Posts from the ‘Uncategorized’ Category

Handbrake is a powerful video transcoding program than is greatly misunderstood. A lot of discussion goes into which preset to use for the best quality and fastest encode times. However, Handbrake’s presets are designed for best compatibility of devices regardless of quality. If you are planning to playback on VLC or Roku Media Player not to much to worry about in terms of comparability. However, if you have an Apple TV or a Bluray Player look at the support guide. The presets are basically useless and will lead to worse encode times when stringent format support is not required.

When encoding for home consumption your looking for

• Smallest File
• Quickest Encode Time
• Highest Quality

A lot of people after watching youtube videos think GPU encoding is perfect for this job, you would be wrong. Here are the issues with GPU encoding is that GPUs are geared towards live encoding for game streaming. The reason they can get these incredible encode time is they are doing a lot less optimized encoding. How video encoding gets great compression is by finding similarity between frames or reducing quality. Finding similarity between frames means you have to look at more frames before outputting a frame. Looking at more frames increases latency thus why it is generally avoided in GPU encoding as it has worse compression. Below I have video by YouTube Maraksot78 about compare file with different encoders. Intel quick sync encoding H.265 provided the smallest file of any of the hardware accelerated systems it still was 40% larger than the CPU encoded file and 1.6 times faster encode time.

Also, an annoying thing about handbrake is that custom presets are compatible across operating system or version. Which means you’re going to have remake them every time which is very annoying. 

Here is what you need to compress video

• HandBrake
• MediaInfo

First quick way to reduce encode time

• Copy content to internal drive if you have enough space. This way reading the file does not slow down encode time
• Make sure your using fast external storage (use a usb 3.0 drive or better yet an external drive enclosure)
• Use MediaInfo to learn more about the data you want to encode 

Step 1 Learn more about your content

Is the video you want to encoded a rip from a Video Disc (DVD or Bluray or Maybe an HD DVD?)

Is the video progressive or Interlaced?

What format is it encoded in? as you may not get additional compression 

Here are some examples from MediaInfo using the tree view 

Example 1: A video from my digital camera

Here are the fields to pay attention to Complete name this is the name and location of the file. The format field tells you the container that is used to storage the video file. A container is basically like when you store a bunch different documents in a zip file so they are linked together. A container is always used because inside a video file is literally a separate audio, video and subtitle files. A container also limits what type of files maybe stored inside of it. The field format profile tells you what video profile your movie in this affect things like maximum quality that can be stored. 

Now we will look at the video section. The fields you care are format, format info and Scan type. The format field tells you what codec this video is encoded in it currently encoded in AVC which stand for Advanced Video Codec as stated by the format info. AVC is also known as H.264 which is the name of the ITU standard for the video format. Then finally the Scan type field tell you if this progressively scan like modern display or interlaced. Knowing the scan type can help save encode time and do necessary processing.

Now finally onto the audio field. The fields of interest are Format, channel(s) and Compression mode. The format tells you what audio format is stored in. When encoding you never want to modify this format because increase computation and can hurt audio quality or even convert 5.1 to surround sound. Many will say you should convert it to save space but in the case of this file the audio only accounts for 2% of the total file size so your no saving anything.

Example 2 A rip of DVD movie

Notice that we have additional field in this such as text and menu along with additional audio fields. This has multiple audio tracks in this case English and Chinese. The text fields contain subtitles and the menu contains the DVD chapters.

So big difference is that the format field is Matroska. This is a MKV file and my preferred containers it allows a lot more file types to be stored and is an open format.

The fields are format, format version, frame rate and original frame rate. The format and format version tell you that this is encoded in MPEG-2 this is a very common format on DVD. The very important field is frame rate and original frame rate. DVDs can only store video at frame rates of 25 or 29.97 (this is known as 30 fps broadcast but it is not really 30fps because of reason that are beyond the scope of this article look up NTSC or watch the playlist by Technology Connections. But in short summary when your encoding video you need to set the frame rate to the frame rate field ,but handbrake will show the frame rate as Original Frame rate field.

The big difference is that this has surround sound with its six channels. The format is is AC-3 which has the commercial name of Dolby Digital. Look the field language this is the english track ,but Audio #2 is the Chinese track.

I am showing it for information consistency but you will not have to do much with it in handbrake.

This allow you to use scene selection.

Example 3 interlaced DVD

Here is the important fields to look out for is scan type, scan order and Standard. The scan type field shows that this video file is interlaced and that the first field is at the top of the screen. Film Maker IQ has a great video (The Joys and Sorrows of Interlacing) explaining what interlaced vs progressive video. However because this interlaced it will need to be de-interlaced to properly be displayed on modern display. The Standard will tell you what standard you need to set it and guess at the true framerate.

