Youtube Encoding: Upscaling to 4K before uploading doesn't work

Started by 50mm1200s, August 14, 2018, 05:33:59 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

50mm1200s

TL;DR: Upscaling to achieve better youtube quality is a waste of time.

Following previous discussion, I decided to do my own tests since I couldn't find any good article about this.
The hypothesis: upscaling 1080p videos to UHD (3840x2160px) before uploading to youtube increases the video conversion quality for 1080p.
The argument for this is that youtube uses VP9 codec for resolutions higher than 1080p, instead of H.264.

Test Methodology

1- 20s video using 50D MLV with resolution of 1920x1080px
2- Processed with MLVApp (no sharpness or denoise filter) and exported to DNxHR HQ 8bit. This file was used as the reference file.
3- Interpolated the reference file to 3840x2160px on Premiere Pro v12.0.0 (with "Use Maximum Render Quality" box enabled) and exported to DNxHR HQ 8bit. I didn't wanted to use some better interpolation algorithm, like NNEDI3 or SuperResolution-based. Most people just use Premiere or FinalCut to do this.
4- Uploaded both and waited for youtube conversion. Note: youtube is now accepting direct DNxHR uploads.
5- Downloaded with youtube-dl using format code 137 (standard for 1080p H.264 files) and 248 (standard for 1080p VP9 files). Youtube-dl is a python script that gets the original files from googlevideo server, so there's not conversion in-between.
6- Converted to raw YUV420p using ffmpeg
7- Analyzed with VQMT tool (600 frames), to get MSSSIM, SSIM and VIFP values. Output to CSV.
8- Graphs using LibreOffice Calc

Non-Interpolated link: https://youtu.be/HtR-hJe-2Wo
Interpolated link: https://youtu.be/JMbeoiblla8


Note

Interpolation adds artifacts and downscaling can add what seems to be sharpness. As you can see in the example bellow, the "re-interpolated" crop (using Bicubic algorithm), seems to be sharper. This can affect the test, but I did not applied any sharpness on any of the files.



Results

Averaged from 600 frames. The closer to 1.0, the better:







Averages

MSSSIM:
Non-Interpolated H.264: 0.989555
Interpolated H.264: 0.989429
Interpolated VP9: 0.991146

SSIM
Non-Interpolated H.264: 0.956130
Interpolated H.264: 0.956211
Interpolated VP9: 0.962942

VIFP:
Non-Interpolated H.264: 0.635854
Interpolated H.264: 0.636947
Interpolated VP9: 0.677739

Percentage difference between the Non-Interpolated H.264 and the Interpolated VP9 averages:
- SSIM: 0.68%
- VIFP: 4.18%
- MSSSIM: 0.15%

Conclusion

The difference is placebo or induced by the added sharpness from the re-interpolation.
Besides wasting time, processing power and network bandwidth, you'll generate interpolation artifacts (I noticed some distortion in fine details on shadows). VP9 has a better compression ratio than H.264, this doesn't mean youtube keeps the same bitrate as H.264. In fact, that's the point of the lossy compression research: less data for the same perceptual quality, so you need less storage and load the video faster.
Haven't tested with HDR uploads (10-bit, Rec.2020).

tonij

Screenshot of both videos at 1080p,
Upscaled upload looks way better?






masc

Quote2- Processed with MLVApp (no sharpness or denoise filter) and exported to DNxHR HQ 8bit. This file was used as the reference file.
3- Interpolated the reference file to 3840x2160px on Premiere Pro v12.0.0 (with "Use Maximum Render Quality" box enabled) and exported to DNxHR HQ 8bit. I didn't wanted to use some better interpolation algorithm, like NNEDI3 or SuperResolution-based. Most people just use Premiere or FinalCut to do this.
4- Uploaded both and waited for youtube conversion.

But this way you upscale also the artefacts coming from conversion in point 2?! I think this way the upscaled version can't be better (until this step). What happens if you resize in MLV App? Maybe the resize algorithm from ffmpeg is not as good as the one from Premiere, but you don't upscale compression artifacts.
5D3.113 | EOSM.202

50mm1200s

Quote from: tonij on August 14, 2018, 11:32:10 AM
Screenshot of both videos at 1080p,
Upscaled upload looks way better?

