Pixelation when rendered by Vegas Pro 15

Comments

AveSatanas wrote on 2/2/2020, 5:22 PM

Im trying alot of test runs to get vokouder to render some quality stuff whilst retaining the 15mbps

 

Musicvid wrote on 2/2/2020, 5:45 PM

If you're using Voukoder assist, head math tells me you "may" be happier at 20Mbps for motion quality, although YT will always output the same file size. Play around and do let us know what you come up with.

Disclaimer: I haven't tested Voukoder on my Intel yet.

john_dennis wrote on 2/2/2020, 6:29 PM

**** Radeon Alert **** Radeon Alert **** Radeon Alert **** Radeon Alert ****

I've been rendering 1080-29.97p at 28 Mbps target / Max 40 Mbps using a Magix AVC/AAC AMD VCE template customized from the UHD starter.

For casual UHD renders, I used the following custom template...

Note the actual PQ measurements in the screenshots. So far, I haven't had to back out of the door when I deliver the video to others.

fifonik wrote on 2/2/2020, 6:31 PM

Yes, YT will re-compress and you cannot change this.

However, you can change the quality or video you supplied to YT. If poor quality supplied, the final quality will be worse than if good quality supplied.

Forget about bitrate with x264. Use CRF regardless frame size and frame rate. The x264 encoder will choose bitrate itself based on frame size, frame rate & scenes to meet requested quality. For example, scenes with many small items and movements (grass, leaves, water) will use higher bitrate while scenes like "talking head" will use much lower bitrate. Value between 18-22 are good starting point (smaller value gives you better quality). I'm using 18 for storing (in extreme cases 16, never used lower values) and 22-23 for sharing.

Last changed by fifonik on 2/2/2020, 6:42 PM, changed a total of 2 times.

Camcorder: Panasonic X1500 + Panasonic X920 + GoPro Hero 11 Black

Desktop: MB: MSI B450M MORTAR TITANIUM, CPU: AMD Ryzen 5700X, RAM: G'Skill 16 GB DDR4@3200, Graphics card: MSI RX6600 8GB, SSD: Samsung 970 Evo+ 500GB (NVMe, OS), Samsung 870 Evo 1TB, HDD WD 4TB, HDD Toshiba 4TB, OS: Windows 10 Pro 22H2

NLE: Vegas Pro [Edit] 11, 12, 13, 15, 17, 18, 19, 22

Author of FFMetrics and FFBitrateViewer

AveSatanas wrote on 2/3/2020, 2:36 AM

i rendered in x264 with crf and tested till i got between 15mbps and i must say there is a huge difference, it looks absolutely quality, cheers @fifonik

fifonik wrote on 2/3/2020, 2:51 AM

As I said already, forget about bitrate. Start thinking in term "visual quality". That is CRF. With the same CRF sometimes you will have higher average bitrate, sometimes -- lower. This is normal as scenes in your videos will be different.

Last changed by fifonik on 2/3/2020, 2:53 AM, changed a total of 1 times.

Camcorder: Panasonic X1500 + Panasonic X920 + GoPro Hero 11 Black

Desktop: MB: MSI B450M MORTAR TITANIUM, CPU: AMD Ryzen 5700X, RAM: G'Skill 16 GB DDR4@3200, Graphics card: MSI RX6600 8GB, SSD: Samsung 970 Evo+ 500GB (NVMe, OS), Samsung 870 Evo 1TB, HDD WD 4TB, HDD Toshiba 4TB, OS: Windows 10 Pro 22H2

NLE: Vegas Pro [Edit] 11, 12, 13, 15, 17, 18, 19, 22

Author of FFMetrics and FFBitrateViewer

AveSatanas wrote on 2/3/2020, 3:27 AM

I sort of have to worry about bitrate though, if it was a project that i was doing for my own use "so be it" but YT compresses the more you go above the recommended bits so i am worried...

Musicvid wrote on 2/3/2020, 3:37 PM

No, YouTube compresses everything to exactly the same parameters, no matter what you give it.

