[0:00] I make a lot of videos related to
[0:02] virtual desktop, like a lot. But one
[0:04] question keeps coming up repeatedly.
[0:07] Which is the best codec to use? So today
[0:10] I'm going to address this question with
[0:12] a comparison under the same strict
[0:15] testing regime where all settings are
[0:17] the same and the only variation is the
[0:20] actual codec itself. Let's get straight
[0:23] into it then. And remember, we're born
[0:24] to respawn.
[0:27] the test. My PC is running a 12th gen i9
[0:31] CPU and RTX 4080 GPU with 64 gig of DDR5
[0:36] RAM. I run a TPLink Archer AXC75 Wi-Fi
[0:40] 6E router in access mode exclusively for
[0:43] virtual desktop to maximize wireless
[0:45] performance which sits at a steady 2.4
[0:48] Gbits pers on the 6 GHz channel. Virtual
[0:51] desktop's PC streamer app settings are
[0:53] as follows. On the options tab, open XR
[0:57] runtime. I am using virtual desktop's
[1:00] own VDXR runtime. Adaptive quantization
[1:03] and two pass encoding are off as they
[1:06] only support the 10- bit codeex.
[1:08] Automatically adjust bit rate is off. I
[1:11] have manually set each bit rate to its
[1:13] maximum allowed so that there are no
[1:16] anomalies. On the advanced tab, I have
[1:18] the horizontal and vertical tangent
[1:20] sliders at 100% with the FOV stencil
[1:24] ticked just to ek out a bit more
[1:27] performance. In headset, VR graphics
[1:29] quality is at ultra. VR frame rate is 90
[1:32] fps. Bit rate is set to the maximum
[1:34] allowed by the codec. For example, H.264
[1:38] allows up to 500 megabits per second.
[1:41] Synchronous spacew walk is off.
[1:43] Snapdragon game super resolution is off.
[1:45] and video buffering is also off. These
[1:49] settings are purely to eliminate any
[1:51] interference and keep the test as pure
[1:54] as possible. I made sure that Virtual
[1:57] Desktop was running the latest build and
[1:59] that there were no interfering apps
[2:01] either. I am using my MetaQuest 3
[2:04] headset and all footage you will see was
[2:06] recorded internally so that you could
[2:08] see the virtual desktop performance
[2:10] overlay in real time. All in-game
[2:13] graphics were on the ultra preset except
[2:16] for trackside TV displays which are
[2:18] turned off and the grid was restricted
[2:20] to 20 cars. That's it for the settings.
[2:22] As I stated, all these will remain the
[2:25] same throughout all five tests and the
[2:28] codec will be the only variable. I will
[2:31] do one lap of Monza in my favorite
[2:34] McLaren GT3 car at midday with the same
[2:37] weather for each test. I will show the
[2:39] whole lap with the virtual desktop
[2:41] performance overlay showing in real time
[2:43] so that you can see how each codec
[2:46] performed. Then I will run the same laps
[2:48] side by side for a direct comparison.
[2:51] The final part of the video will be my
[2:53] conclusion which if you haven't fallen
[2:56] asleep by then will be less
[2:58] science-based and more on the feeling of
[3:00] which codec delivered the best balance
[3:03] of latency and visual performance for
[3:04] me. So, if you're a proper geek like me,
[3:09] make sure to stay till the end of the
[3:10] video for that. Ready? Let's go.
[3:15] H264,
[3:17] the oldest codec here, running at 200
[3:20] megabits pers. It should deliver the
[3:23] lowest latency, but with video quality
[3:25] compromised. Let's see how it does.
[3:31] Yeah.
[3:36] [Music]
[3:44] [Music]
[3:54] Heat.
[3:58] Heat.
[4:00] [Music]
[4:12] [Music]
[4:17] Heat.
[4:21] [Music]
[4:34] Heat. Heat.
[4:46] [Music]
[5:02] [Music]
[5:07] [Music]
[5:13] H264
[5:15] Plus. This is a modification of the
[5:18] H.264 codec, allowing for better
[5:20] compression efficiency and can run up to
[5:24] 500 megabits per second.
[5:27] [Music]
[5:40] Hey,
[5:42] hey,
[5:45] [Music]
[5:54] hey.
[6:02] Heat. Heat.
[6:03] [Music]
[6:13] [Music]
[6:22] Heat. Heat.
[6:24] [Music]
[6:40] [Music]
[6:42] Heat. Heat.
[6:46] [Music]
[7:01] [Music]
[7:12] HVC,
[7:14] sometimes referred to as H.265.
[7:17] This is the successor to H.264 and
[7:20] should allow better compression and
[7:22] quality over its predecessor.
[7:25] [Music]
[7:28] Hey,
[7:34] [Music]
[7:40] hey,
[7:44] hey.
[7:47] [Music]
[7:55] Heat.
[7:58] [Music]
[8:07] [Music]
[8:14] Heat.
[8:16] [Music]
[8:19] Help.
