WEBVTT

00:00:00.240 --> 00:00:07.720
let me start by saying this video was a metric ass ton of work I mean not for me

00:00:05.279 --> 00:00:11.200
mind you Ed did all the heavy lifting so make sure that you show him some love by

00:00:09.400 --> 00:00:15.879
liking the video or following him on Twitter or something because his

00:00:12.799 --> 00:00:19.119
findings are absolutely fascinating if

00:00:15.879 --> 00:00:22.800
you consume or create video content on

00:00:19.119 --> 00:00:26.480
the web welcome to 4K better video

00:00:22.800 --> 00:00:26.480
quality or a big fat

00:00:28.960 --> 00:00:35.399
placebo

00:00:35.399 --> 00:00:43.840
Intel is bringing ddr4 to the mainstream with their allnew core i76700 K and core

00:00:40.800 --> 00:00:46.079
i56600k processors check out the link in

00:00:43.840 --> 00:00:51.280
the video description to learn more so Ed's investigation into 4K video quality

00:00:48.600 --> 00:00:57.359
was actually inspired by a Panasonic camera known as the GH4 at a mere $2,000

00:00:55.480 --> 00:01:03.039
trust me that was amazing when it came out the GH4 could shoot 4K to an

00:01:00.039 --> 00:01:04.439
internal SD card at a bit rate of 100

00:01:03.039 --> 00:01:11.720
megabit per second and you can learn more about bit rate and why it's important here which for the first time

00:01:08.680 --> 00:01:13.960
put 4K video production like prograde

00:01:11.720 --> 00:01:17.920
production in the hands of regular consumers who happen to be in the market

00:01:15.600 --> 00:01:22.520
for what is also a fairly capable mirrorless Stills camera balling right

00:01:20.560 --> 00:01:26.360
we sure thought so we picked one up pretty much right away and the vast

00:01:24.479 --> 00:01:32.759
majority of our b-roll has been shot on the GH4 for months with us going as far

00:01:29.360 --> 00:01:37.360
at one Point as to shoot 4k on our aam a

00:01:32.759 --> 00:01:40.040
Sony fs700 4K on our bcam the GH4 and

00:01:37.360 --> 00:01:46.119
edit on a 4K timeline with a native export of 4K video that we upload but

00:01:44.119 --> 00:01:51.320
this proved to be very challenging at the time the fs 700's High bit rate 4K

00:01:49.000 --> 00:01:55.719
footage was a total bear to work with the file sizes were huge putting

00:01:53.880 --> 00:01:59.520
significant strain on our networking infrastructure when multiple editors

00:01:57.399 --> 00:02:04.240
were working at the same time and we experien other workflow interruptions

00:02:01.600 --> 00:02:08.280
ranging from poor editing timeline uh scrubbing responsiveness to annoying

00:02:06.840 --> 00:02:12.280
file splitting during long clip recording to even export errors when

00:02:10.239 --> 00:02:18.599
rendering out the final videos so we went back to 1080p downscaled from 4K

00:02:15.879 --> 00:02:23.879
using our external recorder on the aam since the fs700 sensor performs so well

00:02:21.200 --> 00:02:29.560
at resolving fine details regardless of the output file resolution anyway and

00:02:26.280 --> 00:02:32.000
continued shooting b-roll at 4k natively

00:02:29.560 --> 00:02:36.040
on the GH4 since while that does slow down the editing timeline it's not in a

00:02:34.280 --> 00:02:40.840
manner that completely cripples the productivity of our editors like the 4K

00:02:38.640 --> 00:02:47.120
native FS footage did so what you've been seeing out of us since then this is

00:02:42.840 --> 00:02:50.080
really important is a 4K timeline with

00:02:47.120 --> 00:02:56.280
downsampled 4K footage off the fs700 that is then re-up sampled back to 4K

00:02:53.319 --> 00:03:02.200
and Native 4K footage from the GH4 that is then exported at 4K when we create

00:02:59.159 --> 00:03:03.799
the file to upload to Youtube or vessel

00:03:02.200 --> 00:03:08.280
fast forward almost a year and we were moving into a new office and rolling out