See the chart. The static image scores higher on all metrics for VP9, but the movement is down on all metrics too. The average difference between the two is 0.15% on MSSSIM, this is not perceptual in moving images (we can only perceive this looking on a single frame, not on 24fps).
Also, your screen shot is not the correct way to test it, as you don't know if youtube is using H.264 or VP9, since youtube adapts the format and resolution according with your network speed (using DASH - Dynamic Adaptive Streaming over HTTP). The higher perceived quality can also be from the interpolation added sharpness.

Quote from: masc on August 14, 2018, 12:00:59 PM
But this way you upscale also the artefacts coming from conversion in point 2?!

Yes, but DNxHR has virtually no artifact due to high bitrate. As I said on step "3", I tried to keep the 'normal' workflow as the average user would do (upscaling using Premiere). Another point: if you upscale before editing, you'll need much more space to store each sequence. Even more impractical then just upscaling after the finished project.

Quote
I think this way the upscaled version can't be better (until this step).

The file will be re-encoded by Youtube anyway. The rationale behind this "upscaling to 4k" idea is that YT converts the "4k" file directly to VP9 instead of low-bitrate H.264. But, as the results show, the perceptual quality is very little. The biggest difference was on averaged VIFP, with 4.18%.

Quote
What happens if you resize in MLV App?

I can test again, but the results will be the same...

Audionut

I've added some code to your post to reduce the image sizes until the images are clicked.
Nice work.  I hope to have some more time in the future to add further comment.

Regards.

allemyr

First thanks for doing the test.

If someone want to check the downloaded files from the example in this thread please go ahead. Downloaded with the youtube-dl. And uploaded to my dropbox with links below.

1080p
https://www.dropbox.com/s/sqjs3wi33klybmk/original%20dnxhr%20hq%208bit%20%28569MB%29-HtR-hJe-2Wo.mp4?dl=0

UHD
https://www.dropbox.com/s/3t2qpxb6fzua0u4/interpolated%20dnxhr%20hq%208bit%20%282%2C20GB%29-JMbeoiblla8.webm?dl=0

I know you can download with different quality settings with youtube-dl, this is set to "bestvideo" in command prompt and this is the videos that are streamed when viewing the highest quality avaiblie on Youtube.

If you want 50mm1200s you can upload the files you downloaded from Youtube with youtube-dl, please do that i guess they wont be bigger then 200mb each. Since this is not the original DNxHR files, its the streaming one we are looking at.
Yes FFMpeg/Youtube support DNxHR as you say and as I say in all the previous threads, they have done that for a while.

I know raw YUV420p tends to be very big especially at UHD resolution, but can you please provide us here a short version of that here maybe just 3-10 frames instead of 250 or similar. My UHD 250 frame raw YUV420p file was 8gigabyte so I understand that thats hard to share.

Measuring image/video quality with SSIM is ofcourse a good tool.

I dont want to go to much further, but even if the quality is not equally good in your video I see the same difference between Youtube UHD and Youtube 1080p, but thats when viewing with my eyes as I have done for years when checking for different compression artifacts when viewing videos at online streams. That way I went with uploading Prores 422 on Vimeo at 1080p since UHD is very limited with the Vimeo Plus plan.

Please respond in this thread 50mm1200s.


trailblazer

Been lurking for a long time, just registered to post this, I just can't really understand what's going on here.

I've uploaded three clips to youtube in 1920, 2520 and 3840 horizontal resolution. Source clips are 2520x1080 10-bit lossless from a 650D.

Here goes (please mute the sound, forgot to mute those noisy nitro rc cars :) )

1920  https://www.youtube.com/watch?v=M41l0S_TJ-8

2520 https://www.youtube.com/watch?v=CqfdynevqwU

3840 https://www.youtube.com/watch?v=RsknYTpXGTk

On my HD screen I definitely see a difference between the clips. The 3840 clip seems to have more information in the shadows, and the colors are true to the source clip. Yes, bold statements, but that's what I see on the youtube playback. Check for yourself at 14sec, the red on the lifebouy changes between the 3840 and the other uploads. Also the bark on the trees are more defined on the 3840 clip.

