MobileDiffusion: Rapid text-to-image generation on-device

Kind of funny that they show the iphone 15 pro and the Samsung S24 in the comparison chart, but not their own phone the google pixel 8. (I know it will perform worse than both phones)

Re: MobileDiffusion: Rapid text-to-image generation on-device

#5

Post by minimaxir » Wed, Jan 31, 2024, 11:36 PM UTC

> With superior efficiency in terms of latency and size, MobileDiffusion has the potential to be a very friendly option for mobile deployments given its capability to enable a rapid image generation experience while typing text prompts. And we will ensure any application of this technology will be in-line with Google’s responsible AI practices.

So I'm interpreting this that it won't ever get released.

Re: MobileDiffusion: Rapid text-to-image generation on-device

#6

Post by cuuupid » Thu, Feb 01, 2024, 12:03 AM UTC

Google has fallen so far. Both Inception and Mobilenet were released openly and changed the entire AI world.

Nowadays we just get blog posts about results that were supposedly achieved, an accompanying paper that can’t be reproduced (because of Google’s magical “private datasets”), and some screencaps of a cool application of the tech that is virtually guaranteed to never make it to product.

Re: MobileDiffusion: Rapid text-to-image generation on-device

#7

Post by minimaxir » Thu, Feb 01, 2024, 12:08 AM UTC

cuuupid wrote: post #6

Google has fallen so far. Both Inception and Mobilenet were released openly and changed the entire AI world. Nowadays we just get blog posts about results that were supposedly achieved, an accompanying paper that can’t be reproduced (because of Google’s magical “private datasets”), and some screencaps of a cool application of the tech that is virtually guaranteed to never make it to product.

No one ever filed multimillion dollar lawsuits over Inception and MobileNet.

The calculus is different now.

Re: MobileDiffusion: Rapid text-to-image generation on-device

#8

Post by Legend2440 » Thu, Feb 01, 2024, 12:13 AM UTC

minimaxir wrote: post #5

> With superior efficiency in terms of latency and size, MobileDiffusion has the potential to be a very friendly option for mobile deployments given its capability to enable a rapid image generation experience while typing text prompts. And we will ensure any application of this technology will be in-line with Google’s responsible AI practices. So I'm interpreting this that it won't ever get released.

I would interpret it as "expect to see this powering some features in the next-generation Pixel".

Re: MobileDiffusion: Rapid text-to-image generation on-device

#9

Post by jsheard » Thu, Feb 01, 2024, 12:21 AM UTC

Legend2440 wrote: post #8

minimaxir wrote: post #5

> With superior efficiency in terms of latency and size, MobileDiffusion has the potential to be a very friendly option for mobile deployments given its capability to enable a rapid image generation experience while typing text prompts. And we will ensure any application of this technology will be in-line with Google’s responsible AI practices. So I'm interpreting this that it won't ever get released.

I would interpret it as "expect to see this powering some features in the next-generation Pixel".

We've already seen this progression - they debuted Magic Eraser as a cloud feature, then with the Pixel 8 they got it running locally on the device. But they also introduced Magic Editor with the Pixel 8, running on the cloud, and the Pixel 9 or 10 will probably run it on-device.

Re: MobileDiffusion: Rapid text-to-image generation on-device

#10

Post by kccqzy » Thu, Feb 01, 2024, 12:58 AM UTC

minimaxir wrote: post #5

> With superior efficiency in terms of latency and size, MobileDiffusion has the potential to be a very friendly option for mobile deployments given its capability to enable a rapid image generation experience while typing text prompts. And we will ensure any application of this technology will be in-line with Google’s responsible AI practices. So I'm interpreting this that it won't ever get released.

I'm interpreting it as they will be adding a layer of safety restrictions. Understandable given the furore of the recent Taylor Swift generated image incident.

Everyone needs to do this and probably is already doing this. Search for "ChatGPT lobotomized" and you'll see plenty of complaints about the safety filters added by OpenAI.