00:03:05.680 --> 00:03:15.599
three new servers a rendering server powered by two Intel Zeon E5 2699 v38

00:03:12.239 --> 00:03:19.720
core processors a high-speed SSD based

00:03:15.599 --> 00:03:24.080
20 gbit network storage server and a 100

00:03:19.720 --> 00:03:27.000
tbte plus archival storage server these

00:03:24.080 --> 00:03:32.560
potentially addressed almost all of the previous issues Network performance file

00:03:30.159 --> 00:03:36.400
size and rendering horsepower not to mention that adobe hasn't been sitting

00:03:34.360 --> 00:03:40.200
on ass for the last year and Premiere has improved a fair bit as well in terms

00:03:38.319 --> 00:03:45.560
of how it handles high bit rate and 4K files so it was time to evaluate whether

00:03:42.599 --> 00:03:49.920
it was worth it to go native 4K again as part of an investigation into how we

00:03:47.280 --> 00:03:54.799
could use our 36 core video crunching server to improve our workflow so we

00:03:52.319 --> 00:03:59.400
started by taking a closer look at some of the 4K footage that we'd shot

00:03:56.640 --> 00:04:02.879
recently on the GH4 comparing the visual quality

00:04:00.280 --> 00:04:08.000
of the Native 4K footage and then that very same footage down sampled to 1080p

00:04:05.959 --> 00:04:12.640
with a variety of codec trying to figure out how to get timeline performance

00:04:09.680 --> 00:04:17.359
perfect and the results shocked us when viewed on a 4k monitor they were

00:04:15.199 --> 00:04:23.080
indistinguishable from each other and in fact even when compared using overlaid

00:04:20.440 --> 00:04:28.960
still captures in Photoshop and by setting the blending mode to difference

00:04:25.440 --> 00:04:31.919
they were nearly identical so then was

00:04:28.960 --> 00:04:36.960
there any point recording in 4k at all we needed to dig deeper so we

00:04:33.960 --> 00:04:41.280
specifically recorded identical Clips at

00:04:36.960 --> 00:04:43.680
4K and 1080p natively on the GH4 camera

00:04:41.280 --> 00:04:49.400
and compared those against each other this time with the 1080p footage

00:04:45.840 --> 00:04:52.240
upsample to 4K same result down sampling

00:04:49.400 --> 00:04:57.080
4K to 1080p upsampling 1080p to 4K they just all look the same so we broke out

00:04:55.120 --> 00:05:01.680
the big guns and used our proper professional grade camera our Sony fs7

00:04:59.880 --> 00:05:07.320
100 to do the same comparison now we can see the difference

00:05:05.120 --> 00:05:12.240
watching the files over our own local network between the upsampled and the

00:05:09.639 --> 00:05:17.240
native 4K footage with the text on our figurine appearing to be more in focus

00:05:15.000 --> 00:05:22.120
when watching back the native 4K file even though the camera and subject were

00:05:18.840 --> 00:05:24.479
not moved in between shots but what does

00:05:22.120 --> 00:05:29.280
this difference translate to in terms of a real world benefit we all know that

00:05:27.120 --> 00:05:33.280
YouTube has had a 4K option for quite some time but should content creators

00:05:31.520 --> 00:05:38.280
even be worried about shooting and editing in 4k should they export in 4k

00:05:36.240 --> 00:05:43.479
for their upload should you the viewer select 4K when watching we have all

00:05:41.240 --> 00:05:46.919
those answers but Buckle in because if the rest of this video sounded pretty

00:05:44.840 --> 00:05:51.080
technical this is probably going to be a little worse I'll start with the answer

00:05:48.680 --> 00:05:56.080
for the content creators since this affects the content consumers as well

00:05:53.919 --> 00:06:01.960
regardless of your camera working on a 1080p timeline with 1080p footage will

00:05:59.360 --> 00:06:06.639
not not result in a degraded experience for your viewers thanks to the video

00:06:04.160 --> 00:06:11.280
processing done by YouTube the benefit of your fancy camera is basically

