I provided a source above, though it was sort of 'hidden' in an edit, sorry.
This was recorded using Dxtory and 'UtVideo YUV422 BT.709 VCM' 1080p 60fps. Outputing as AVI to a 7200RPM drive separate from where the game was installed.Edit: Here's the link to a source video as requested.
I did forget to link the 'cooked' YouTube version though.
That was run through Vegas and Handbreak with:
Render Settings (Video for Windows .avi)
Re sampling disabled.
For this example the quality came out more or less what I would want to see. But then the next video I upload will look substantially worse. Now I can see where you would say that youtube's handled the video differently, leading to a different result. But then it just makes me ask, how do other youtube content creators maintain constant quality across every thing they upload no matter what the content of the video is?
+ Reply to Thread
Results 31 to 47 of 47
Last edited by midiout; 29th Jun 2015 at 20:14.
Nope the cooked version is the one you upload to YouTube, we want 10 seconds of that uploaded to this site obviously not a copy from YouTube.
I provided a link to the source recording video in my post, uploaded to your site. Though it's only 7 seconds, I was told 'a few seconds long' before I uploaded it.
Source recording uploaded to this website. No compression or rendering: http://files.videohelp.com/u/243643/Dxtory%20UtVideo%2060fps%201080p.avi
That same footage run through Vegas and Handbreak, uploaded to this website.:
The Vegas/Handbreak version, uploaded to youtube:
Or am I misunderstanding your request?
For this example the quality came out more or less what I would want to see (although still not as good as my previous videos). But then the next video I upload will look substantially worse. Now I can see where you would say that youtube's handled eachvideo differently based on their content, leading to a different result. But then it just makes me ask, how do other youtube content creators maintain constant quality across every thing they upload no matter what the content of the video is?
Last edited by midiout; 29th Jun 2015 at 20:43.
The only consistent thing across the board is consistently is poor quality.
The only videos that look fairly good consistently are those with "ideal" content, like "talking heads" , low motion, studio lighting, those sorts of videos.
All YT versions fail exactly where you would expect them to fail - exactly in those sections that stress the compression from lots of activity, motion etc... This is a basic fundamental principal like gravity. You can't get away from it. Certainly all gameplay videos fall into that category. I see the exact same problems everywhere . Or show me some examples of what you mean by "consistent" .
The only thing I've seen in this entire thread that is inconsistent and unexpected, is your test with the same video 2 weeks ago. And I'm not entirely convinced unless you proove that it was the same video, same settings etc...
I just did it realizing that it could be useful for some gopro footage, shaky, in reality similar to those games. When youtube encoding might just brake apart. So I wanted to see that.
@newpball - no , post 33 explains it
The AVI is the original
"Dxtory UtVideo 60fps 1080p - Vegas - Handbreak.mp4" is the exported version from Vegas, encoded with Handbrake => that is the version uploaded to YT
I had a closer look and everything is consistent and expected with the other videos - including the videos of other users in the first post. They fall apart exactly where you expect them to, they look better exactly where you expect them to. The differences like edits/ blurred backgrounds in various sections all help with compression. Peripheral overlays (the red border) mean less differences between frames, easier to compress. Everything is explainable and consistent behaviour. The motion and content differences easily explains the observed differences between the videos
The only INconsistent thing is the video uploaded 2 weeks ago vs. Today. You would expect that level of quality to be similar (ie. similary poor), even if your uploaded videos were slightly different. That is inconsistent. For some reason YT applied worse encoding settings, yet higher bitrate. The quality is definitely worse. That one has me stumped.
Looking at the AVI you uploaded, there are some small optimization things you can do , like playing with higher AA settings ideally (or post AA) - recall that sharpness will "tax" compression . Things like jaggy edges will eat up bitrate. Also, things like blurring - adding motion blur . Basically reducing the fine details so YT has an easier time compressing it. But in the end, all those optimizations will have only a tiny effect
Let me try again: what is the bloody point in trying to understand what goes wrong on YouTube if the OP does not want to bother downloading a few frames of the file he uploads to YouTube.
Am I just dumb or what?
The OP wouldn't need to DOWNload anything, because everything should be local already for him. Unless you meant the version YT serves. Then he could just download it from YT.