At 33 seconds the 3840 upload handles motion WAY better than the lower resolution uploads, resolution and colors are also way better kept at 38 seconds in.

On average the difference is subtle, but it's there. Sometimes it's not subtle at all, at 38 seconds I really see a big difference. In my eyes, if you want to retain the absolute highest possible quality from a 2520x1080 source to YT, a 4k upscaled upload seems to be necessery (for now).

I have not downloaded these clips from youtube and tested them like 50mm1200s did (anybody is welcome to do that!), but is the actual playback on YT good enough to spot perceived quality differences? I mean at 33 and 38 seconds the perceived difference in quality is obvious. Although I do not know if it's VP9 or h.264.

My workflow is as follows: MLV app -> CDNG -> Resolve. For the 3840 upload i chose a 4k timeline. All clips were exported using "force sizing to highest quality" and "force debayer to highest quality". Exported using DNxHR 444 12-bit. Sharpening is the same (in camera raw in Resolve).

50mm1200s

Quote from: Audionut on August 16, 2018, 06:31:18 AM
I've added some code to your post to reduce the image sizes until the images are clicked.
Nice work.  I hope to have some more time in the future to add further comment.

Regards.

Thanks. Feel free to change anything if you want, I'm sure there's some typos there (english is not my first language).

50mm1200s

Quote from: allemyr on August 16, 2018, 12:40:15 PM
I know you can download with different quality settings with youtube-dl, this is set to "bestvideo" in command prompt and this is the videos that are streamed when viewing the highest quality avaiblie on Youtube.

The "-f best" will download the highest quality and that's not what we are testing here. We are testing 1080p vs 1080p, not 1080p vs 4K...

Quote
If you want 50mm1200s you can upload the files you downloaded from Youtube with youtube-dl, please do that i guess they wont be bigger then 200mb each.

Why? You can download with youtube-dl. It will be the same file. Here's the SHA256 (same files used on the test before transforming to raw YUV):

Non-Interpolated H.264: 9ACAB309BA66A05D4B8F8D592C2CF6CC00C1817F6CD4D5F6EDCF977A7F9DD45D
Interpolated H.264: 9B271ACD4833E9F3BEFBCAE05D8432CFEADD18833F361E5E1461B3CB87DCF8AF
Interpolated VP9: D501E3DCF287EFADB9A9A5C661C178DC7A8341E60E7625424DC14B2E00A888D1

Quote
I know raw YUV420p tends to be very big especially at UHD resolution, but can you please provide us here a short version of that here maybe just 3-10 frames instead of 250 or similar. My UHD 250 frame raw YUV420p file was 8gigabyte so I understand that thats hard to share.

Here's the original DNxHR:
https://drive.google.com/file/d/1h8w-jbb6mhEnDxw3qsaGxwOcmDxWPrTQ

You can convert the original file to raw YUV420 using ffmpeg:


ffmpeg -i input.mov -c:v raw -pix_fmt yuv420p out.yuv


Quote

I dont want to go to much further, but even if the quality is not equally good in your video I see the same difference between Youtube UHD and Youtube 1080p,

That's because you're probably comparing 1080p with 4K, so the resizing done by your browser adds some sharpening. This doesn't mean the compression is better...

You cannot compare using SSIM with different resolutions (it needs to match the reference footage), but I could try to download the 3840x2160px VP9 (format code 313) and resize it to 1920x1080p as the browser would do through youtube player in a fullscreen FullHD display. Not sure if this would work.

50mm1200s

Quote from: trailblazer on August 16, 2018, 08:53:38 PM
Been lurking for a long time, just registered to post this, I just can't really understand what's going on here.

Can you send me the original file you uploaded? I can do the test again, using your files, if you want. It needs to be without interpolation, without sharpening and in a format without much lossy compression, like ProRes or DNxHR.

allemyr

Quote from: 50mm1200s on August 17, 2018, 03:35:58 AM
The "-f best" will download the highest quality and that's not what we are testing here. We are testing 1080p vs 1080p, not 1080p vs 4K...