That means that the only thing that matters is uploading optimal quality for YOUR source, without undershooting or grossly overshooting bitrates for YOUR optimal source quality.

Overthinking will work against you in this game. Here:

https://www.vegascreativesoftware.info/us/forum/wagging-the-dog-effects-of-hyperoptimal-youtube-upload-bitrates--114098/

AveSatanas wrote on 2/3/2020, 4:30 PM

cheers @Musicvid. hence why i upload in 1440p upscale so the video on YT gets unlocked to 30mbps and it has more quality and less pixelation, im trying to do more 1080p stuff but it gets locked to avc and not vp09. Vokoduer has helped alot though, it gives a better quality but takes alot longer to render but hey its worth it

Musicvid wrote on 2/3/2020, 4:59 PM

So you're software-upscaling and hardware-encoding 1080 to 1440 on the belief that triggering VP9 at YouTube's end will improve the download quality more than overcoming collateral losses?

I would love to see quantified testing on that alchemic theory, but as I've stated before, I have absolutely no interest in testing it myself. When you've figured out a way to rule out placebo / virtual Pygmalion effect with actual SSIM and PSNR data, I'll be the first to view it, I promise.

Keep inquiring; you're one of the good thinkers, but I'm still a big fan of Newton and Einstein. Did I tell you that when I was twelve, I hooked up a motor to a generator, and created perpetual motion -- for almost two seconds!

 

AveSatanas wrote on 2/3/2020, 6:01 PM

With VP09 and Upscaled to 1440p 60FPS:

 

WithOUT VP09 and 1080P 60FPS:

 

Why i upscale, because YT gives more bitrate. No it does not improve quality as you mentioned @Musicvid but it does allow better playback and less pixelation due to YT allowing more bitrate to playback at 1440p 60fps 30mbps.

fifonik wrote on 2/3/2020, 6:28 PM

Yes, the upscaling video before uploading to YT is known trick to got higher final YT btrate. But I believe that the higher bitrate would only be on the 1440 fake resolution and would be the same for 1080. Will it be better for viewing? Possible. With more bandwidth required, more GPU/CPU required on target device for downscaling.

For proper testing you should create one small project. Then render it with different parameters and upload. Only then it would be possible to compare.

For example, I clicked on your second video somewhere and saw a mesh. This is hard for encoder so if you used to use "bitrate target" for the video it is possible that the scene was rendered in poor quality. Then YT re-encoded it one more time and made it even worse. Funny thing: the first video (better quality) YT shows me in 480 by default, but the second one (very poor quality in my taste) in 1080.

Camcorder: Panasonic X1500 + Panasonic X920 + GoPro Hero 11 Black

Desktop: MB: MSI B450M MORTAR TITANIUM, CPU: AMD Ryzen 5700X, RAM: G'Skill 16 GB DDR4@3200, Graphics card: MSI RX6600 8GB, SSD: Samsung 970 Evo+ 500GB (NVMe, OS), Samsung 870 Evo 1TB, HDD WD 4TB, HDD Toshiba 4TB, OS: Windows 10 Pro 22H2

NLE: Vegas Pro [Edit] 11, 12, 13, 15, 17, 18, 19, 22

Author of FFMetrics and FFBitrateViewer

AveSatanas wrote on 2/3/2020, 6:30 PM

im going to try 15mbps vocoder and x264, slow with average bitrate. that 1080p vid was 18mbps hardware encoded and the 1440p is 30mbps hardware encoded

Musicvid wrote on 2/3/2020, 6:34 PM

Eagerly awaiting your quantified tests (meaning with numbers), published testing protocol, and data comparisons. You've introduced way too many untrapped variables to make visual comparisons meaningful.

Here's an example of the scientific method (that happens to disprove another popular and enduring street myth).

https://www.vegascreativesoftware.info/us/forum/10-bit-vs-8-bit-grading-the-musical--111748/

 

AveSatanas wrote on 2/3/2020, 6:52 PM

...lads....i can render but my files are not appearing in the destination :O

AveSatanas wrote on 2/3/2020, 6:53 PM

This is a first...

