Why are deepfake devs so bad

Seriously. There's not a single deepfake software in existence that works out the box. They all require some hacky source code edits to work. All of them have 50+ open issues. Why is this so difficult.

Attached: EBFoE4vUwAAN3Is.jpg (827x736, 47K)

that's not true dumb virgin

provide a counterexample

DFL and faceswap are piss easy to run though.
>All of them have 50+ open issues.
That's not uncommon for bigger project. Also faceswap has 7 at the moment.

im running into a clusterfuck of errors involving a bunch of modules like plaidml.

What steps did you take until now (as detailed as possible)?

>Thinly veiled request for a spoonfeed
Fuck off

>He can’t find a counterexample

Is there a good source on how these all work? I understand some use autoencoders and some use GAN but I can tell which open source tools are using what method since they dont really document how their github programs work

>she thinks bobs and vagane coders can make something high quality

Just goes to show deepfake users are entitled brats. It's FREE you moron. Someone sat down and wasted his life on technology's final frontier - AI - and you bitch and whine. They should never have developed it, nor put up the source code.

GAN is surprisingly rare when it comes to common deepfake software. There's faceswap-GAN and DFL seems to offer a half-assed RecycleGAN implementation. That's it afaik.

Deepfake is the most worthless fucking software on the planet and everything it generates looks horrendous. How anyone could get turned on by nipples and vaginas that look like plastic surgery disasters is beyond me. Anyone who uses deepfake is pathetic, is the lowest type of subhuman, and needs to be immediately killed. I say kill instead of murder, because your death would be an amazing thing, and a true act of righteousness. Mashallah, every word I have said is beyond reproach. Destroy yourselves immediately.

I think they can look good with the right amount of effort and source material.

You're thinking about DeepNude you niggermost retard.

This, deep nude is too ambitious

GAN does work better I think?

Alhamdulillah, every word I have said is beyond reproach.

They’re too busy jerking it.

There’s no future in deep fake software. It’s not something you can really put on a resume or tell people about. And the people who work on it are fucking losers who masturbate 10 times a second

True. Jerking it between each commit

In theory, yes. But in practice nobody made a decent GAN implementation yet. DFL's RecycleGAN collapses after a few thousand iterations, last time I checked anyway, and shaoanlu seems to be the only one able to achieve good results with his faceswap-GAN. And faceswap's port of faceswap-GAN (yeah, it's a bit confusing) was broken from the start and stayed that way for a long time. Maybe it's still broken (I haven't followed development for quite a while).

You have been visited by the Laura of not great, not terrible threads.

This thread is currently reading 20 replies (not great, not terrible).

Attached: Laura.jpg (1275x637, 108K)