Why? You can download with youtube-dl. It will be the same file. Here's the SHA256 (same files used on the test before transforming to raw YUV):

Here's the original DNxHR:
https://drive.google.com/file/d/1h8w-jbb6mhEnDxw3qsaGxwOcmDxWPrTQ

You can convert the original file to raw YUV420 using ffmpeg:


ffmpeg -i input.mov -c:v raw -pix_fmt yuv420p out.yuv


That's because you're probably comparing 1080p with 4K, so the resizing done by your browser adds some sharpening. This doesn't mean the compression is better...

You cannot compare using SSIM with different resolutions (it needs to match the reference footage), but I could try to download the 3840x2160px VP9 (format code 313) and resize it to 1920x1080p as the browser would do through youtube player in a fullscreen FullHD display. Not sure if this would work.

Hi again.

1. Ok so you don't download the video files from Youtube either with "bestvideo" nor "-f best", ok, but what setting do you download them with? No we are not mainly comparing 1080p v UHD/4k we are comparing what details is left after uploading, and what is smudged out.

2. Thanks for the SHA256 stuff, don't know of SHA256. And I guess not to many of the other users here know how to download with Youtube-dl, and especially not with SHA 256 adress.

3. No I won't use the original DNxHR video to make any comparission, we are talking about Youtube compression here, not any local video in 1080p or another resolution that hasnt been uploaded to Youtube.

Well, we won't come any further i suppose. In the previous thread that you link to, you havent really watched my screen captures at all? The loss in detail after uploading a 1080p video to Youtube is so obvious to me, so if you cant see that with your eyes, I dont know what you really can observe with your own eyes sorry.

If you want do make it easier for us. Please give us the setting you use with Youtube-dl script, if it's not "f-best" or "bestvideo", what is the setting you download them from youtube with?
Further, that SSIM tool doesnt seem to be able to measure resolutions like 3840x2160 as its to highres, I get a message when I try to do that anyway. at 1920x1080 it says that it has to be divided by 8 so 1920x1088 pixels cause 1080/8=135 which is a odd number, and 1088/8 is 136 which seams to be fine. Do you upscale the video to 1920x1088 before doing the SSIM test?

Please provide the video you download from Youtube as I asked for before, not in SHA256 please.

Second: Can you provide us with 3-10 frames of each raw YUV420 you do tests on.

This turned out to be quite ridiculous conversasion. Guess we wont get anything out of this, since we havent manage anything until now.


50mm1200s

@allemyr everything you asked is on "test methodology" section. Just read, please.

chrisfrancia

I read your original post but none of the replies

I was under the impression the increase in quality for uploading 1080p as 4k was not by upscaling the 1080 but by exporting in the 4k preset. I am not sure what difference if any this would make if any over what you did but, from personal experience (in tests SEVERAL years ago) there was a definite increase in quality when selecting 4k over 1080p in youtube although I dont think i compared a 1080 export in my test.

My test:

I shot 1080p ML RAW, I edited in 1080p, I exported 4k youtube preset, I compared the 4k button on youtube vs the 1080p and it looked better in 4k.



Also! I am not trying to start an arguement, I would love if you told me that it was pointless and I could save time by just exporting to 1080p.
[5D3 1.2.3 ML] [6D 1.1.6 ML]

allemyr

Quote from: chrisfrancia on August 27, 2018, 04:40:12 AM
Also! I am not trying to start an arguement, I would love if you told me that it was pointless and I could save time by just exporting to 1080p.

It's pointless and it doesn't work!

Edit: A joke ofcourse, since if you get a reply it will be that, read the first post again.

What preset are you talking about? From what application, video codec and container? There are tone of different presets when exporting a video i don't think that suites this thread. 50mm1200s and me are exporting to Youtube in high quality with DNxHR HQ codec in UHD and 1080p, no presets!! And no H264 or other at 5mbit/s.

Check "DNxHR HQ" out if you want to!

allemyr

Quote from: 50mm1200s on August 23, 2018, 07:50:49 PM
@allemyr everything you asked is on "test methodology" section. Just read, please.

