Topaz Video AI v3.4.4

When will you fix warps in Apollo fast preset? It’s been about 5-6 updates already. You haven’t even put it in known Issues :slight_smile:

Artemis always producing many white dots.

Does anyone have a faster way to export?
A 1-hour video must be exported over 12 hours (4K resolution Proteus).

Split the video in two half and process them at the same time via a second instance:

Which can be faster but isn’t in all cases. RAM can be a limiting factor here, also the used model is important. Of course the used architecture is playing a major role, too. The more GPU cores you have the higher the chance Topaz will not fully use them with one single encode.

While I do get a slight speed boost running two instances of e.g. Artemis this isn’t really the case with Nyx. And e.g. three instances of Nix 4K are a little too much here for my 64GB of RAM, so it’s using swap thus no speed gain (and making the whole machine extremely sluggish) even with RAM set to 10% , running at 0.4 / 0.4 /0.5 fps, two Nyx 4k processes are running at 0.6/0.7 fps while I can use the machine for other “normal” tasks (office, web-browsing) without noticing the heavy load on the GPU.

1 Like

This release v3.4.4 defaults the Bitrate > Constant to 1Mb/s, all previous versions the default was 16Mb/s, What exactly should it be? there should be an auto and manual setting for Bitrate with auto choosing a recomended value?

3.4.4 no Theia no Artemis

Seriously, no Artemis? Why? Why degrade the software, why remove useful features? :frowning: :frowning: :frowning:

2 Likes

I see Theia & Artemis in 3.4.4.

Me too.
I’m actually coding with Artemis in 3.4.4 right now, as I type this message.

Maybe bug:

  1. Changed Iris to Relative to Auto. Moved all sliders to 100 (except last 2).
  2. Preview video.
  3. Changed Iris to Auto.
  4. Preview video.
    On step 4 a new preview is generated but it looks same as on Relative to Auto with maxed params.
    Auto does not work or relative does not realy switch to Auto.