You are not logged in. Please login or register.
SmoothVideo Project → Posts by Mystery
really? That would be possible to download H264+Vorbis, but then the output will have to be a MKV container instead of MP4
Yes, there are non-DASH MP4 files. What confuses me, however, is that I saw a non-DASH MP4 that had a higher audio bitrate than the DASH audio.
OK. I'm looking at a file, and the 360p files, for both WebM and MP4, contain both the audio and video. So I should just ignore both and discard MP4 and WebM files that contain both the audio and video?
huh... how can I know whether it is VP8 or VP9? And how can I know the 360p really is 240p before downloading the file? http://www.clipconverter.cc/ only offers the 240p option so there must be a way.
I have re-deployed the project and the source code on GitHub
There's one weird issue with video downloads that I don't know how to handle. A few videos report having a 360p version available, but when I download it I get 240p. The YouTube player only offers 240p, but I don't know how to know whether the 360p is real or not.
That's an easy one: regional settings. I fixed it. Download and install the same file again.
You can try with this version. It has UT Video Codec embedded into it and will download WebM format unless it is over 35% smaller than the MP4 version.
https://www.spiritualselftransformation … 2-beta.exe
There's also a mass video downloader/updater that isn't finished yet.
I have embedded UTVideo into the setup itself. This solves it.
I see. They changed the website. Perhaps the best solution would be to host the latest version on my own server, since they have no official reliable download link.
Alright, I have done some proper test to compare MP4 and WebM based on their bitrate difference. Here's a list of videos with the file size difference of the video stream (ignoring the audio).
30%: HAVASI - Spring Wind (https://www.youtube.com/watch?v=rwMYcyo0slc)
31% Girls' Generation - Love & Girls (https://www.youtube.com/watch?v=6ex1YkDXelo)
34% Ellie Goulding - Burn (https://www.youtube.com/watch?v=CGyEd0aKWZE)
36% Alizée - J'ai pas vingt ans (https://www.youtube.com/watch?v=eBUEyHiWMUw)
38% Alizée - À contre courant (https://www.youtube.com/watch?v=EyS9HeCgJL8)
40% Donatan & Cleo - We Are Slavic (https://www.youtube.com/watch?v=VJ920cN2HmA)
43% HYUNA - Bubble Pop (https://www.youtube.com/watch?v=bw9CALKOvAI)
The 4 first videos look better in WebM format, while the 4 last videos look better in MP4 format. This means that to get the highest quality, I should take MP4 if WebM is over 35% smaller than MP4 format.
I also realized that they lowered the bitrate of MP4 videos. Some videos I downloaded previously were larger. For example, Love & Girls had 144MB on my hard drive while the YouTube MP4 version is now 95MB. The current WebM version looks better, but the MP4 version I had from before looks even better. All that has to be taken into consideration to decide whether to re-download a video, and which version to download.
Huh? You're the only person reporting having issues with downloading that. If you try again, it still fails?
In any cases, installing that codec manually should work. It's simply used to convert videos faster with the Media Encoder.
OK I've done plenty of comparison with many videos of various resolutions. For videos in 240p, 360p or 480p, I haven't yet found a single video that looks better in MP4 than in WebM. For 1080p videos, however, all those I compared looked better in MP4 format. As for 720p videos, some looked better in MP4, some better in WebM, depending on the WebM bitrate.
So... some say WebM discards twice more data, some say WebM stores twice more data per byte. If the WebM version has half the bitrate of the MP4 version, then it nullifies any benefit. Perhaps the best solution would be to look at the file size of both, and select WebM only if it is no more than 30% smaller than the MP4 version.
From what I can observe, WebM thinks in terms of vectors while MP4 thinks in terms of pixels. For low-quality videos, the shapes get preserved with WebM while a lot of the noise gets discarded. With MP4, it just takes more bitrates to store the noise. So, it seems to be giving good results on SD videos from what I observed so far. For 1080p, however, I'm still not convinced, because the bitrate is too low to keep enough details to make it look natural. WebM is supposed to get increased benefits with 2K and 4K videos.
Did you try the Media Encoder within the software? One person told me it wasn't working with his computer (issues with Spanish characters, and perhaps something else). Does it work on your computer?
I plugged the 1080p TV and compared some videos. I'm not convinced with WebM yet. It seems to do better at preserving edges while making images that look more "cartoonish". Furthermore, I had to lower SVP settings just for WebM videos.
For now, I re-downloaded all my FLV into WebM. With some videos taking half the space, the quality is MUCH higher because FLV has always done a terrible job with videos.
As for MP4 vs VP9, you've taken the time to do plenty of comparison yourself? Are you 100% sure that VP9 provides a better quality than MP4 at half the bitrate?
I'm not convinced of it for all videos, and one option could be to look at the file sizes of MP4 and WebM before downloading and chosing based on the file size, but if you can tell me that you've done your homework and can guarantee that VP9 is a better option, then I could simply go for that.
I'm French Canadian.
I'm already using MPC-HC v1.7.9
Lastly, you're falling for the video equivalent of the MHz myth - more bit rate is only higher quality if the format is the same. Allow me to link to a post I made on Doom9 showing direct comparisons between the formats:
http://forum.doom9.org/showthread.php?p … ost1719952
2kbps VP9 for sure is much higher quality than 1kbps VP9.