Hi,

Yes i missed the format code in your first post 137 for the fullHD upload and 248 for 1080p from UHD upload.

This question is still current tho: "Can you provide us with 3-10 frames of each raw YUV420 you do tests on?"
Or can you describe the upscaling process to 1088 from 1080 and more about the processing with settings when going from downloaded Youtube 1080p file to raw YUV420p?

If there is not a high loss in quality within that 1080p to raw YUV420p I'am very surprised that the VQMT tool can't spot the difference in details and quality.

I downloaded my own videos with the 137 and 248 code to get both at 1080p. Shure the 1080 VP9 download from the UHD upload has less details, but only by a touch, still visible tho.
The big difference comes to the 1080p no upscale upload, its very compressed and a big loss in details.

Before giving the example images I would like to talk about this UHD upscale thing and workflow. If your workflow is just from MLV to CDNG with no filter/effect (just plain MLV to 16bit CDNG) then straight to Resolve where edit and grade happen, and after that export the final video from Resolve. The UHD/4k upscale thing is almost ZERO extra effort. But if you workflow is something else I know there are a tone of different workflows including MLVFS/Lightroom/Premiere/Resolve what ever it is, like exporting video and proxies thru different application, this is ofcourse a lot of extra hassel, so yes, I shouldnt do this upscale thing with a complicated workflow, your workflow isnt suitable for that. With that said here comes an update I don't know if anyone is going to look at these but I upload them anyway.





UHD downscaled to 1080p with bicubic(smooth gradients) in Photoshop - This is for me the highest quality of the image





1080p from UHD - not to far behind in quality but still less details and more compression then viewing UHD on Youtube (even at a 1080p screen)





1080p from non upscaled 1080p video from Youtube - This is a big decrease in image quality and detail with a high visible compression. Big loss in details, highly visible in the brickfacade on the black building.





No crop UHD screen shot






No crop 1080p from UHD screen shot





No crop 1080p no upscale screen shot.

https://www.youtube.com/watch?v=M3j4NP3YchY UHD Youtube video

https://www.youtube.com/watch?v=QQcfUymb8bc 1080p Youtube video
 
Please dont talk about added sharpening or filters, its bicubic on the UHD downscaled example. You can't add sharpening to the no upscale 1080p Youtube upload to get those details in the brick facade, they are smuded out and lossed.

trailblazer

Quote from: 50mm1200s on August 17, 2018, 03:38:31 AM
Can you send me the original file you uploaded? I can do the test again, using your files, if you want. It needs to be without interpolation, without sharpening and in a format without much lossy compression, like ProRes or DNxHR.

Didn't read your original post correctly, thought this was about 4k playback also. When I look at my upscaled 4k upload in 1080p I agree with what you said in the original post, the difference is almost non-excistent.

However, I still stand by what I said in my above post. When comparing the clips, I viewed the 3840 upload in 4k. The 4k playback of the upscaled 2520x1080 material looks higher quality than the 1080p playback of the 2520x1080 and downscaled 1920x823 uploads, even though I only have an HD screen.

Here are some examples of the 4k playback vs 1080p (2520x1080 upload) on my screen:

4k


1080p


4k 100x100 crop scaled x6


1080p 100x100 crop scaled x6


4k


1080p


4k 100x100 crop scaled x6


1080p 100x100 crop scaled x6 (notice the red has slightly shifted and is sort of "melting", the yellow on tab on the left is almost gone, red bars in center are gone)




allemyr

Quote from: trailblazer on August 27, 2018, 03:14:19 PM
Didn't read your original post correctly, thought this was about 4k playback also. When I look at my upscaled 4k upload in 1080p I agree with what you said in the original post, the difference is almost non-excistent.

However, I still stand by what I said in my above post. When comparing the clips, I viewed the 3840 upload in 4k. The 4k playback of the upscaled 2520x1080 material looks higher quality than the 1080p playback of the 2520x1080 and downscaled 1920x823 uploads, even though I only have an HD screen.

