If my video encoder will only let me save up to 60 Mbps (megabits per second) in 1080p (1920x1080), but up to 200 Mbps in 4K (3840x2160 or 4096x2304), then is it kosher to save animations of detailed 1080p images to 4K to maintain their quality? Because they’re going to be damaged in the encoding process no matter what anyway, so I figure save to the best quality possible. Or is that cheating? Will people be angry downloading 4K on YouTube and say this is just upscaled…? (I figure upload the highest to youtube and let it do the rest) ?
Thanks for your help as always!
60 Mbps and you are asking about quality ?
60 Mbps 1080p is quite high quality 1080p… presumably… I assume your encoder doesn’t just makes a high bit rate junk quality… that could happen too
What I think you have is two settings which is suggesting " 60 Mbps is a great quality for 1080 and 200 Mbps is a high quality for 4K".
I think these are just default profiles, and you can set lower quality , but thats up to you to learn about. trial and error ,etc…
and yeah it is cheating to post 1080p as 4k …but its only a minor upscale, and your encoder probably does it with proper smoothing… still it limits your audience as they start thinking they will have the issue … or perhaps youtube audience really doesn’t matter , you can say HD/4K … to advise its suitable for either…
Ah, but if your encoder does a high quality upscale, then thats a win for the viewer who has 4k hardware but not so good upscaler built into it. (and fast internet or is downloading it ?? depends on your expectations of the viewer/customer, I guess… )
So the upscale from 1080p to 4k is no problem to anyone, but might cost you time, if you have to wait 4 more hours or something… why bother ?
If the original image doesn’t have that 4K quality, you’re not preserving anything; all you’re doing is pad up the size and slow down every system involved. Does the image actually have detail that will get lost in 1080p?
Which video encoder are you using?
If you don’t have actual 4k content then upscaling is of minimal benefit, especially for web distribution. It takes more time and space to encode to 4k, then 4x the time to upload.
You will save yourself a lot of time by encoding to H.264 1080p and uploading that.
If you have any doubts take a short test clip encode it both ways, examine before upload, then examine after upload. If you can see a slight difference after encode but you can’t see that difference after uploading, then there’s no benefit to wasting the extra space, encode time and upload bandwidth.
YouTube will reduce your 1080p file down to around 12Mbps, maximum. So I do not see the point.
I also doubt that a 200Mbps upscaled 4k video will be better quality than a 60Mbps 1080p file.
Since you have a high quality version of the animation, it’d probably make the most sense to encode it at around what YouTube uses. Here are YouTube’s recommendations (click on Bitrate.)
You might be saving a lower quality image by going with 4k, IIRC 4k is about 4x the data of 1080p, so by upscaling to 4k you have to save 4x the amount of data in your case to display the exact same thing, and at 200 Mbps that would be equal to 50 Mbps at 1080p.
Add to that if you upscale the video you need some of that bitrate to store details that are not really there, but a result of the upscaling.
Now with compression I don’t know how it would do this, but your final product at 4K I believe would be of lower quality due to higher compression to fit in that bitrate and saving upscaling data that does nothing for image quality.
Thank you for the info but I don’t think you guys quite get animation. Say I have 1080p files of 3 Mbytes (or 24 Mbits) each [1 Mbyte is 8 Mbits]
then 1 second at 30 fps = 720 Mbits, needing 720 Mbps (Mbits/s) for lossless quality.
60 Mbps (the max PowerDirector14 lets me do in any encoding, in H264 AVC)
already cuts the quality to EIGHT percent (8.33%)
It discards over 91% of my render! (When you render fractals, every single pixel takes calculation time, pixel for pixel. So it’s just gruesome to render for months and have to discard almost all of that data.)
12 Mbps (YouTube’s higher-level bitrate upload recommendation) is only 1/5 of THAT, meaning under 1.7% of the quality of the frames. (Imagine if you got paid for 1.7% of the hours you worked! That’s what my computer gets for its work if I upload at 12 Mbps.)
If I can upload 200 Mbps, then at least I can get 200/720 or 28% of my quality uploaded, then YouTube has the highest data quality possible to make whatever versions it wants, right?
No, because higher resolutions are associated with higher bitrates. Consider that: 1) I don’t have the option of setting any bitrate I want in PowerDirector14. If I want more than 60 Mbps then I have to go higher than 1080p. 2) YouTube recommends 8-12 Mbps for 1080p, but 53-68 Mbps for 4k 2160p.
Just about any image will lose quality when encoded to lossy (non-lossless) formats. If you take a 40 Mbyte PNG and encode to a 10 Mbyte JPG, you’ve lost potentially 75% of your information (not necessarily, the format might have worked well).
Because YouTube allows a higher bitrate for 4k 2160p, up to 68 Mbps (on the chart linked to above), so at the least, YouTube will allow 9.44% quality of my animation (as opposed to 1.7%), assuming they maintain the quality of the upload recommendations.
60 is 30% of 200. That means the player is sending 30% of the data that it would send at 200, right? Isn’t encoding into a format that will transfer over 3x the data going to maintain more quality?
If YouTube likes 8-12 Mbps (1.1% to 1.7% quality) for 1080p, isn’t it going to mangle a 60 Mbps 1080p file?
The big point here is that SIZE seems fused with QUALITY in these systems.
It doesn’t seem possible to specifcy a high-QUALITY but lower-SIZED animation.
This seems to explain to me why there are so many videos labeled “4k” with comments labeled “that’s upscaled!”. Why shouldn’t people “upscale” if it will allow massively more data/quality for the end viewer?
Thanks for all your help!!
If you’re forced to save a lossless file to a lossy format, your best bet is to save to the highest possible resolution available. A 160x90 pixel PNG will lose quality to 160x90 JPG, but if you save to a 16-trillion by 9-trillion JPG, you have a better chance at preserving all the quality.
The 1080p may be marginally better. Assuming 1080p60 vs 4k60, and starting with lossless uncompressed RGB:
(192810803 bytes)(60 Hz) in bytes/s = 2985.984 Megabits/s
(384021603 bytes)(60 Hz) in bytes/s = 11943.936 Megabits/s
60/2985.984 = 0.02
200/11943.936=0.0167
So the 60Mbps at 1080p actually has a slightly better compression ratio, and should retain more of the original information. (This still holds for 30FPS, just multiply by two).
This is because you’re upscaling, of course, if your video were natively in 4k downsampling wouldn’t make sense because the information you lose doing so would outweigh the worse compression ratio.
I’ll admit I’m not 100% this analysis is 100% sound, but I don’t think upscaling is worth it. Youtube compression is known to be awful regardless.
Edit: You can’t just use the original bitsize, because after upscaling and reencoding you don’t have any control over which pixels get preserved; a lot of repeated pixels will be encoded, it’s not a 1:1 conversion. It also depends on whether your upscaling method does any nearest neighbor smoothing or just repeats pixels or whatever else. Depending what it does, upscaling and re-downsampling may actually look worse.
Not edit: Note that my ratios aren’t actual compression ratios, the ACTUAL lossy compression ratios are going to preserve far more than 2%/1.6% of the original information, but as a naive heuristic I think my calculation is okay.
If you care about quality, don’t use YouTube is the answer. Vimeo supports uploading as ProRes files. A ProRes file uploaded to 1080p will look great. Or you can upload a high bit rate h264, in either way the result will be far superior to YouTube and you can still monetise content on Vimeo if that’s what you are doing.
What do you mean by “cheating”? There are no Codec Police who will arrest you for using the wrong settings. In the real world, there’s only how good it looks (how good it actually looks, not just what percentage the “quality” slider tells you), and how big the file is. Somewhere, there’s a happy medium where the combination of those two considerations is optimal. Do that.
And you sort of crippled the discussion by referring to the video as “animations”. That usually means “cartoons”, which are generally low-detail and thus very robust against information loss (a low-quality cartoon will look nearly as good as a high-quality one). Fractals, however, are the very opposite of low-detail.