Here this is just to point out that DTS surround audio exist.

---

Now onto handbrake Settings. The goal is to show you how to encode a HEVC/H265 file passing thru all audio and subtitles into an MKV container.

Picture Top menu

Here is how we break Down the top menu

Open Source – This is what you click to open a video or folder of videos you want to encode

Add to Queue – If you imported a folder of videos this allows you to setup the same transcoding for all video in the imported folder. This allows you to set a bunch of settings on one video and apply it to all videos.

Start – will start the encoding of videos

Pause – pause allows you to pause encoding when you need more CPU powerful

Presets – Shows you list of available presets. If you are trying to encode for specific device this can be useful ,but they will not be used in this tutorial. 

Preview – this allows you to preview your video encoding setting this very useful and can save you a lot of time compared to just rendering out a whole a video. This will also allow you to estimate how long an encode will take.

Queue – Allows you to see video that are in the encode queue and this opens up a new window

Activity – handbrake is basically a wrapper for a command line program and this just a cleaned up version of the command line output. Can be helpful for figure additional stats about encoding speed but generally you will will not need to interact with it.

Source – is the current video file that you are working on.

Angle – would allow you to selection another camera view if you have a file with multiple video streams

Range – allows you to select the region of the title you would like to encode. The selection range can be anything from chapters, seconds or frames

Preset – is the currently selected preset

Summary pane

This pane summarizes the encode settings

save as – controls the output file name

encode time – would tell you the progress of an ongoing encode

Dimensions Pane

Basically set for the bare minimum any cropping on downscaling will cause additional computation.

Orientation & Cropping

Unless the video has clear visible black bar around the edges you want everything in Orientation & Cropping off. Flipping should be unselected, Rotation should be at zero and cropping should be set to none. If your platform does not have a none option then should set it to custom and put zero in all the boxes.

Resolution & Scaling

Resolution Limit – None. This will allow you to reuse this as a template. 

Anamorphic – Automatic

Optimal Size – Leave Checked

Allow Upscaling – leave unchecked. If checked this will drastically increase computation time depending on what your resolution limit is set to.

Borders

fill -none. You do not want to add a border to your video

Final dimensions

can be ignored let handbrake do it for you.

Filters Pane

Anything in this pane will be done with CPU and increase encode time

I mention the setting Detelecine (link wikipedia article). What happen is this processing is about undoing the process to turn convert film to television frame rate called telecine. Converting Cinema (shot on film) frame rate (24fps) to television frame rates 25 fps and 29.97 fps involves some interesting process but when converting to a progressive frame rate we want it to be at its original frame rate. Animated content is almost always rendered at 24fps because it is cheaper to do it this way. for reference for you save 60 frames every minutes when you compared to 25fps. You save 358 frames per minute compared to 29.97 fps. If you assume a film is approximately 90 minutes long that mean at 24fps saves you 32238 frames compared to rendering at 29.97fps and 5400 frames compared to 25 fps. In other terms renders in 29.97 compared to 24fps is like making your movie 22 minutes longer and almost 4 minutes longer at 25 fps. So assume it is at 24fps if it is animated or your regions equivalent.

If your content is progressive or the scan type is not provided then you want to set every filter to off. Playing with these filters will not really improve your encoding with out massive amounts of tweaking and then only marginally. If your frame rate differs original frame rate and is 24 fps or 23.97 or 23.976 enable the Detelecine this will help improve the quality of motion. 

If your content is interlaced only turn on Interlace detection to default then set Decomb and choice the preset EEDI2 this will give you the best Deinterlace quality. If you know the content original came from film or is animated turn Detelecine to default.

Video Tab

Video Encoder – H.265 (x265) x265 that will encode your video

Frame Rate – the simplest and the least computational expensive option is same as source. However, if you video was originally from film set the frame rate to the appropriate frame rate for your region.

Make sure your have handbrake set to variable frame rate this means that only when a change on the screen this can drastically decrease your file size if you have a lot of static content. Make sure your quality is set to Constant Quality this will give you the best looking output by varying how large each frame needs to be to have maximum quality. HandBrake provide suggested values in the document “Adjusting Quality”. The smaller number the value the better the quality but do not go 2 less than suggested as this will just increase file size without really improving quality. Remember zero just render a whole saves each frame as an uncompressed image and reduce file size is about only storing the difference between frame. 