Hi,

Its a lot of different resolution involved here maybe, but please read carefully again. The 2520x1080 is probably processed in a similar way as 1080p is.
What we are comparing is a non upscaled 1080p video compared to other videos, especially UHD/4k videos that has been upscaled. You are agree on the wrong thing in your first sentence sorry. You are comparing a 4k upload with a 4k upload when viewing the same video at different resolution. Please compare different uploads as you do with your comparisson in the images. We are comparing UHD/4k uploads viewed at 1080p or UHD  with 1080p uploads,

The 6x crop is not 1:1 pixels so difficlut to see, it is in fact as you say 6 pixels describing 1 pixel crops should be 1:1 pixel, no need to do a little crop and then interpolate by zooming in 600%.

I can see in your "2520test" video that the quality is higher in the shadows compared to UHD.

Glad you are comparing yourself but the comparison might be slightly defective since you realy on sharpening filters on the browser when viewing Youtube videos with 2520 width on a 1080p screen, same thing as the UHD video. Shure you can see a difference but they don't say as much since its different filteres applied on browser when going to various resolutions.


trailblazer

Quote from: allemyr on August 27, 2018, 04:13:35 PM
You are agree on the wrong thing in your first sentence sorry. You are comparing a 4k upload with a 4k upload when viewing the same video at different resolution. Please compare different uploads as you do with your comparisson in the images. We are comparing UHD/4k uploads viewed at 1080p or UHD  with 1080p uploads

Sorry for the confusion, but I don't understand what you mean I agree on the wrong thing. I do not compare the same upload with itself in different resolutions. Originally I have three different uploads.  Youtube downscales the 2520x1080 clip to 1920x822, check "stats for nerds" by right clicking on the playback. 4k upload is looked at in 4k. What I agreed with in my second post was that there is not much difference when I view all three of my uploads in HD playback on youtube, non-excistant might be a little harsh :) When I look at your examples there are more details on the 1080p 248 UHD sample then your 1080p 137 no upscale sample

Quote from: allemyr on August 27, 2018, 04:13:35 PM
The 6x crop is not 1:1 pixels so difficlut to see, it is in fact as you say 6 pixels describing 1 pixel crops should be 1:1 pixel, no need to do a little crop and then interpolate by zooming in 600%.

It's a 100x100 crop scaled to 600x600 in gimp with "none" set at interpolation, so we can get an idea of what's different up really close (of what I see on my screen anyways)

Quote from: allemyr on August 27, 2018, 04:13:35 PM
Glad you are comparing yourself but the comparison might be slightly defective since you realy on sharpening filters on the browser when viewing Youtube videos with 2520 width on a 1080p screen, same thing as the UHD video. Shure you can see a difference but they don't say as much since its different filteres applied on browser when going to various resolutions.
2520 clip doesn't have 2520 width on yt, so don't worry about that. Are you sure about added sharpening by downscaling from 4k in the browser? The 4k playback on youtube looks very similar to my non-upscaled 2520 DNxHR clip, can't say I see any extra sharpening going on. Wish I could see all this on a 4k screen though

So I'm not really sure what to make out of all this, in my experience I have to upscale before uploading to youtube and view it in 4k if I want to keep the quality of my source footage, even if it's not 2520x1080. It's a little silly, yes, would much rather just export and upload in source resolution.

allemyr

Quote from: trailblazer on August 27, 2018, 08:38:23 PM
Sorry for the confusion, but I don't understand what you mean I agree on the wrong thing. I do not compare the same upload with itself in different resolutions. Originally I have three different uploads.  Youtube downscales the 2520x1080 clip to 1920x822, check "stats for nerds" by right clicking on the playback. 4k upload is looked at in 4k. What I agreed with in my second post was that there is not much difference when I view all three of my uploads in HD playback on youtube, non-excistant might be a little harsh :) When I look at your examples there are more details on the 1080p 248 UHD sample then your 1080p 137 no upscale sample

1 When I look at my upscaled 4k upload in 1080p I agree with what you said in the original post, the difference is almost non-excistent.

2 However, I still stand by what I said in my above post. When comparing the clips, I viewed the 3840 upload in 4k. The 4k playback of the upscaled 2520x1080 material looks higher quality than the 1080p playback of the 2520x1080 and downscaled 1920x823 uploads, even though I only have an HD screen.

