InvokeAI - Adding Models (v2.2.5)

This software is free and open source - You can download and run it here, but make sure to check system requirements.

With 2.2.5, InvokeAI allows you to easily add both standard and custom Stable Diffusion models directly from the UI. This video covers the "How To".

A huge shout-out to all of the translators who helped to translate InvokeAI - Which is now available in **8 languages**!

Join the Discord for help and tips!

- Thanks from the InvokeAI Team

----------------

Music by FASSounds from Pixabay

[MMOStars]

The constant development is really good, don't stop 👍

[GraemeHindmarsh]

Thank you! InvokeAI just got a whole lot more useful for me!

[RNmarkie]

Loving the software, and very much enjoying the frequent releases. Thanks so much!

[hangry]

This is huge. I'm so stoked for this to get even better.

[BeetlebombD]

I checked out InvokeAI about 4-5 months ago and was absolutely stunned by how amazing the UI looked, and your outpainting approach is unrivaled. However, I was disappointed for it's lack of easy custom model integration, but this update solves everything! I'll have to download invokeAI and give it another try :)

[DanielAngel]

"May your hands be perfect!" 🤣

[GameInTheory]

I think you're creating something marvelous here, its a very powerful tool, that I think is honestly going to shake up the creative industry. (In a good very way)

[ianjohnson]

Thank you! I was having trouble adding models manually. I can't wait to try this out.

[theresalwaysanotherway]

what an incredible app, it's making a huge difference to the accessibility of stable diffusion, and when I'm asked to help someone install stable diffusion, I'll probably guide them for Invoke over automatic1111. Unfortunately it still has yet to reach feature parity for me, as I'm waiting on:
1) v2 support
2) xformers support
3) ability to easily switch VAEs independently of the model
4) community plugins/extensions (absolutely the hardest/biggest one)
5) WD1.4 tagger
6) LORA support
Which I can only imagine at this level of polish could take over a year to complete, assuming you'd even want if you're trying to avoid clutter and a maintenance nightmare. Either way, keep doing what you're doing, your rate of progress is staggering!

[billybobnz]

“May your hands be perfect” good humour

[HalkerVeil]

I've been using AI for awhile and I have never heard of this magical VAE file that we are supposed to magically know all about and just pipe it in like normal...
wtf?

[JD-qowk]

Loving InvokeAI so far. However I'm keenly awaiting implementation of Stable Diffusion v2! Any news on when we can support for v2 to be added? That would really catapult InvokeAI way ahead of a lot of the other popular tools around at the moment.

[playstuff]

Love InvokeAI and the Updates. Will Face Restoration be coming to the Unified Canvas?

[HarryJPotter]

ckpt folder? YAY! clicking that update button now...

[madshader]

Can you please support safetensors? A lot of new models these days are safetensor only.

[AimerYui]

Can you please add the InvokeAI version to the GUI, for example add it beside the InvokeAI logo. This will really help the user to know what version they are using.

[ellenripley]

I know this is a bit off topic but what's the best tool for an artist to train their own models with their own artwork?

[andrinSky]

Hello, is it possible to import/Add Model the fantastic "Chilloutmix" to Invoke AI? If yes, how?

[HalkerVeil]

A full year later, still no info on this VAE file?
Every time I try to add the VAE from Automatic1111, and copy it over, it throws errors. And there is no way to choose the VAE within the "easy" UI.

[DataJuggler]

Anytime line for Stable Diffusion 2.0 support?