 inches / 65 cm from the wall (so about normal viewing for a person). Normal office lighting (fluorescent).
• All test calls were made between 2 computers on our internal network.
• For the WebRTC calls, I used talky.io running in P2P mode as well as Meet Jitsi https://meet.jit.si/ (which supports HD).
• I tested using the Lenovo’s internal webcam and just for fun, I did similar testing using Apple iPhone 11Pro Max using the Zoom native client and Safari.
• I took a screenshot of each image after allowing the connection to stabilize.
• I edited each screenshot in Photoshop purely for size and alignment as there were some slight offsets (but photos were unretouched).

Before you complain, I’d love to see someone more technically capable take on a more scientific review (I devoted an entire hour to this in-depth project!). Sample images below.

To my eye, Zoom looks better (if you’re doing remote eye exams, Zoom is definitely the way to go, everyone is guaranteed to need glasses). I noted that the WebRTC connection seemed to hunt for quality. The image would go from good to bad and back to good. Zoom seemed to stay on mark and didn’t fluctuate (to my eye). While Zoom’s quality wasn’t outstanding, it was consistent. Check the comparisons below and let yourself judge my unscientific results.

Perhaps this comparison isn’t fair. I didn’t take into account color rendition or how either handled poor network conditions. We don’t normally ‘video’ still images and zoom in on them. But from my basic layperson’s view, Zoom was a tad sharper. Though my message is there is room for improvement.

We’ve all experienced the new 4K video segments of the “nearly there” experience. I have found that YouTube segments with poor video or audio quality struggle to hold my nanosecond attention span. While network speeds have improved, they may never achieve consistent speed/quality to achieve the silky smooth and sharp 4K imagery we simply will expect to hold our attention.

Below is where we are and indeed, we’ve come a long ways (from when video didn’t work except in rare circumstances with a ton of fiddling). Network speeds and delivery might get marginally better in the future, but I’m thinking we’re going to have to re-imagine the entire notion of how video is transmitted and rendered. Unknowing what we know or believe to apply new techniques to the same old problem is perhaps our future.

Below is comparison WebRTC vs Zoom using my Lenovo Laptop default.

As a follow-on, here’s a comparison between the native Lenovo webcam and and Apple iPhone 11Max pro on Zoom. The iPhone clearly is trying to brighten the scene automatically.

I also looked at Agora.io using their native demo vs a WebRTC P2P session using my Lenovo webcam.

Finally, here is same with a Logitech C920 HD Pro with Zoom feature set to HD mode and HD mode using Jitsi Meet. The HD setting enables 1080p. HD obviously needs more bandwidth and reliably if it’s to work well.

Caveats: While I did numerous test runs this was NOT a 100% controlled testing environment and your own results are likely to vary. Did I see a clear winner? No. Some tests seemed sharper than others, but this sharpness also introduced some level of pixelization which could be perceived as bad. And as I noted, video isn’t about remote eye exams, it’s usually about seeing another person and here perceptions may change.

The post Zoom vs WebRTC Video Quality appeared first on Chris Kranky.

Within the tech security community, Zoom got called out today for a potential security vulnerability for the usage of the Mac webcam. Nothing strikes fear in a user’s heart than the notion their webcam is secretly watching them. This fear is so great that users install plastic little covers to slide over the webcam and newer laptops of often have an embedded physical cover.

The headline “Zoom Communications can access your webcam without your knowledge” is fodder for a USAToday article.

With the posting of Jonathan Leitschuh blog article today the pin was pulled to potentially become a huge PR disaster for Zoom. What did Eric Yuan do as CEO to counter this? He joined a Zoom conference call with 250 security hackers!

Calmly, thoughtfully and with respect, he patiently responded to the group in technical terms. Eric is an engineer after all! Hard to imagine the CEO of a public company with a $25 billion market cap joining without a cadre of lawyers in tow. Solid move and the security group clearly respected the move by Eric.

The clickbait blog article clearly was heart stopping for the CEO of any public company (particularly as it was early in the trading day). This article wasn’t drop bombed on Zoom. Rather the security group had informed Zoom of their findings as far back early March 2019.

Let’s dial this back a notch to try and understand what the real issues are. Zoom has long focused on making the application work fast and flawlessly. To enable this capability, they have to do some under the hood engineering moves.