Ok no my mistake, it was I who missread then was a bit confused about the previous post with 1 and 2, which I pasted above.

Its good that you make your own comparison, and also check others, and I agree this upscale thing is a bit confusing because of the change in the amount of pixel, but the thing that it takes ways around that heavy compression on Youtube, Vimeo had better quality of 1080p footage and has had for a really long time, like 8 years?.

I have a UHD screen myself mostly for when working in applications, not so much to watch stuff in UHD videos. I think 1080p is plenty of pixels and think this UHD/4k commercial gimick is pretty funny. But even tho 1080p is enough, I'am allergic to heavy compression. When 1080p on Youtube was the high standard like 8 years ago, it looked much better then 720p uploads, but that was the same thing there that the 720p was bad because of more compression artifacts then 1080p, it was not the amount of pixels with 1080p uploads that made it look so much nicer it was the less compression.

Yes I agree about 1080p is enough pixels. This hole thing is about compression as everyone here understand. When I uploaded things to Vimeo with Plus member I always used Prores instead of the recommended 5mbit/s when uploading. But still 1080p on Vimeo could even be less compressed then it is. I think UHD/4k on Youtube is very high quality for a stream that you don't pay for with a membership plan, better quality then Vimeo 1080p. Its all about compression :)

If you want to watch and make crops and stuff in UHD/4k you could download Youtube-dl and use it in Windows Command Prompt, there you can download all various videoqualities that you stream on Youtube, watch it in VLC at 1:1 or something or just extract single images directly in your image application. But take care of resolution and different scaling. Always when you scale things there is a filter applied, there are many different variation, but always some sort of filtering involved both in image application or your web browser.

If it suits your workflow to upload 2520px wide videos or 1920px wide videos you should ofcourse do that, I have a workflow where its so little effort to do the upscale thing, so I do that since I as I said I'am allergic to compression.

chrisfrancia

Quote from: allemyr on August 27, 2018, 07:34:07 AM
It's pointless and it doesn't work!

Edit: A joke ofcourse, since if you get a reply it will be that, read the first post again.

What preset are you talking about? From what application, video codec and container? There are tone of different presets when exporting a video i don't think that suites this thread. 50mm1200s and me are exporting to Youtube in high quality with DNxHR HQ codec in UHD and 1080p, no presets!! And no H264 or other at 5mbit/s.

Check "DNxHR HQ" out if you want to!

I was refering to Adobe Premiere's H264 Youtube 4k Preset
[5D3 1.2.3 ML] [6D 1.1.6 ML]

allemyr

Quote from: chrisfrancia on September 06, 2018, 03:15:43 AM
I was refering to Adobe Premiere's H264 Youtube 4k Preset

Ok but still its a only a preset for setting the H264 options automatically. Well you can ofcourse use that option if you want but its not suited for this forum thread.

Your PPro 4k Youtube preset is 5megabytes/s which will demolish a big part of the quality especially after Youtube re-encodes it, DNxHR HQ is over 80megabytes/s and more like a Prores format and FFMPEG support it and therefore Youtube aswell.
If you want good quality you should use that if you are on Windows, if you are on MacOS you might choose Prores instead.

UHD/4k is a lot of pixels and definetely not suited to go into 5mb/s compression and after that compress it again.

My advice don't use this Youtube Premiere Pro preset.

allemyr

This is a fun thread, now when watching the general popular creators on Youtube they do the same, shooting in 1080p and uploading in 4k/UHD.

For example PotatoJet explains he do that when shooting with his A7s3 in 10-bit in this video https://youtu.be/Sc4hSKkDDIU
And also says its the same for all of them like Peter McKinnon, Matti Haapoja. So what I saw in quality has to be right. And I have to say that PotatoJets 1080p video looks superb uploaded in UHD.

To bad 50mm1200s found MagicLantern forum, and glad I stayed with what I saw and didnt view his graphs that compared nothing. I should move on and go forward yes. But its cool that 2021 people still shoot in 1080p and upload to UHD