AveSatanas wrote on 2/3/2020, 6:54 PM

Eagerly awaiting your quantified tests (meaning with numbers), published testing protocol, and data comparisons. You've introduced way too many untrapped variables to make visual comparisons meaningful.

Here's an example of the scientific method (that happens to disprove a popular and enduring street myth).

https://www.vegascreativesoftware.info/us/forum/10-bit-vs-8-bit-grading-the-musical--111748

 

I dont think id understand how to do that? lol sorry friend

AveSatanas wrote on 2/3/2020, 6:59 PM

seems like vokouder doesnt like rendering in Average bit rate

Musicvid wrote on 2/3/2020, 9:05 PM

I dont think id understand how to do that? lol sorry friend

Well, you could start by using the same file for your tests.

And uploading your source files to Drive or Dropbox so an objective test is even possible.

The rest can be learned. Your pixelation in the second (larger) 1080p file is from artifacts caused by excessive bitrate somewhere in your encoding chain, that's all.

Expections can be ruinous. That's why we don't believe subjective tests.

AveSatanas wrote on 2/3/2020, 9:31 PM

hmmmm ill have to research on excessive bitrate cheers @Musicvid

AveSatanas wrote on 2/3/2020, 9:33 PM

Interesting, check what this guy says

 

Hey Ralph,

My personal preference on my PC is to avoid H264 at all costs, particularly if it's done through Apple QT codec - however, sometimes that can't be helped.

Nevertheless, as someone suggested on the Adobe forum, it may be that your footage already have artifacts in it, that is being exposed through re-encoding it.

My initial suggestion would be to do an uncompressed export and compare that with the original. If it doesn't happen there, then it is your H264 settings that you need to work on.

The other thing to try is to turn off
"Render at maximum depth"
&
"Use Maximum Render Quality"
as you are already working compressed and you are not currently adding any graphics or fx to the video, I see no advantage in having those boxes ticked - if anything, it by trying to make it too beautiful, this may be the reason for the artifacts coming out.

Although it doesn't give you 60fps, have you run a test through the "H.264 Blu-ray" codec also included with PPro?

It is all about finding an output that gives you a minimum of artifacts, whilst being close to your desired result. Do keep mind though out that you are re-encoding already compressed footage, and that it will never improve - unless you use specific software to remove the "noise" with, but even that is not always perfect depending on your final method of distribution.

All the Best
Mads

@madsvid, London, UK
Check out my other hangouts:
Twitter: @madsvid
http://mads-thinkingoutloud.blogspot.co.uk

fifonik wrote on 2/3/2020, 10:08 PM

Again?

Just try to render 30 secs of your current project in source media resolution using Voukoder/CPU with the single CRF=18 option (do not touch any other options in Voukoder).

Then try all other methods that are in your mind or you've found in internet.

Name them 1, 2, 3, etc (with no information about processing/encoding), upload all these videos to YT and ask someone to rank them (or provide links here).

Last changed by fifonik on 2/3/2020, 10:14 PM, changed a total of 2 times.

Camcorder: Panasonic X1500 + Panasonic X920 + GoPro Hero 11 Black

Desktop: MB: MSI B450M MORTAR TITANIUM, CPU: AMD Ryzen 5700X, RAM: G'Skill 16 GB DDR4@3200, Graphics card: MSI RX6600 8GB, SSD: Samsung 970 Evo+ 500GB (NVMe, OS), Samsung 870 Evo 1TB, HDD WD 4TB, HDD Toshiba 4TB, OS: Windows 10 Pro 22H2

NLE: Vegas Pro [Edit] 11, 12, 13, 15, 17, 18, 19, 22

Author of FFMetrics and FFBitrateViewer

Musicvid wrote on 2/3/2020, 10:31 PM

Upload your original to Drive or Dropbox.

Link your YouTube files here.

I wouldn't be able to help you otherwise.

Internet quotations are not useful, sorry.

AveSatanas wrote on 2/4/2020, 1:31 AM

cheers lads will keep ypu both posted