Preset – set this to faster. A lot of people believe that this a quality setting including the handbrake team. It is not a quality setting but a compression vs quality. Basically increasing from faster to fast might reduce your file size, but it will definitely increase your encode time and remember since you are using constant quality encoding it will just increase the file size in specific regions to maintain quality so I would really only go from faster, medium & slow. I have not noticed visual quality drops or drastically bigger files using faster but have noticed the speed up in encode time. I have noticed quality drops using anything below faster.

Tune – If it was originally shot on film set to grain and if it is animated content set to animated. Otherwise set to none. This will also help reduce file size.

Profile – auto. Only the main and main still pictures exist so no reason to set it to anything other than auto.

Level – set to auto. (I am not even going to try to explain this. If your player says it only supports a specific level in h.265 set it to that level otherwise just leave it at auto)

Here is a link to the documentation for x265 the encoder used by handbrake to encode the files this better explains tunes and profile. You can learn a lot by reading this document it is very well written actually. A wikipedia page about the different types of frames used in encoding will help explain things like b-frames.

Audio

So here I am gonna to tell you to basically throw away everything that is already enabled away.

Click the button “Selection Behavior”

in the new window

in the new window

Track Selection Behavior – All Matching Selected Languages

Languages – Any. Or your preferred language. Since audio takes so little space compared to video no reason to really not keep all the tracks. Then the person who prefers subs is always comfortable.

Auto Passthru – Select all of the check boxes you do not want to waste cycles re-encoding anything. Then set Passthru fallback as Flac 16-bit so you preserve the maximum quality audio in a lossless format. In most cases it will not do anything but pass the audio. 

Then click the minus button at the bottom and make sure the pane is completely empty. 

Then add the following

Codec – Auto Passthru

Mixdown – 7.1

Samplerate – Auto

Bitrate – cannot be set as this is passed from the file itself

gain – 0

DRC – 0

NOW click ok.

In the Audio Pane click reload

(if you do not reload you will not pass thru all the changes.)

This will load every audio track including stereo files of the selected language because by selecting mix down 7.1 it will never actually mix down audio.

Subtitles Pane

then click selection behavior

Track Selection Behavior – All Matching Selected Languages

Languages – Any 

Options – Add Closed Caption when available select this option. Add foreign audio search.

Burn-in Behavior – None and deselect DVD & Blu-ray subtitles.

Click OK 

Then hit reload in the subtitle pane.

Now all the subtitles are added. Once again these are text files do not add much to the final output.

Chapter Pane

If you know or want to name the chapters you can in this pane.

---

Create New Preset

Click New preset

Resolution Limit – None

Audio – Repeat the selection from the audio pane it does not pull that from your current settings which is annoying.

Subtitles – Repeat the selection from the subtitle pane.

Now that you are done you will want to save a preset for different qualities based on resolution and maybe even different tunes.

Why I do not like handbrake’s presets 

In general the presets basically never use H265 which will give you the maximum compression possible without loss in quality. Only streaming services and website do not use H265 because they do not want to have to pay a licensing fee, but as an end user who has devices that support the format it saves you a ton of space. Yes, the encode time is longer, but I have found with the faster preset my H265 files are shrunk by a 1/3 on average (big variance can occur depending on the source material) without visual drop in quality just by going to H265. You can fiddle with the knobs of h264 all you want your not gonna get that drop in file size. Many presets forces a frame rate which could cause you to have to waste time generate additional frames or lose frames.

Should you even do this?

A lot of YouTuber are pushing services like Plex and Jellyfin and show these encoding tutorials. However, the issue with home media servers is that ripping DVD and Blurays then transcoding video takes longer than watching the films themselves. A NAS and Libredrive compatible Bluray drive is going to cost more than just buying a nice Bluray player. Also, in getting an optimal video deinterlacing and upscaling in most cases you just need to pop in a disc into a Bluray/DVD Player. You can buy a 4k capable Bluray player for between $150-1000 which will get you some of the best Bluray player in the world. The cheapest two drive NAS that I could find was around $160 with no storage drives which means for a lower end Bluray player (which the Sony ones are actually quite good) you could be watching a bluray instead of ripping them. So, unless you have videos that raw video files or video downloads it is more worthwhile for you time and money to just play your disc on a player. You can read my article about optical storage media ,but it has a much longer life than common wisdom states. Optical media has already out lasted the other storage media at this time so backing up to SSD or hard drives is a waste of time.

An important question we need to ask is what is the purpose of the reviewer? Most people watch or read reviews to find out if something is worthwhile to purchase and to see how it compares to other in that price range. An important aspect of reviewers is to help their audience gain context about the current products they are reviewing, the available competition and that category in general. It is also, important to try to understand your audience that your targeting. 