Anti-SOEr's must talk a different language
A few frames of the file I upload to YouTube (Vegas/Handbreak):
The YouTube Result:
This is the only way I can imagine trying to further prove that the June 15th video is the 'same' as the one I rendered and uploaded yesterday.
The project was last saved June 13th.
I open it, don't change any settings and manually select a loop region (in this case the exact loop region as my original upload).
I know that the original video was a wmv.
The only wmv settings I've ever used are saved and favorited here, and I render using that same preset.
From there I render it, the same way, from the same file, last saved June 13th and as we know the outcome is poor.
In terms of the quality across my videos coming out as expected based on the cirumstances, I can't disagree with that any stronger.
Here for example: https://www.youtube.com/watch?v=nK9Im1eP3DI This is TotalBiscuit, one of my most respected youtubers. His videos are always top quality, no matter what's on screen. Sure you say YouTube quality is bad and that's expected, but there's different degree's of poor quality. Sure it's compressed, not true 1080p, and not lossless, but that video looks fantastic compared to anything I put out. I'm not comparing lossless vs youtube, I'm compairing youtube vs youtube. He can be standing still, he can be in a room with tons of bright colors with lots of motion it DOES NOT MATTER what is on his screen, the length of the video, the ammount of breaks and downtime, nothing matters. It ALWAYS looks better than what I'm uploading. I highly doubt he has to think to himself,'hmm this video doesn't have 10% motionless scenes, my entire video is going to go to crap'. He publicly talks about how much work and experimentation he goes through in order to achieve the quality he deems fit for his channel, and it's worked. Example 1. Example 2. For example one if I download that game and get on that same map, follow his exact path/movement, record it and upload. It will be look as good as his.
His content looks consistent no matter what, yet mine does not. This is what I'm trying to achieve. Saying, 'It's not possible because youtube' just isn't enough for me because I could provide countless examples of videos in the same game looking much better than mine, no matter what the content is (motion, colors, lighting, etc etc...). In which case I have to change something, because I must be doing something wrong, but I don't know what that is. I don't have any knowledge on this subject, I'm just trying to pick things up as I go. I was hoping with making this thread that someone would tell me 'open vegas, change these settings, try enabling this etc and see if that helps', not, 'YouTube ruins videos, deal with it'. It's quite clear to me that there's more going on here than that, and I really don't think that I'm being delusional when thinking that based on all of the good quality examples I've provided.
Last edited by midiout; 30th Jun 2015 at 10:04.
You're joking right ? His videos look typical for youtube. They fail exactly where you you expect them to fail, they look decent exactly where you expect them to look decent - So yes, they are consistently "poor", which is "normal" for youtube. Nothing you have shown me suggests otherwise.
I think you've fallen into this "groupie" mindset, where one guy says his videos look good, and everyone agrees. Why don't you objectively look at the videos - they are terrible in quality, but that's "typical" for youtube. It's not his fault, it's YT. Download his video and examine it locally. It breaks apart on high motion scenes - blocky messes. Static setions look fine. It's all content related
I get it, you're saying the videos you upload look "more terrible" than his . You're comparing youtube to youtube - I understand what you're saying. It might be a bit worse, because of your gameplay settings, AA settings , maybe a bit of optimization, slightly different motion... etc... but it won't be that much worse. Those optimization techniques discussed earlier do end up making minor differences in the end. They deal with making video easier to compress for YT. There are many threads and examples on this. They do make a difference. And if those are the minor optimzation differences you're referring to, you have your answer.
If , however, you're saying you get something like your "today" video, despite similiar content (and that is the key, the videos you posted were very different content wise in terms of compression) then something is definitely wrong. That is more than a minor difference. That is not typical. You seem keen not to pursue that any farther, but really that's the only interesting example you posted, because it's unexpected. Everything else is expected and consistent behaviour.
I agree that his video's look typical for youtube, they're at the standard I would expect. I'm saying mine are not and that I don't know what I need to do in order to correct that.
It might be a bit worse, because of your gameplay settings, AA settings , maybe a bit of optimization, slightly different motion... etc... but it won't be that much worse. Those optimization techniques discussed earlier do end up making minor differences in the end.
