avcodec.py issue... #779
Comments
|
Also, since switching that around, I FINALLY got hevc encoding to work via nvenc hardware, but on a rare occasion I will get something like the following... (otherwise, it's working AWESOME!) C:\Users\chads\Desktop\sickbeard_mp4_automator-master>manual.py -a T:\Movies\MOVIENAME.mp4 |
|
Looking back now, looks like maybe the FLAC audio caused an issue w/ eac3? |
|
Changing from eac3 to ac3 corrected the issue. It now works on all files. Re-encoding using this method has given me back 600 Gigs by running 2 threads overnight, in a single night. |
|
Good catch, I just pushed a commit that fixed that encoder being wrong |
|
Also 600 is no joke, how's the quality on all these videos? Any noticeable loss? |
|
Sorry for taking so long, I was away. Anyhow, it was a marked decline in picture quality. I was doing S1 of 24 (the old one) and filesize of each episode went from ~ 4.1g down to ~700m, but it was enough of a drop in IQ to go ahead and stop processing my library, as much as I wanted to keep doing it to save space. The Nvidia encoder is super fast, but the quality just isn't there (without some tweaks of some kind). I've been experimenting w/ StaxRip and encoding using both Intel H.265 and Nvidia H.265. It has a nice graphical user interface to enable various options, but I'm not up to speed on which options do what. But setting both of these to encode the same movie, max settings (a few options have "best" or "quality" avail, so using those) it's clear to me that the Intel H.265 is lightyears beyond nVidia's (mind you, I'm using Kaby Lake here), not in speed, Nvidia encoder clearly has that crown, but in image quality. If this helps: StaxRip Intel H.265 settings.... StaxRip Nvidia H.265 settings... Taking a 30G movie file in h264, Intel H.265 takes it down to about 9g whereas Nvidia H.265 takes it down to 4-5g but the Intel H.265 looks every bit as good as the original (to my eye) and the Nvidia one has a noticeable decline in IQ (but way better than just using ffmpeg via the script, so those options in StaxRip definitely made a difference in IQ). |
|
I would think the Intel vs NVidia shouldn't make a difference in size or quality, but rather you're getting a different quality setting with the lower sized one which is why it looks worse |
|
Yup, that makes sense, I just am not sure what settings are doing what. I would love to find some IQ settings to enable for the Nvidia one, but I'm not familiar with what does what. This is what the -help has on it (sorry for the long post following)... |
|
Almost forgot, you asked for these (assuming you just want the mp4 part here). Also, I was going for speed over everything else, since Plex doesn't need tags or posters or subtitles (using plex channel for those, if needed). I used CRF 14 to try and get a better picture from the Nvidia encodes. Nothing special in mine really.... [MP4] |
|
Just wanted to update the above for you mdhiggins. I've changed my settings, getting excellent results running 2 instances of this at the same time. With these settings, I did Grey's Anatomy (13 seasons) with ~1.4 to 1.5 gig mkv files, and is outputting .4 to .5 gig mp4 files in hevc via nvenc. Quality is not too bad. Not a deal breaker for the savings in space, imo. If I could just get someone to tell me how the preopts and postopts options work (below), I could add some more nvidia custom settings to my options below to make it even better, but as it stands now, I only can pull off 1 option for each. [MP4] |
in the avcodec.py file,
class NVEncH265(H265Codec):
"""
Nvidia H.265/AVC video codec.
"""
codec_name = 'nvenc_h265'
ffmpeg_codec_name = 'nvenc_hevc'
...but shouldn't that be: hevc_nvenc on that last line instead? Also, should read HEVC instea of AVC. So I replaced mine with the following...
class NVEncH265(H265Codec):
"""
Nvidia H.265/HEVC video codec.
"""
codec_name = 'nvenc_h265'
ffmpeg_codec_name = 'hevc_nvenc'
The text was updated successfully, but these errors were encountered: