#Parallel Image Generation
17 messages · Page 1 of 1 (latest)
Curious how this would benefit those with lower VRAM, as I would expect it to require more, since it's generating more than 1 image at a time
also ... bringing it back? was it ever an option? heh
Yes, it was an option in very early versions of Invoke.
Depends on how low.
i don't believe it was ever an option, and yes it would use more vram, as well as be slower, so i'm not sure how it'd help people
It was and the amount of time to generate 4 images opposed to one too only a fraction longer.
what version?
But didn't take 4x time to generate 4 images.
i believe the feature was short-lived because SDXL then came out, and now we've got sd 3.5 and flux where the trick is to get them to even fit on a gpu for one generation
in the days of SD 1.5 it made more sense
https://www.restack.io/p/invokeai-answer-adetailer-cat-ai?t&utm_source=perplexity
see "Batch Processing with invokeai-batch"
Explore the features and capabilities of InvokeAI ADetailer for advanced image generation and manipulation. | Restackio
Makes sense.
I've been using invoke for several years now, and I don't remember invoke ever having a parallel generation feature. Batching taking x times longer for x images would make sense, since I'm pretty sure this just means 'generate x images before I need to hit invoke again'