Now as far as the today vs last week video. I'm genuinely really curious why that's happening. But it seems to be making an impasse for me to improve my quality (though I'm sure it's fixable, no one here seems to be able to tell me how)So I'm trying to 'start over' and just assume I've never uploaded a video. Say for example I linked you the 'today' version, and it was my first ever upload. Based on the responses I've been getting in this thread, I would be told 'yeah that's youtube handling your video poorly, that's what youtube does.' Where as because it's the second upload I can prove that it's possible to look better. So say for example I started this thread showing my 'today' upload and asked, how can I improve my quality. What would your suggestions be? I feel like I haven't gotten one real response telling me how to optimize my video, I'm just getting blanket statements telling me that I need to optimize it. I can't change the type/frequency of the motion if that's a big problem (it seems to impact my videos but not total biscuits, so what else is there ?
Last edited by midiout; 30th Jun 2015 at 11:09.
These optimization techiques ALL deal with making it easier for YT to compress. I have to re-iterate it's not going to be a huge difference, but they do make minor improvements in the end . So don't expect miracles. I briefly mentioned a few already, and they are discussed in more detail with examples in other threads
First you need to improve your gameplay settings. The antialiasing settings expecially. Those sharp jaggy edges ("stair stepping") have a negative impact on compression efficiency. If you have a weak GPU or slow hardware, you can do some if it afterwards in software, but the AA quality is always a lot worse than if you did it correctly in-game
Low passing and reducing fine details spatially is basically what AA is doing, but you want to reduce it farther. At the bitrates YT uses, you're not going to be preserving wall textures, those sorts of things anyways. So you want to blur them out. Recall I mentioned the sharper, the more detail, the harder it is on compression? You want to selectively blur the things that are unimportant to you , rather than let YT wreck everything - that's the general idea
Using a temporal smoother or denoiser , even on low settings will improve compression efficiency. This makes it easier on youtube because the differences between frames are smaller. Recall that's how compression works - but storing the differences between frames. By reducing the differences between frames, you are making it easier to compression which will result in a higher quality at a given low bitate typical of what youtube allocates
So basically you want to "degrade" your video before you upload it to youtube by applying filters. A lossless high quality video actually ends up looking worse on youtube because it struggles to do preserve everything and fails miserably at everything. All these techniques attempt to make the job easier for YT
The other common technique is stablization. But - it usually doesn't apply to gameplay videos, unless you want to change your playing style, or if altering the field of view in post is acceptable. Usually game players don't want to do any of that
But I'd be more concerned about why your "Today" video looked like that. Even if you applied all those techniques, it wouldn't make up for the difference observed there. Something very weird happened there
The game it's self has very limited options that have an impact on the videos, AA for example there's ON/OFF and changing them have no real impact. So what optimization should be I doing out of game (video card/vegas/handbreak/dxtory settings etc)?
If your game or GPU settings don't allow for higher AA settings, your only option is to apply them afterwards . It not ideal because strong AA settings applied afterwards will "hurt" fine details like small font text overlays , HUD text etc.. often making them illegible. However, if applied in-game, they will still be crisp, and only the unwanted jaggy textures will be affected. Ideally you only want to "degrade" selectively - you want to keep those HUD and text crisp (so you can read legibly player names , text popups, etc...) . There are ways around it - you can composite different layers and use masks, but its' more involved and more advanced. You really want to look for AA settings in game or maybe a GPU setting if you can - or if there is a game mod that allows you to control it - it's worth your time investment looking into it.
All these filters and techniques can be found in free software like avisynth, virtualdub . For example avisynth has over a dozen antialiasing filters , hundreds of variations on denoising/ smoothers - all free. There are commercial plugins available too, e.g you can find various denoisers, smoothers etc.. some of them work directly in vegas if that's where you want to apply them
Well I don't seem to be having any luck no matter what I do, my videos look like a distorted mess in comparison to everyone else. Would anyone know some other forums related to this subject I could post on? I'm pretty curious about my massive quality downgrade over the last week and I haven't seem to have gotten any answers here. Or maybe there's some more information I can provide, or tests I can do to get some answers? I've been banging my head against a wall for the week trying to fix this with no luck and I'm beginning to go crazy.