old but relevant, it&#x27;s like a production system. We can continue to tune and squeeze 1mp pictures out of it in high quality on relevant localized hardware without breaking a sweat. 1-3 seconds on a m3 system, and then cascade of tooling below stays the same.

There&#x27;s also sdcpp ( <a href="https:&#x2F;&#x2F;github.com&#x2F;leejet&#x2F;stable-diffusion.cpp" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;leejet&#x2F;stable-diffusion.cpp</a> ) . It is for people that know command line shells like bash or Powershell. SDCPP is like llama.cpp but it is for using text to image AI models. Not everyone likes the node&#x2F;graph programming interface in Comfy or a Webpage interface. Some people need these GUI&#x2F;web interfaces and get lost and frustrated in shell scripting.

A1111 (or rather Forge Neo at this point) can still be pretty useful (for a Windows user) if somebody just want a batteries-included approach because it makes it trivial to move back and forth between masking, manual inpainting, tacking on a high‑res fix and&#x2F;or upscale, and swapping things out on the fly quickly.Caveat: No idea how well A1111 and its variants function on Mac though.These days I mostly use ComfyUI because after building out a node‑based workflow, it’s easy to export as a JSON file with websocket outputs that I can integrate into my programs.<a href="https:&#x2F;&#x2F;github.com&#x2F;Haoming02&#x2F;sd-webui-forge-classic&#x2F;tree&#x2F;neo" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;Haoming02&#x2F;sd-webui-forge-classic&#x2F;tree&#x2F;neo</a>

A1111 also had its time. Comfy takes a couple hours to learn, and then it’s vastly superior. Like, if you are already on Mac there is no reason to also kick yourself in the face trying to go back to 1111.

SD 1.5 is so insanely old at this point im not sure the point of this when you have Mini Max H3 on metal <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=49252179">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=49252179</a>

I love AI content. I hate content written by AI.It&#x27;s not serious. It shows that the author doesn&#x27;t care about showing what he&#x27;s up to.He cares about making content to post on a blog.And that&#x27;s disrespectful to the reader.

If anyone is triggered by AI content they surely avoid HN

Invoke AI is also really good if you prefer working in a canvas-style environment, in a more traditional graphical way. It easily has the nicest workflows for outpainting&#x2F;regional&#x2F;inpainting by far.<a href="https:&#x2F;&#x2F;github.com&#x2F;invoke-ai&#x2F;invokeai" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;invoke-ai&#x2F;invokeai</a>

What a crazy repo. Some cyber archeologists need to do a case analysis of the entire diffusion model ecosystem circa the original Stable Diffusion up to now.

Automatic1111 wasnt being updated, people moved to Forge Neo, ComfyUI and Maestro.

Yes it includes LoRA composition as well, a bunch from the internet, and they work just fine!

The speedup includes running LoRAs, right? Which ones are you using?

A1111 is such a throwback. It&#x27;s been ages since that &amp; AI image generation was cool. Back before the massive amount of slop and when things were just a fun experiment

Automatic1111 for Apple metal, 40% speed up sd1.5