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

FLUX is fast and it's open source

replicate.com

Post reply on HN
Print view View original

41–50 of 127 posts

Re: FLUX is fast and it's open source

#41

Post by jsemrau »

ruthmarx wrote: post #36
jsemrau wrote: post #30

My favorite thing to do with Flux is create images with a white background for my substack[1] because the text following is amazing and I can communicate something visually through the artwork as well. [1] https://substackcdn.com/image/fetch/w_1456,c_limit,f_webp,q_...

That example you gave is a good reason why artists get pissed off IMO. The LLM is clearly aping some artists specific style, and now missing out on paid work as a result. Not sure I have an opinion on that, technology marches on etc, but it is interesting.

I understand your point, but in 0% of all cases would I hire an artist to create imagery for my personal blog. Therefore, I would think that market doesn't exist.

Re: FLUX is fast and it's open source

#42

Post by popalchemist »

swyx wrote: post #35
weird-eye-issue wrote: post #20

Earlier quoted context omitted.

I mean it literally explains why in the second paragraph. It returns the actual file data in the response rather than a URL where you have to make a second request to get the file data

thats not "making the image models much faster", thats just making getting the image back slightly faster

The "making the image models much faster" part is model optimizations that are also explained in the post.

Re: FLUX is fast and it's open source

#43

Post by teh_infallible »

sorenjan wrote: post #23

Text to image models feels inefficient to me. I wonder if it would be possible and better to do it in separate steps, like text to scene graph, scene graph to semantically segmented image, segmented image to final image. That way each step could be trained separately and be modular, and the image would be easier to edit instead of completely replace it with the output of a new prompt. That way it should be much easie…

I am hoping that AI art tends towards a modular approach, where generating a character, setting, style, and camera movement each happens in its own step. It doesn’t make sense to describe everything at once and hope you like what you get.

Re: FLUX is fast and it's open source

#44

Post by trickstra »

Non-commercial is not open-source, because if the original copyright holder stops maintaining it, nobody else can continue (or has to work like a slave for free). Open-source is about what happens if the original author stops working on it. Open-source gives everyone the license to continue developing it, which obviously means also the ability to get paid. Don't call it open-source if this aspect is missing.

Only the FLUX.1 [schnell] is open-source (Apache2), FLUX.1 [dev] is non-commercial.

Re: FLUX is fast and it's open source

#45

Post by WithinReason »

kqr wrote: post #40
sorenjan wrote: post #23

Text to image models feels inefficient to me. I wonder if it would be possible and better to do it in separate steps, like text to scene graph, scene graph to semantically segmented image, segmented image to final image. That way each step could be trained separately and be modular, and the image would be easier to edit instead of completely replace it with the output of a new prompt. That way it should be much easie…

Isn't this essemtially the approach to image recognition etc. that failed for ages until we brute forced it with bigger and deeper matrices? It seems sensible to extract features and reason about things the way a human would, but it turns out its easier to scale pattern matching purely done by computer.

This is Sutton's Bitter Lesson:

https://www.cs.utexas.edu/~eunsol/courses/data/bitter_lesson...

Re: FLUX is fast and it's open source

#46

Post by thomashop »

llm_trw wrote: post #32
ZoomZoomZoom wrote: post #28

Earlier quoted context omitted.

The issue with this is there's a false assumption that an image is a collection of objects. It's not (necessarily). I want a picture of frozen cyan peach fuzz.

https://imgur.com/ayAWSKr Prompt: frozen cyan peach fuzz, with default settings on a first generation SD model. People _seriously_ do not understand how good these tools have been for nearly two years already.

You can do this dynamically with Pollinations URLs too:

https://pollinations.ai/p/frozen_cyan_peach_fuzz?seed=1

https://pollinations.ai/p/frozen_cyan_peach_fuzz?seed=2

https://pollinations.ai/p/frozen_cyan_peach_fuzz?seed=3

Disclaimer: I'm behind Pollinations.AI

Re: FLUX is fast and it's open source

#47

Post by thomashop »

If you want to play with FLUX.schnell easily, type the prompt into a Pollinations URL:

https://pollinations.ai/p/a_donkey_holding_a_sign_with_flux_...

https://pollinations.ai/p/a_donkey_holding_a_sign_with_flux_...

https://pollinations.ai/p/Minimalist%20and%20conceptual%20ar...

It's incredible how fast it is. We generate 8000 images every 30 minutes for our users using only three L40S GPUs. Disclaimer: I'm behind Pollinations

Re: FLUX is fast and it's open source

#48

Post by 112233 »

Does someone know what FLUX 1.1 has been trained on? I generated almost hundred images on the pro model using "camera filename + simple word" two word prompts, and it all looks like photos from someones phone. Like, unless it has text I would not even stop to consider any of these images AI. They sometimes look cropped. A lot of food pictures, messy tables and appartments etc.

Did they scrape public facebook posts? Snapchat? Vkontakte? Buy private images from onedrive/dropbox? If I put as the second word a female name, it almost always triggers nsfw filter. So I assume images in the training set are quite private.

See for yourself (autoplay music warning):

people: https://vm.tiktok.com/ZGdeXEhMg/

food and stuff: https://vm.tiktok.com/ZGdeXEBDK/

signs: https://vm.tiktok.com/ZGdeXoAgy/

[edit] Looking at these images feels uneasy, like I am looking at someones private photos. There is not enough "guidance" in a prompt like "IMG00012.JPG forbid" to account for these images, so it must all come from the training data.

I do not believe FLUX 1.1 pro has radically different training set than these previous open models, even if it is more prone to such generation.

It feels really off, so, again, is there any info on training data used for these models?

Re: FLUX is fast and it's open source

#49

Post by ErikBjare »

popalchemist wrote: post #42
swyx wrote: post #35

Earlier quoted context omitted.

thats not "making the image models much faster", thats just making getting the image back slightly faster

The "making the image models much faster" part is model optimizations that are also explained in the post.

Where? I don't see any explanation of model optimizations in the linked post.

Re: FLUX is fast and it's open source

#50

Post by ruthmarx »

jsemrau wrote: post #41
ruthmarx wrote: post #36

Earlier quoted context omitted.

That example you gave is a good reason why artists get pissed off IMO. The LLM is clearly aping some artists specific style, and now missing out on paid work as a result. Not sure I have an opinion on that, technology marches on etc, but it is interesting.

I understand your point, but in 0% of all cases would I hire an artist to create imagery for my personal blog. Therefore, I would think that market doesn't exist.

Yeah, I get that completely, I'm the same way. I just think it's interesting. It's kind of the same argument as piracy, since most people wouldn't pay for what they download if it wasn't free.
Post reply on HN