A review that upon first reading irked me upon further reading is actually a good review not flawless, but good. In the BuzzFeed’s review of the M1 iMac John Paczkowski talked about how the original iMac was so much better than his 1995 Compaq Presario. This seem like a strange reference to mention a nearly 3 decade only computer, but it gives the reader the context of why people cared about the original iMacs. He then goes on to mention his current computer a 2015 iMac and how it is much slower than the M1 and cannot have as many applications open. He then proceeds to say wirecutters top pick HP All-in-One 24-df1036xt is worse because it has worse performance, a worse screen and is not worthwhile even with its lower price. My complaint is that the reasoning behinds these opinions is not well supported. This review does not provide benchmarks or a quantitative Data besides he could have more tabs open. However, the reader is given actionable information that will help them make a purchase decision. I am now going to show you how reviewers can go wrong with quantitative numbers and be lead astray by numbers and specifications. I will be comparing the M1 Ultra Mac Studio with the Alienware R13 the media coverage.

(Many) Reviewers are Basically Useless

Both Youtube channels Linus Tech Tips and Gamer’s Nexus have done reviews of prebuilt computers. I am going to focus on Gamer’s Nexus Coverage of the Alienware R13 and compare it to other outlets. I am picking this computer because it is the flagship device from a major PC manufacturer. It also cost around the same as Apple’s Mac Studio another flagship device. We can see how these computers compare in benchmarks as well. 

Gamer’s Nexus released 4 Videos about the computer and its components

1. Crazy Bad $5000 Alienware Gaming PC: R13 Aurora Tear-Down (April 24 ,2022)
2. The Worst Pre-Built We’ve Ever Reviewed: Alienware R13 $5000 gaming PC Benchmarks (May 1, 2022)
3. “Fixing” the Alienware R13 Dumpster Fire (May 22, 2022)
4. Sometimes Dell Actually Tries: Dell RTX 3090 Review, Tear-Down & Benchmarks (July 2021)

A quick summary of the problem that Gamer’s Nexus found was that the Alienware ran so hot and was so power limited that its CPU the i9-12900KF performed like an i7-12700K. A power limit means that the system did not provide enough power to the components it was due to software in Alienware’s case, but can also be a power supply issues as well.

Looking at Amazon on July4 , 2022 the i9 cost $549.97 and the i7 cost $377.98 (USD). That means that in the a flagship computer your CPU cannot perform to its full potential and performs like a CPU that cost more than $150 less than the one inside the computer. Also, the computer uses a R13 specific case, motherboard & power supply so you cannot reuse any of the components for a new build. Removing the front panel caused the internal temperature to drop by 10 degrees and during load the CPU was nearing the dangerous maximum temperature. The fans are loud. All the ram is green on an all black Motherboard & Graphics card PCB it does not match the aesthetic. The RAM that Alienware put inside the computer is Slow DDR5. The computer does not have a thunderbolt ports. The R13 is not meaningfully upgradeable as you are thermally and form factor limited. The front inputs are on the motherboard instead of a separate board. If your front inputs get damaged it requires you to replace your entire Motherboard not just the input board.

This is not the case with custom built computers and this is not the case with M1 Ultra Mac Studio as in seen in the Max Tech Teardown . Apple has more replaceable components so that if any port gets damaged you replace the breakout board not the whole motherboard. However, Apple locks it components to CPUs so enjoy trying to fix it but it can be physically be fixed. It also does not have 10gigabit which is used for High end connected storage and is standard on the Mac Studio. My custom built computer which is half the price of this computer has 10gigabit ethernet and thunderbolt so it is frankly unacceptable. When we visit the website OpenBenchmarking.org the i9-12900K (A variant that has same boost and clock but more enterprise features) it is in the 75th percentile of a site that allows dual core servers workstations. The i9-12900K in many benchmarks in in the 99%+ percentile so saying it is faster than any other system is a given. The real question is what value does Alienware bring to this components! 

Other Media Coverage

Many of Many (An Australian Men’s LifeStyle site)

They mention the loud fans but are not as annoyed as Gamer Nexus. They never really compared it to other systems on the market or talked about thermal throttling. They never compared it to an external test bench. They did mention that boutique builds would use nicer material than was used on the Alienware. They also did not compare it to other offering.

Tech Targeted Media

PC Mag

When scrolling across the page they show similarly priced system but the prices are not as configured for the review but as the base system price which is misleading. They also mention that it does not feature glass side panel on the case that high end competitors have, but plastic. This computer cost $4.5k+ so maybe Alienware could put some nice glass on it I mean it is only a couple of months of rent in a two bedroom apartment in NYC (https://www.apartmentlist.com/renter-life/cost-of-living-in-new-york). They do not mention that it does not have comparable connectivity to other products in this price range and that the device thermal throttles. The Alienware system also performed below the test bench they used in the review for the i9-12900K they reviewed standalone which was not mentioned. System to system variance is real, but for $4.5k you can tune a system. Compared it to other offering and it was worse with a more powerful GPU. Also, the benchmarks never ran for a longtime so the thermal system was never really taxed. So, more benchmark and no real additional useful information.

