Thread: VirtualDub2
View Single Post
Old 18th June 2018, 22:58   #525  |  Link
Revan654
Registered User
 
Revan654's Avatar
 
Join Date: May 2004
Posts: 324
Quote:
Originally Posted by LigH View Post
That's the important difference.

There are many HEVC / H.265 encoders out there. But not all of them can be implemented into VirtualDub2, based on libraries available to ffmpeg (e.g. most of them are not available as source code): x265 is one of them, but it will not use any GPU encoding. GPU encoders like NVEnc may be available too, where the hardware supports it, but it won't compete with the efficiency (quality retention per bitrate) of x265.

HEVC encoding in a GPU is way too limiting: less available RAM to store many frames to compare with each other to find similarities; slow uploading of the frames from system RAM into video RAM; limited set of instructions in the GPU, optimized rather for 3D graphics than for video compression ... using SSE, AVX and other modern SIMD instruction sets, and having possibly more than 8 GB RAM available with rather fast access, is much more interesting for video compression with good quality preservation, which requires a higher complexity in the encoding process.
There is x265VFW, However it's limited when it comes to pixel type.

Maybe can just use the GPU as acceleration option. Since x265 can be taxing on a system.

Or an option to use external encoder while in capture mode. Right now all the codecs are pretty much useless to me since I need to add additional flags to the codecs and there no option for that (Except for x264).

Quote:
Originally Posted by davidhorman View Post
Are we talking about x265 as in the specific encoder here, or using it to refer to HEVC/H.265? Just trying to work out what everyone's saying...
Just a reference to HEVC.
Revan654 is offline   Reply With Quote