Whether 1kbps VP9 would be higher quality than 2kbps MP4, however, would need to be proved. The VP9 seems to look a bit more like plastic to me.
Overall, MP4 looks more grainy when reducing bitrate, while VP9 maintains the shapes but looks more like plastic. It's probably working with vectors of shapes.
I'm already using MPC-HC v1.7.9
OK here's something interesting. I started having high CPU usage problems after uninstalling K-Lite Codec Pack and trying to run without it.
All my medias still play fine. HOWEVER, CPU usage for playing videos is higher. After re-installing it, CPU usage went down.
What could be causing this performance difference?
I got a HD WebM file that is playing with 50% CPU usage WITHOUT SVP!!
I'm not sure WebM is any better than H264 when it comes to playing it locally with SVP. Unless the YouTube source is 60fps.
Edit: Woah... now after pausing and resuming, it plays at 20-30% *without SVP*, so there's definitely something else going on in terms of high CPU usage.
Another video, 1080p 24fps, was also lagging. I tried lowering all the settings and saw little improvement, until I lowered SVP Shader from Complicated to Standard and then the CPU usage when WAY down. I could raise all other non-expensive settings with no issue. But then, when setting SVP Shader to Complicated with those higher settings, it plays fine!
It's really strange. It was playing at 100% CPU usage, then at 50% CPU usage, then after pausing and resuming, at 70% CPU usage, and now it's back at a stable 50%...
I just had the issue again. I was playing a 1080p 30fps video and it was lagging with CPU topping at 100%. I considerably lowered the settings so that the CPU usage would be 80-90%, but it still played with lots of dropped frames. I de-activated and re-activated OpenCL. I tried shifting to the old core libraries to see if it would do better, but it just made the CPU usage even higher. Then after reverting back to the latest core libraries, the same video plays fluidly with those same settings with ~50%-70 CPU usage.
The profile also keeps switching while playing this video
1080 47 24. Duration: 795 ms
22:17:43.426; ===== Playback
30.363 * (47 : 24) = 59.461 fps
Auto crop: [0:0:0:0]
22:17:44.595; FalseVFR ffdShow-framerate: -Dwn-> 29.968 fps
22:17:44.595; express clear interval
22:17:44.595; stabilized ffdShow-framerate: 29.968 fps. SVP Index: 0.29/0.79
22:17:44.597; SVPMgr: main "C:\ProgramData\SVP 3.1\" 1920 1080 29.968 1920 1080 2 1. Duration: 782 ms
22:17:45.379; ===== Playback
29.968 * (2 : 1) = 59.936 fps
Auto crop: [0:0:0:0]
22:17:53.948; ===== rewind forward 00:02:31 / 503194:33:01
22:17:57.911; FalseVFR ffdShow-framerate: -Up-> 34.04 fps
22:17:57.912; this is changing #2. ffdShow-framerate: 33.081
22:17:58.845; this is changing #3. ffdShow-framerate: 31.781
22:17:59.789; this is changing #4. ffdShow-framerate: 31.177
22:17:59.789; express clear interval
22:17:59.789; stabilized ffdShow-framerate: 31.177 fps. SVP Index: 0.92/1.01
22:17:59.795; SVPMgr: main "C:\ProgramData\SVP 3.1\" 1920 1080 31.177 1920 1080 21 11. Duration: 614 ms
22:18:00.409; ===== Playback
31.177 * (21 : 11) = 59.52 fps
Auto crop: [0:0:0:0]
22:18:01.638; FalseVFR ffdShow-framerate: -Dwn-> 29.968 fps
22:18:01.638; express clear interval
22:18:01.639; stabilized ffdShow-framerate: 29.968 fps. SVP Index: 0.44/0.78
22:18:01.646; SVPMgr: main "C:\ProgramData\SVP 3.1\" 1920 1080 29.968 1920 1080 2 1. Duration: 634 ms
22:18:02.280; ===== Playback
29.968 * (2 : 1) = 59.936 fps
Auto crop: [0:0:0:0]
Nintendo, downloading WebM instead of H264 gives mixed results... not sure what to think about it.
This video is 18.6MB in H264 and 22.9MB in WebM, which means the quality has to be definitely better.
https://www.youtube.com/watch?v=E80TUBjw_Vs
This one, however, is 69.5MB in H264 and 30.3MB in WebM. I understand that it must be compressing better, but I can't imagine having higher quality with half the file size!
https://www.youtube.com/watch?v=NB3NPNM4xgo
It seems that the WebM format is much less stable in terms of bitrate, with some videos being higher and others much lower. The video quality is probably sometimes higher, sometimes lower.
What are your thoughts on this?
I'm not concerned about the file size, but I want the highest quality available.
Nintendo Maniac 64
and I've found a way to make it revert to normal performance levels with 100% success rate.
Simply put, disable GPU acceleration in SVP, and then re-enable it.What about other ways to restart the script?
- change number of threads
- modify any option in the profile
- ... etc ...
Playing with the options didn't seem to solve it for me, although it does allow lowering CPU usage by lowering settings.
SmoothVideo Project → Posts by Mystery
Powered by PunBB, supported by Informer Technologies, Inc.