For YT video record at 60fps or 30fps?

[Deleted]

Someone mentioned on thread earlier that to maintain best quality its better to work with same type files from recording--> editing-->render.
If I am making youtube videos, should I be getting better quality from recording at 60fps or 30fps?

[Butler to Baby Sloths]

https://support.google.com/youtube/answer/1722171?hl=en

Bookmark that reference document

[Ghâzh]

30fps will be enough unless you wanna use slow motion effects in the editing. If you do you should record as high fps as you possibly can to make the slowed video appear as smooth as possible. Otherwise it won't matter too much even if you recorded at 60 fps youtube will play it at 30 fps although there's a small advantage since blending down from 60 to 30 should produce "sharper" frames. In reality the difference isn't really anything spectacular.

[fixx]

there's a small advantage since blending down from 60 to 30 should produce "sharper" frames

Other way around. Blending down from 60 to 30 fps will make smooth motion blur effect that people are used to seeing in TV. If you just drop half of the frames from 60 to 30 it will look super-sharp. Both have advantages and disadvantages, and will produce distinctly different "look" for the video. Kinda like "normal" vs HFR version of The Hobbit.

[Ghâzh]

Other way around. Blending down from 60 to 30 fps will make smooth motion blur effect that people are used to seeing in TV. If you just drop half of the frames from 60 to 30 it will look super-sharp. Both have advantages and disadvantages, and will produce distinctly different "look" for the video. Kinda like "normal" vs HFR version of The Hobbit.

Hmm yeah I suppose I was just confusing the two. To rephrase myself; Youtube does infact frame dropping, not blending, when you upload at 60 fps. The effect is still the same though like you said it will make it look sharper.

[Ghâzh]

If you're going to drop frames rather than blending then there's no noticeable difference between capturing at 30 and 60 or 120 so you might as well just do the lower data rate version.

They are all gonna obviously be the same if you look at a still picture. In motion however there's a small albeit noticeable difference in smoothness between recording at 30 versus 60 and frame dropping.

[Ghâzh]

Can you explain your reasoning please?

If any particular frame considered stand-alone looks the same, then presumably getting 30 frames and considering them stand alone will also show that each of those 30 frames looks the same. How will 1 second of video consisting of 30 frames look 'noticeably smoother' when every frame is shown for the same amount of time and each of those frames contains the same information?

Magic?

Wait, I got this. Maybe it's the youtube script that's intelligently dropping the "bad" frames and leaving the good ones to achieve smoother looking output? Honestly I don't have any more technical explanation, seems to be hard to find more information on how exactly youtube handles the conversion. All I know is that it looks smoother to me when uploaded at 60 fps instead of 30.

[Gaebryel Quintyne]

I think, from all the research I have done, that if you are recording gameplay to upload to YouTube, and your FPS dips below 60 FPS at any time while you are playing, you should record at 30 FPS because if you record at 60 FPS, then the frames that are not at 60 FPS or some sections in the video that are not 60 FPS will be kinda messed up or something... Correct me if I am wrong on this...

[Ghâzh]

I'll post a few videos of the same scene captured at either 30 FPS or 60 FPS dropped to 30 FPS in a playlist. I'll post the "answers" here in an encrypted format so I can't lie once you've seen them. You watch the videos and tell us what you think are the dropped frame versions and which were captured at 30 FPS. Afterwards I'll post the key to decode the file and we can see if you were correct.

If you get 7/10 or more correct--because you can get right 5/10 times--I'll send $10 to a charity of your choice. If not - you send $10 to a charity I pick (or we can do something like blizzard store/curse memberships/etc). Deal?

No deal. I'm aware that it could very well be nothing but a placebo effect, I think it's not but there's always a chance. I'd be more then willing to do the test for the name of science though. Just to get some evidence like you said, one way or another.