[8:21] [Music]
[8:37] [Music]
[8:39] Heat.
[8:43] [Music]
[8:57] [Music]
[9:09] HEVC 10bit. An evolution of HEVC where
[9:13] the 10 bit refers to each color red,
[9:16] green, and blue being represented by 10
[9:18] bits and evolution required for higher
[9:21] resolutions like 4K and 8K.
[9:32] [Music]
[9:38] Heat.
[9:40] [Music]
[9:47] [Music]
[9:49] Heat.
[9:59] Hey,
[10:09] heat.
[10:16] [Music]
[10:28] Hallelujah.
[10:35] [Music]
[10:38] Heat.
[10:41] [Music]
[10:58] [Music]
[11:07] AV1 10 bit offering better compression
[11:10] and color than older codecs. This is
[11:13] currently the most modern codec used by
[11:15] virtual desktop.
[11:18] Heat.
[11:20] [Music]
[11:30] [Music]
[11:40] Heat.
[11:43] [Music]
[11:54] Heat.
[11:57] [Music]
[12:07] Heat.
[12:15] Heat.
[12:17] [Music]
[12:26] Heat.
[12:33] [Music]
[12:54] [Music]
[13:02] Right, let's sit back and run them all
[13:05] side by side for a direct comparison.
[13:11] [Music]
[13:19] [Music]
[13:27] Heat. Heat.
[13:29] [Music]
[13:44] [Music]
[13:54] [Music]
[13:56] Conclusion. Unsurprisingly, all the
[13:59] codecs performed well with consistent
[14:01] frame rates and manageable latency with
[14:03] my PC and wireless setup. As expected,
[14:06] H.264 264 at 200 megabits per second
[14:09] consistently gave the lowest latency
[14:11] overall between 31 to 48 milliseconds
[14:15] with good frame rate between 87 and 90
[14:17] and an average across most of the lab of
[14:20] 90 fps with 37 milliseconds latency. But
[14:23] to me the visual fidelity was clearly
[14:26] lower. H264 plus at 500 megabits per
[14:29] second had higher latency between 34 to
[14:32] 55 milliseconds. probably due to the
[14:34] higher bit rate. A good frame rate
[14:36] between 86 to 90 with an average 90 fps
[14:40] across most of the lap with 40
[14:42] milliseconds latency. And like H.264,
[14:45] my opinion is that the visual clarity
[14:47] wasn't as good. HEVC at 200 megabits per
[14:51] second had slightly higher latency at 37
[14:54] to 52 milliseconds. Frame rate between
[14:56] 84 to 90 with an average across the lap
[14:59] at 90 FPS with 48 milliseconds latency.
[15:02] However, the visuals were a definite
[15:05] step up from H.264 and 264 Plus. HEVC
[15:09] 10bit and AV110 bit had an almost
[15:12] identical performance with both
[15:14] latencies between 32 to 56 milliseconds.
[15:17] Frame rates between 86 to 90 with an
[15:19] average across the lap at 90 FPS with 37
[15:24] milliseconds latency. In my opinion,
[15:27] both these codecs gave a sharper, more
[15:29] colorful image than the three previous
[15:31] codecs. And if I had to choose between
[15:33] them, it's a difficult one as the
[15:36] differences are very very marginal. For
[15:38] me, after this test, I would choose HEVC
[15:43] 10 bit just for the tiny difference in
[15:45] consistency. But of course, this is my
[15:48] opinion. At the end of the day, it is
[15:50] your choice. Prefer latency, pick H.264,
[15:53] which consistently had the lowest
[15:55] latency overall. prefer graphical
[15:57] sharpness and better colors, then either
[16:00] HEVC 10bit or AV110 bit would have you
[16:04] covered. Plus, for consistency across
[16:06] the test, I had no adaptive
[16:09] quantization, no two passing coding, no
[16:13] adjustment of the horizontal field of
[16:15] view sliders, plus synchronous spacew
[16:17] walk and snapdragon game super
[16:19] resolution were turned off. These
[16:21] settings can have a massive influence on
[16:24] performance, but can also cause issues.
[16:26] I have done a full video on this subject
[16:28] here. So, if this video has piqu your
[16:31] interests, go check it out for some
[16:33] serious performance gains with just a
[16:35] few little tweaks. As always though,
[16:38] what do you think? Do you prefer lower
[16:40] latency over graphical fidelity, or are
[16:43] you like me, a massive graphics tart who
[16:46] must have all the shiny things? You know
[16:49] the drill. Get involved and comment down
[16:51] below. Well, that's it for today. If you
[16:54] enjoyed this content, please hit the
[16:55] like button. The algorithm loves the
[16:57] likes. If you love this content, please
[17:00] join my channel membership like these
[17:02] lovely people did. You get custom
[17:04] badges, emojis, and early access to most
[17:06] of my content. If you want to watch more
[17:09] content from me, please click here or
[17:11] here. Thanks for watching. I'll see you
[17:15] on the other