Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Imagen, a text-to-image diffusion model

gweb-research-imagen.appspot.com

Post reply on HN
Print view View original

551–560 of 661 posts

Re: Imagen, a text-to-image diffusion model

#551

Post by zimpenfish »

ALittleLight wrote: post #284

Interesting to me that this one can draw legible text. DALLE models seem to generate weird glyphs that only look like text. The examples they show here have perfectly legible characters and correct spelling. The difference between this and DALLE makes me suspicious / curious. I wish I could play with this model.

The latent-diffusion[1] one I've been playing with is not terrible at drawing legible text but generally awful at actually drawing the text you want (cf. [2]) (or drawing text when you don't want any.)

[1] https://github.com/CompVis/latent-diffusion.git [2] https://imgur.com/a/Sl8YVD5

Re: Imagen, a text-to-image diffusion model

#553

Post by richrichardsson »

bowmessage wrote: post #489
qz_kb wrote: post #437

I have to wonder how much releasing these models will "poison the well" and fill the internet with AI generated images that make training an improved model difficult. After all if every 9/10 "oil painted" image online starts being from these generative models it'll become increasingly difficult to scrape the web and to learn from real world data in a variety of domains. Essentially once these things are widely availa…

I also worry about the potential to further stifle human creativity, e.g. why paint that oil painting of a panda riding a bicycle when I could generate one in seconds?

One reason:

A digital picture of an oil painting != an actual oil painting

Of course once someone trains an AI with a robotic arm to do the actual painting, then your worry holds firm.

Re: Imagen, a text-to-image diffusion model

#554

Post by mdp2021 »

LudwigNagasena wrote: post #300
mdp2021 wrote: post #177

Earlier quoted context omitted.

Very certainly not, since use is individual and thus a function of competence. So, adherence to meaning depends on the user. Conflict resolution? And anyway - contextually -, the representational natures of "use" (instances) and that of "meaning" (definition) are completely different.

Definition is an entirely artificial construct and doesn't equate to meaning. Definition depends on other words that you also have to understand.

You are thinking of the literal definition - that "made of literal letters".

Mental definition is that "«artificial»" (out of the internal processing) construct made of relations that reconstructs a meaning. Such ontology is logical - "this is that". (It would not be made of memories, which are processed, deconstructed.)

Concepts are internally refined: their "implicit" definition (a posterior reading of the corresponding mental low-level) is refined.

Re: Imagen, a text-to-image diffusion model

#555

Post by PoignardAzur »

dragonwriter wrote: post #246
renewiltord wrote: post #229

Earlier quoted context omitted.

Copenhagen ethics (used by most people) require that all negative outcomes of a thing X become yours if you interact with X. It is not sensible to interact with high negativity things unless you are single-issue. It is logical for Google to not attempt to interact with porn where possible.

> Copenhagen ethics (used by most people) The idea that most people use any coherent ethical framework (even something as high level and nearly content-free as Copenhagen) much less a particular coherent ethical framework is, well, not well supported by the evidence. > require that all negative outcomes of a thing X become yours if you interact with X. It is not sensible to interact with high negativity things unless…

> The idea that most people use any coherent ethical framework (even something as high level and nearly content-free as Copenhagen) much less a particular coherent ethical framework is, well, not well supported by the evidence.

I don't have any evidence, but my personal experience is that it feels correct, at least on the internet.

People seem to have a "you touch it, you take responsibility for it" mindset regarding ethical issues. I think it's pretty reasonable to assume that Google execs are assuming "If anything bad happens because of AI, we'll be blamed for it".

Re: Imagen, a text-to-image diffusion model

#556

Post by wongarsu »

ithkuil wrote: post #547
SirHound wrote: post #532

Earlier quoted context omitted.

Most images you see from these services will not have a watermark on them. Cropping is trivial.

Perhaps a watermark should be embedded in a subtle way across the whole image. What is the word? "Steganography" is designed to solve a different problem and I don't think it survives recompression etc. Is there a way to create weakly secure watermarks that are invisible to naked eye, spread across the whole image, resistant to scaling and lossless compression (to a point)?

Invisible, robust watermarks had a lot of attention in research from the late 90s to the early 10s, and apparently some resurgence with the availability of cheap GPU power.

Naturally there's a python library [1] with some algorithms that are resistant to lossy compression, cropping, brightness changes, etc. Scaling seems to be a weakness though.

1: https://pypi.org/project/invisible-watermark/

Re: Imagen, a text-to-image diffusion model

#557

Post by londons_explore »

mistrial9 wrote: post #119

Reading a relatively-recent Machine Learning paper from some elite source, and after multiple repititions of bragging and puffery, in the middle of the paper, the charts show that they had beaten the score of a high-ranking algorithm in their specific domain, moving the best consistant result from 86% accuracy to 88% accuracy, somewhere around there. My response was: they got a lot of attention within their world by…

If you worked in a hospital and you managed to increase the survival rate from 86% to 88%, you too would be a hero.

Sure, it's only 2%, but if it's on a problem where everyone else has been trying to make that improvement for a long time, and that improvement means big economic or social gains, then it's worth it.

Re: Imagen, a text-to-image diffusion model

#558

Post by avn2109 »

richrichardsson wrote: post #553
bowmessage wrote: post #489

Earlier quoted context omitted.

I also worry about the potential to further stifle human creativity, e.g. why paint that oil painting of a panda riding a bicycle when I could generate one in seconds?

One reason: A digital picture of an oil painting != an actual oil painting Of course once someone trains an AI with a robotic arm to do the actual painting, then your worry holds firm.

Surely this already exists right?

Re: Imagen, a text-to-image diffusion model

#559

Post by actionfromafar »

joshspankit wrote: post #460
rhacker wrote: post #448

Earlier quoted context omitted.

Look at carpentry blogs, recipe blogs. Nearly all of it is junk content. I bet if you combined GPT and imagen or dalle2 you could replace all of them. Just provide a betty crocker recipe and let it generate a blog that has weekly updates and even a bunch of images - "happy family enjoying pancakes together" I can see the future as being devoid of any humanity.

I’d much rather skip the blog format and replace them with an AI that can answer “Please provide a pie recipe like my grandparent’s”, or “I’d like to make these ribs on the BBQ so that they come out flavourful, soft, and a little sweet.”

- 100mg of Zoloft

wash it down with water.

Re: Imagen, a text-to-image diffusion model

#560

Post by geonic »

Can anybody give me short high-level explanation how the model achieves these results? I'm especially interested in the image synthesis, not the language parsing.

For example, what kind of source images are used for the snake made of corn[0]? It's baffling to me how the corn is mapped to the snake body.

[0] https://gweb-research-imagen.appspot.com/main_gallery_images...

Post reply on HN