00:06:08.960 --> 00:06:16.080
limited to better performance on a green screen and better color grading and

00:06:13.720 --> 00:06:18.840
punching options in post- production something that is arguably much more

00:06:17.759 --> 00:06:27.280
important to the appearance of the footage than sheer resolution anyway so I'm not poo pooing fancy cameras but we

00:06:23.759 --> 00:06:30.680
also discovered that upsampling to 4K

00:06:27.280 --> 00:06:33.479
for your final export is worth the extra

00:06:30.680 --> 00:06:37.400
file size and the extra rendering time because when YouTube detects a

00:06:35.120 --> 00:06:43.479
resolution of 4k on their side they will offer a 4K viewing option to the viewer

00:06:40.000 --> 00:06:45.720
that features a higher playback bit rate

00:06:43.479 --> 00:06:52.080
and therefore sharper image quality on both 1080P and on 4K displays which

00:06:50.080 --> 00:06:56.440
leads us finally to the answer for consumers conventional wisdom would

00:06:54.520 --> 00:07:00.879
dictate that watching content at the native resolution of your display is

00:06:58.680 --> 00:07:05.840
ideal this is something that most of our viewers seem to be aware of based on

00:07:03.039 --> 00:07:11.000
these poll results but if you have the internet connection and bandwidth cap

00:07:07.840 --> 00:07:13.800
limits to handle it this is not the case

00:07:11.000 --> 00:07:21.160
for online video or at least on YouTube watching at 4k on a 4K or 1080p monitor

00:07:18.199 --> 00:07:25.479
yields a noticeable visual quality improvement regardless of whether the

00:07:23.080 --> 00:07:30.639
video was natively shot at 4K or upsampled thanks to YouTube delivering

00:07:27.639 --> 00:07:32.319
the video to you at a higher bit rate

00:07:30.639 --> 00:07:35.879
which leaves us finally with a last message for the content creators out

00:07:34.039 --> 00:07:41.240
there if there's anything to take away from this video it's don't get too

00:07:38.160 --> 00:07:43.759
focused on 4K or resolution in general

00:07:41.240 --> 00:07:48.879
in the same way that a 4K recording phone is not a professional video

00:07:46.039 --> 00:07:55.199
recording device entrylevel 4K cameras are not equal to their 4K prograde

00:07:51.759 --> 00:07:57.599
Brethren but don't worry most Pros are

00:07:55.199 --> 00:08:02.599
just using their 4K gear to have more flexibility during the creation of their

00:07:59.639 --> 00:08:08.479
1080p final product anyway so if you're producing Web video you may actually be

00:08:05.199 --> 00:08:11.280
better served by a superior 1080p camera

00:08:08.479 --> 00:08:17.400
at this time because lenses encoding settings codecs camera sensor size and

00:08:14.639 --> 00:08:21.800
sensor quality all impact the finished product and on that subject stay tuned

00:08:19.840 --> 00:08:26.039
for the results of our investigation into editing codec it's going to be at

00:08:23.960 --> 00:08:29.560
least as interesting as this one especially For You video editors out

00:08:27.759 --> 00:08:33.919
there thanks for watching guys if you disliked this video you know what to do

00:08:31.919 --> 00:08:37.519
but if you liked it please click that like button maybe support us by buying a

00:08:36.039 --> 00:08:42.360
cool t-shirt like this one changing your Amazon bookmarked one with our affiliate code we have the instructions for how to

00:08:40.279 --> 00:08:46.240
do that in the little eye in the corner or even by supporting us directly

00:08:43.880 --> 00:08:49.000
through our community Forum now that you're done doing all that you guys are

00:08:47.760 --> 00:08:54.120
looking for some more videos to check out boom you should check out our recent

00:08:50.959 --> 00:08:57.680
video on the Acer 34in curved gaming

00:08:54.120 --> 00:08:59.440
monitor that I absolutely freaking love

00:08:57.680 --> 00:09:05.870
uh we just did that one and it's not for 4K but uh is it ever sexy see you guys

00:09:03.880 --> 00:09:07.079
next

00:09:07.079 --> 00:09:10.079
time