V3.0 Avisynth Input

Topaz Video AI General
1

I asked this in the long thread about 3.0, but I think it got lost in the noise so I’m asking it in a new post. The lack of frameserver input in 3.0 is a complete dealbreaker for me, and I understand the issue is the distribution of ffmpeg binaries with GPL code enabled. It should, therefore, be possible to build a custom ffmpeg from the redistributable source code here:

However, I’m not clear on what exactly needs to happen here. Diff is telling me that the patch is already applied, so am I correct in saying this source tree is already patched with everything it needs for Topaz compability? If so, I compiled ffmpeg from that tree with --enable-gpl and --enable-avisynth and replaced the binary in the Topaz directory, but it didn’t do anything for avs input. So what actually needs to happen here to enable that?

1 Like
2

I emailed support about AviSynth support and all they said was,

“We are unable to offer AviSynth due to a licensing issue at this time.” :-/

3

Here is the relevant post from the big thread by gregory.maddra to which I’m referring:

Unfortunately we cannot distribute a version of FFmpeg with GPL codecs enabled. We’ve reached out to some of the more popular FFmpeg build scripts to see if they’d be willing to have support for our filters as a solution, but it may be a while before that’s ready. Once they’re done, however, you should be able to replace the bundled copy with one with your desired features.

In the meantime, for anyone feeling adventurous, the redistributables for our filters are available on our GitHub

4

https://community.topazlabs.com/t/video-ai-3-0-0-9-b/35298/58

I tried superpipe today.

“avs2yuv.exe” “script.avs” -csp I420 - | “ffmpeg.exe” “-hide_banner” “-nostdin” “-y” “-f” “yuv4mpegpipe” “-i” “-” “-color_trc” “2” “-colorspace” “2” “-color_primaries” “2” “-filter_complex” “scale=w=1024:h=576,setsar=1,veai_up=model=prob-3:scale=0:w=1920:h=1080:preblur=0:noise=0:details=0:halo=0:blur=0:compression=0:estimate=20:device=0:vram=1:instances=1,scale=w=1920:h=1080:flags=lanczos:threads=0:force_original_aspect_ratio=decrease,pad=1920:1080:-1:-1:color=black” “-strict” “-1” “-pix_fmt” “yuv420p” “-f” “yuv4mpegpipe” “-” | “x265-10b.exe” - --y4m --log-level none --crf 22.0 --sar 1 --aq-mode 2 --deblock -3:-3 --output “output.mkv” 2>nul

Script processing and x265 encoding runs on CPU in parallel (“between frames”) with veai upscale on video card. Compared to pure veai the speed drops by 5 percent at most.
Very satisfied!

1 Like
5

Hi, to answer your questions, yes the source is patched, you will need to add --enable-veai to your configure statement and copy the redistributables to somewhere your OS is happy to picked them up from, and a legal installed copy of 3.0.
The patch is there for when ffmpeg updates but Topaz don’t update their version of ffmpeg (note I’m not speaking on Topaz’s behalf here)

You need the to compile all the ffmpeg dependancies too if you don’t have them to hand.

I’m one of the build script developers (for ffmpeg for macOS M1) the quote replies too, and have a build script for the last beta, but not for the release version yet as my actual job is getting in the way at the moment.

Now that the v 3.0 is officially out, I’m planning on getting a which version up, that builds from the pre-patched source, merging in my work to build AViSynth+ in for mac (as an option as it can’t be built statically, as the libs are loaded at runtime, which is one of the reasons I wrote the build script) at the same time.

I’m then planning to moving onto using the patch so I can keep ffmpeg up to date.

2 Likes
6

dealbraker for me as well. i have TVEAI implemented in my daily workflow with multiple avisynth scripts and I simply cannot change that.

I’ve payed a renewal fee shortly before this change happend and quite frankly I want my money back. I cannot use the product without.

Please bring this feature back. the software is useless without.

7

There’s an easy workaround for this. Instead of reading the Avisynth source with the TVAI build of FFmpeg just read it with a different (Avisynth-enabled) FFmpeg build and then pipe its YUV output to the TVAI FFmpeg instance.

8