Zoom allows you to use a standard URL to start a video conference. We’ve all done it. You click a link, a web page momentarily opens and then the Zoom client starts. Works well and frankly, I like how it operates.

However, if you take this same link and embed it into your own website and hide it — any Mac user visiting that page will immediately get a Zoom conference launched. Worse, if I set up the link properly, I can force YOUR video to start immediately showing me your camera. There might be a great reason I would legitimately want to do this, but like all good things, there is a dark side as well.

Similarly, if I were living on the darker side, I could create a Zoom link, embed into my dark website and create a denial of service attack on your computer. Oh, joy! (note this has been patched in the current version).

To make matters worse, the security team discovered the Zoom client was running a small web server on your Mac (note the blog article is Mac focused). This tiny web service was deployed to assist in downloading the Zoom client and start of a Zoom session. Zoom indicates they implemented this server to get around limitations that Apple Safari imposed. Unfortunately, if a handful of things go wrong, this server could be exploited for darker purposes. Having any web server just idling running on your computer is mostly a bad idea.

Are these serious problems?

Yes and no. The blog headlines seem to say blast this is a major problem. In reality, the worse that happens is the Zoom app opens and your camera light comes on. Clearly, this could be surprising and you might well not notice if distracted.

Independent security experts are a mixed blessing for any CEO. They often find things that are a legitimate concern for protecting the user. The Zero Day Initiative Group has a well-prescribed 120-day to cure before going public with their findings. Unfortunately for Zoom, today was day ZERO.

Eric Yuan’s bold and immediate reaction is noteworthy. But the question is where were Zoom’s security team in the last 120 days? Why did it get to day ZERO before the CEO himself had to intervene? These are no doubt questions Zoom internally is trying to answer today.

Zoom remains a valuable communications application that works well so that even a novice can utilize it. With it’s expanding popularity though, it becomes a playground for hackers to exploit. Novice users are easily exploited so it’s in Zoom’s best interest to take seriously these security concerns.

Note: Following broad news today, true to form, Zoom quickly released a patch to address bulk of issues with Mac. Current release Version: 4.4.4 (53932.0709)

The post Zoom Security appeared first on Chris Kranky.

The Krisp SDK promise is to “on the fly” improve audio quality by removing background noises and upscaling the existing audio to HD. Extra benefits are the ability to handle packet loss and dynamically adjust the audio levels. This article focuses on background noise elimination.

2Hz using AI has built their own DNN, they’re calling krispnet, built on a training of 10k speakers, 20k different background noises and 2.5k hours of audio. Is that enough? In the world of AI more training data is better, but a good start.

The Krisp.ai application (available now for Mac and Windows) uses this krispnet SDK/. Krisp creates a new sound input/output device on your computer. You manually configure each comms application to send your audio first to Krisp. Setup is straightforward.

Krisp offers a free full trial of the application. After 14 days though the application will only work for the ‘speaker’ (inbound audio) and NOT the microphone element.

Let’s start with a baseline. Here is a brief unmodified recording I did (using Audacity) on a MacBook Pro using the internal microphone at 44100Hz (high quality).

From a previous posting on microphones, the Mac has a great internal mic. In the baseline sample, my voice is clear and high quality. However, the background noise is similarly HD and competes with my voice.

Below is a 44100 Hz sample with Krisp ENABLED.

The background noise is mostly gone in the Krisp enabled example. Negatively, Krisp has slightly altered my voice and introduced some new artifacts. The artifacts give my voice a more mechanical sound and somewhat less natural. Nonetheless, it’s an improvement over the standard (opinion). The Krisp app did not add any noticeable uptick in CPU utilization.

Let’s step the quality down a bit now. Below are two additional samples at only 8kz sampling rate, a more likely an Internet communications app. Krisp did an even better job here.

At the lower sampling rates AND with background noise, Krisp shines here as you can hear, there is nearly no background noise. While audio quality in both sample isn’t HD, with the Krisp enabled example you don’t strain to pick out my voice (as all the background has been eliminated).

Note: Krisp purports to work on the inbound audio, however, I was unable to notice any remarkable difference when enabled. This raises the question of whether Krisp is of any value past the trial period unless you upgrade to a paid version. Krisp worked well with Zoom in my testing.