Hot Hardware

Mentions fan noise of previous generations. They mention the wifi chip could be better I mean it is only $4.5K+ who needs better wifi. They did not experience any thermal throttling in testing and device even got the Editor’s Choice award even when it did not perform as well as on some of the benchmarks for the standalone i9-12900K that Hot Hardware reviewed. They also did not use almost any of the same benchmarks for the standalone beside Speedometer so it is very hard to show a drop in performance. 

If you look at these website you would think that I am making up this thermal throttling issue, but I picked this device because I know it thermal throttles. So, the YouTuber JustIn Tech Tips made a video called “How to Mine Crypto on the Alienware Aurora R13 – Without Crashing!” he had to dissemble the GPU and apply new thermal paste on a New $4.5k+ computer. The GPU temperature dropped by almost 20 degrees Fahrenheit and allowed him to Crypto mine for 24 hours a day instead of instantly crashing! I am not a fan of crypto and recommend Nicholas Weaver lecture talking about its issues. JustIN Tech Tips also replaced the stock fans with a pair of after market fans (Noctua NF-A12x25) he found that it was quieter even at maximum speed. I am avoiding talking about his testing methodology because that is a whole another article and most channels do it wrong. He did see that when the where fans not at full blast they had worse temperatures. The difference in fan performance could be do to a poorly optimized fan curve or how quick it makes the fan spin for a given load. Remember Alienware only contribution to this system was the power supply, cooling, circuitboard and chassis design. The fact that both JustIn Tech Tips and Gamers Nexus could improve the system means that Alienware’s engineer where not trying very hard. 

Now let’s see how the Alienware compares to the Mac Studio with M1 Ultra. First UFD Tech found that the Mac studio was very efficient and could mine and does not mention the device crashing when trying to mine on the device or the standalone PC he built. In the PC Mag review they mention even under load the Mac Studio fans do not get loud unlike the Alienware. They mention that the Mac studio is not user upgradeable ,but do not talk about how the Dell used not standard parts. In Linus Tech Tips review of the Mac Studio even with using unrealistic workloads they could not get the mac studio to thermal throttle and the fans never spun up past 50% of fan speed.

 The Tech Journalist Pipeline

I could only find a couple, but Glassdoor says it all a Journalist Total pay is $85,580 ($52,551 Base Pay) and for an Electrical Engineer $102,303($81,649 Base Pay) so becoming a journalist will cause you to loses $15k+ in salary on Average. Making it hard to recruit any one from the industry with knowledge. PC Mag’s Glassdoor shows that a staff writer makes somewhere between $76K – $81K Base Pay. This is actually industry competitive base salary, but it does not show the difference in working environment. 

A big incident of the poor culture in tech journalism is the infamous Verge PC Build which showed the incompetence of need for quick video for clicks by releasing sloppy videos. Check out Timmy Joe PC Tech video on the subject for a full context. Eric Jackson a former Contributor for Forbes wrote that “Why is there so little criticism of tech companies by mainstream business journalists today?”. Where he lamented that few journalist add their own opinion because they fear being wrong or losing access to companies. He points out the mainstream Business Press is not writing Long Form stories covering tech companies. Jackson says “Sure, you’ll get the occasionally bright blogger shining a spotlight on truth but the mainstream business press really needs to start pulling its weight.” YouTuber are the bright bloggers of the modern era.

Tech journalism has become tedious product journalism where printing the spec sheets for mass produced consumer products is celebrated as a great story and where there appears to be little understanding of bigger picture stories about how our digital technologies are transforming our industries, cities, and our societies, at a pace and scale that’s never been seen in our history.

Tom Foremski article in “The role of tech journalism in a post-technology world”

His words are still applicable today as very little context is given even in bench mark heavy reviews. For instance the technology outlets who benchmarked the Alienware R13 got burned because Alienware made sure that short benchmarks would run fine on the machine. As someone who works in a companies lab we do know what test your most likely going to run and we will see how we compare to thecompetition before hand.

Closing: Why does this matter?

The average consumer does not have the ability to do high level testing or try a large set of products that are available so they need journalist to provide them with context. The technology companies have basically spent over a decade with minimal journalistic oversight and it shows. Companies are savvy and know what is being tested and reviewers must update methodology to make sure the readers understand the benefits and downside of products.