[Ghâzh]

-aenean egestas ipsum in laoreet = 30 FPS
-consectetur congue augue = 30 FPS
-nec purus quis velit fringilla = 60 FPS
-apien dui luctus ac faucibus = 30 FPS
-proin at sapien ac ante tincidunt = 60 FPS
-nam in egestas tortor = 60 FPS
-pretium libero at tempor egestas = 30 FPS
-mus proin vestibulum nunc = 60 FPS

The seventh clip appears private. Can't watch it so I didn't rate it's counter part either before I can see that one. I got to say though that there's almost no difference at all between these whatsoever. I might be inclined to believe that the most difference is actually the FPS difference between the scenes from clip to clip (I don't think you maxed 60fps at all times?). Nevertheless goes to prove that if there's a difference it's very minimal. Anyway waiting for the results.

[fixx]

My guess is YT handles B or P frames diffrently in 30 and 60 fps video. Maybe even straight passthrough in 30fps and full recompression in 60fps which adds some unnatural movement in game videos.

Or alternatively YT uses a compression optimized for interlaced video (assuming the source material is i60) which leaves shittier quality for progressive frames. It would be dumb, but wouldn't be the first time YT's compressor is shit. They had the infamous /8 height issue few years ago where 1920x1080 looked shit compared to 1920x1072.

[wMw]

Does anyone know best params for Frame-Blending? Amount of Particles and Angle? I've seen some super smooth gameplays of WoW. When I convert 60 FPS to 30 FPS with motion blur, it looks better then Frame-Dropping, but not that great as I saw other YouTubers make.

I'm using Final Cut Pro and Motion 5. I use 8-16 particles and 360-460% angle

[apepi]

An update to post the "answers" to the test. Your guesses:





The "key" needed to decode the block above (after base64 decoding it) is: MGFHsrI1S~ywp9UONTPFCib<IN[l4\2v
You can do it yourself using a tool like http://www.tools4noobs.com/online_tools/decrypt/

-----

So based on our very small sample, we had an unexpected result. You "guessed wrong" every time, however, the fact that you were consistently wrong means its very likely you can spot a difference between 30 and 60 FPS, you just interpreted the difference you detected 'incorrectly'. Presumably you saw the 30 FPS video as somehow better (more detailed, less choppy, less noise) and concluded it was the 60 FPS video, or alternatively you spotted some flaw introduced by frame-dropping and concluded those must be the lower frame-rate captures.

I didn't think we'd see any pattern, you thought you'd prefer 60 FPS and could identify it. There's a small chance you'd just luck out like that, but we should probably take a moment and consider that maybe 30 FPS captures really do look better if you send them to YouTube. Maybe dropping frames causes YouTube video to look worse. It might be worth some further investigation just to see if you were lucky (or unlucky, depending on how you want to call it) or if there really is something going on here.

I checked and double-checked the files before uploading, and I can see the original file names on youtube in the video manager. For example:



So I'm reasonably sure I recorded all the titles/file-names right.

Possible explanations:

• You're used to seeing video captured and uploaded at 30 FPS. It doesn't actually look better, but it's more consistent with what you expect. While you can detect a difference, you prefer video that looks like the stuff you see most often, not what actually looks smoothest (the same sort of complaints made about 48 FPS films)
• You were lucky. Not insanely lucky, but a <10% chance of getting it by sheer luck is still within the realm of possibility.
• I screwed up the captures/encoding/etc. some how. I'm not sure how that could be, and I'd be willing to document my procedure if you think you could spot something - but I can't think of any shortcuts I took that could introduce error.
• Maybe the way frame-dropping happens actually introduces some sort of weird temporal artifact. Maybe they've optimized their system for high-speed cameras and those cause strange issues with computer graphics. I suppose we could correct for that by doing the frame dropping on my end with premiere.
• The frame dropping method introduces a bit of jitter, or perhaps capturing gams running at very high frame rates (>200 FPS in some cases) at 60 FPS causes artifacts when you drop frames. I can kind of imagine a way that could happen but I can't think of a good way to test it on youtube without doing a lot of work.

In any case, I'm reasonably convinced you can detect the difference between a 30 and 60 FPS file sent to youtube. The problem we have to sort out is that you cited the 30 FPS versions as though they were the 60 FPS versions; you noticed the difference and thought lower frame-rate captures looked smoother. That's surprising to say the least.

Well I heard many people not liking the fast frames on the new hobbit movies, maybe that is like this.