The paid version of Krisp is $20 per month which seems high for a ‘feature’ type product ($120 a year if you pay at once). Krisp is keen to get trial use and offers a referral link which extends your free usage period for 2 months. You can use mine https://ref.krisp.ai/u/u456100422

AI for in real time media process is still new. The Krisp app is promising with the microphone element working well for those of us in noisy environments. Krisp is usable today and with additional model enhancements likely to improve in the future.

The post Krisp – AI for voice quality improvements appeared first on Chris Kranky.

Is there something easier for us to take on? As the plane pulled up to the gate I thought I would have to think about this later. Unfortunately, ‘later’ came around immediately as we waited 10 minutes for the plane doors to finally open whilst the hapless ground crew tried to wrangle the Jet Bridge into place. I realized that in my years of travel this ‘waiting for the Jet Bridge’ was quite a common occurrence.

Thus it dawned on me that shortest distance, and possibly least complex, self-driving vehicle opportunity was mere feet from where I was sitting. The airport Jet Bridge.

Connecting a Jet Bridge to a plane is a basic exercise in geometry.  You have a few pivot points and a single extension. “Driving” a Jet Bridge also involves a joystick. Since 99.9% of us don’t use a joystick that often nor understand geometry  – it’s no wonder that properly connecting a Jet Bridge to match the side of an aircraft door is hard. Harder still if you’re sitting inside the plane waiting.

Since airports are busy and expensive, even learning to operate a JetBridge is costly. So costly in fact that one software company has a ‘game style’ JetBridge simulator to help train operators!

So it would seem a self-driving Jet Bridge would be easy. After all, it only has to drive 10 feet (3 meters) on flat ground and mostly only needs to watch out for something big, metal and shiny. But Jet Bridge tires often get ‘flat’ spots which affect their precise handling and a gate often supports different aircraft types during the day affecting the configuration.

The economic benefits of a self-driving Jet Bridge would be enormous. For a commercial airline, the cost of a plane sitting idle at a gate is ~ $65 per minute! If you could reduce the gate connection time by a mere 3 minutes this works out to a saving of $195 per gate movement (I didn’t factor in the undocking process). Considering that in the United States there are approximately 87,000 commercial aircraft movements per day, in theory you could save $17 million in a single day or over $6 billion a year.AI

Below is a list of US airports with > 100 gates and 3-minute reduction in gate time would generate nearly $2 billion in annual savings just for these 18 airports!

A new Jet Bridge typically has a cost of $500-800k and given the current inputs (an on/off switch and joystick) is not super complex.  What would you need for a self-driving Jet Bridge?

• A handful of infrared cameras
• Some infrared lighting
• A Raspberry pi with some cables to the existing circuitry
• Very basic LIDAR to map the plane.
• Perhaps a 2nd camera into ensure the Bridge didn’t run over anyone below. The $99 Xiamoi LIDAR module (shown above) from a robotic vacuum cleaner would easily work.

Even at wild sum $500k per gate for the upgrade, the payback period for largest airports in the US is less than 6 months.

Southwest Airlines, a master of efficiency, is trying to launch a better jet bridge and Shinmaywa Industries in Japan has been working on a self-driving Jet Bridge  They estimated the added cost for a self-driving newJetBridge to be less than $50,000.

The predominate supplier of Jet Bridges in the US is the $1 billion revenue JBT AeroTech Jetway Systems (whose current website doesn’t appear to have any job openings for those with machine learning or artificial intelligence skills).

Innovation on reducing boarding times included the ill-fated Double Docker system which connected two Jet Bridges to the front and rear exits of the plane. The rear Jet Bridge unfortunately spanned over the wing of the aircraft. This experiment came to an abrupt end when on March 31, 2007 the rear bridge fell on the wing of a parked United 757 at Denver’s airport. The message: experiment but don’t destroy. Most of the efforts to automate Jet Bridge operations have been limited to pre-set maneuvers and a large array of sensors.

Today there is much discussion about deep technology in artificial intelligence which is well beyond the knowledge of a normal reader. As well, AI applications remain futuristic and out of reach to the average business. The opportunity remains to couple business application with the technology available to us now.

The post Driving a JetBridge appeared first on Chris Kranky.