Is AuraFlow The Open Source Model Of The Future?

Is AuraFlow The Open Source Model Of The Future? I guess it's still TBD however so far very promising. It's been refreshing to have so many open source models coming out lately as I've found SDXL the same old stuff. AuraFlow's prompt coherency is probably one of the best I've come across so far, dare I say on par with SD3? Keep in mind folks, this is a beta model so there is room for improvements as mentioned in the video. The potential is very, very promising!

⏲Time Stamps
0:00 Intro to AuraFlow
1:25 How to use AuraFlow with ComfyUI
2:49 Remember this is a beta model!!!
3:41 Recommended settings
5:09 Performance on my 3060Ti
5:35 Prompt coherency tests
6:28 Comparing 15-30 steps
7:44 A whole bunch of examples!
12:07 Some advice and question for you

**Disclaimer Affiliate Links Below**

📸 Gear I use

🔦 Find us on:

[MonzonMedia]

Honestly I think if they develop their own controlnets and supporting conditioning methods, this can be the next model that the open source community adopts. Very impressed with a high potential! I'd love to hear about your experiences in the comments.

[bahethelmy]

Thanks for the video!! The model looks good. I'll wait until they release a version that handles photorealism.

[Lazniak]

Great model to produce the base image composition, than you can enchance with some more detail ones. Thanks for video.

[sylvain]

the lower you go on the cfg the better it is for Realistic images between 1.6 to 2.2

[AnthonyWashingtonXK]

Great video. I hope by the end of August to be able yo get something from NVIDEA's 40 series. So many new things in the AI imaged world.

[NadjibIxe]

doesn't work with swarmui ? with the comfyui tab integrated

[alfadelta-dd]

Unfortunately for the project, the recent debut of Flux ends any game for AuraFlow. At the moment, On civitAI we have 100 new LORA coming out each day for Flux. Nothing or near to nothing for Aura...

[Camo_Corey]

I keep getting errors when trying to queue prompt?

[danielc]

Will you try kolors and 900m pixart sigma?

[cmdr_stretchedguy]

30 steps is the standard for Fooocus, and without having to setup all the extra bits and pieces produces some amazing images (via SDXL).

[fixelheimer]

Is upscaling with latent like with sdxl possible?

[splatt]

Ultimately it boils down to will it do lewds. And if so how do said lewds compare to models like pony. Don't hate I don't make the rules.

[yiluwididreaming]

nope. not at wait until the model is further