Patrefans
n00mkrad from patreon
n00mkrad patreon

Flowframes 1.38 Teaser

🕑 Added 2022-07-04 09:07:39 +0000 UTC

Comments

LTSM

Windows 11 should work the same

Lolol Idou

Is this supported for windows 11?, I've got it installed on 10 without proplems before but then I switched to windows 11 on a fresh install and it now it won't download flowframe java python

LTSM

No quality differences, but certain encoding features might be harder to implement. For example, the option to encode two files (one with FPS limit and one without) likely won't work.

g90ak

SUPER EXCITING - no more waiting for 20 min to load large files. Saw that it doesn't need separate encoding anymore as well - COOL. Are there any downsides from a quality perspective from either the processing or encoding changes?

LTSM

Definitely

LTSM

Yes, though it has some minor limitations, FPS limiting does not work yet, deduplication still requires frame extraction, but looping and scene changes work, and all other basic stuff

Milica Zivanovic

So basically the 1.38 version will be using frame serving script? That sounds awesome! Can't wait to try this new one!

baxtyr

Sounds like a massive game changer update! Are we talking this month?

LTSM

No, only a small buffer will be in RAM.

EvilMadness

Sounds good.

kamild_

I take it this exemplary 1.35 GB folder will now take up just as much space in RAM instead?

Shiloh Vanaver

4.1 is a bit better though a little glitchy. But you'll notice that if you export PNGs, that the jumpy frames are actually the input images. You can shift them all down a few pixels in an editing program.

Michael Carychao

Amazing! Looking forward to it.

FamKis

Totally awesome! Love this news! This probably will blow T*paz Video Enhance AI out of the water, speed wise at least.

WaveCut

Can I ask if RIFE 4.1 experimental anyhow different from 4.0?

WaveCut

Oh! My bad. Thanks for the clarification, waiting for the release.

LTSM

Why? NCNN is faster now (with 4.0 model) and does not require 3 GB of dependencies.

WaveCut

That's neat, I wish this could be possible with CUDA sometime


More Creators