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

Releasing weights for FLUX.1 Krea

krea.ai

Post reply on HN
Print view View original

91–100 of 112 posts

Re: Releasing weights for FLUX.1 Krea

#91

Post by vunderba »

dvrp wrote: post #83
vunderba wrote: post #80

Nice release. Ran some preliminary tests using the 12b Txt2Img Krea model. Its biggest wins seems to be raw speed (and possibly realism) but perhaps unsurprisingly did not score any higher on the leaderboard for prompt adherence than the normal Flux.1D model. https://genai-showdown.specr.net On another note, there seem to be some indication that Wan 2.2+ future models might end up becoming significant players in the…

Can you point to a URL with the tests you’ve done? Also, FWIW, this model focus was around aesthetics rather than strict prompt adherence. Not to excuse the bad samples, but to emphasize what was one of the research goals. It’s a thorny trade-off, but an important one if one wants to get rid of what’s sometimes known as “the flux look”. Re: Wan 2.2 I’ve also been reading of people commenting about using Wan 2.2 for b…

The Image Showdown site actually does have Flux Krea images but they're hidden by default. If you open up the "Customize Models" dialog you can compare them against other Flux models (Flux.1 Dev and Kontext).

> FWIW, this model focus was around aesthetics

Agreed - whereas these tests are really focused on various GenAI image models ability to follow complicated prompts and are not as concerned with overall visual fidelity.

Regarding the "flux look" I'd be interested to see if Krea addresses both the waxy skin look AND the omnipresent shallow depth of field.

Re: Releasing weights for FLUX.1 Krea

#92

Post by johnisgood »

orphea wrote: post #63
vipermu wrote: post #7

hey hn! I'm one of the founders at Krea. we prepared a blogpost about how we trained FLUX Krea if you're interested in learning more: https://www.krea.ai/blog/flux-krea-open-source-release

Off topic but did you really hide scroll bars on the website? Why...? .scrollbar-hide { -ms-overflow-style: none; scrollbar-width: none; }

They probably did it because the website might look better without a scrollbar, but they should realize that many browsers hide the scrollbar and they only get displayed when you hover over or when you start scrolling. That said, the scrollbar is always there for me (unless hidden by CSS), and I would not have minded it at all.

Re: Releasing weights for FLUX.1 Krea

#93

Post by iyn »

dragonwriter wrote: post #75
piperswe wrote: post #19

Earlier quoted context omitted.

A parameter can be any size float. Lots of downloadable models are FP8 (8 bits per parameter), but it appears this model is FP16 (16 bits per parameter) Often, the training is done in FP16 then quantized down to FP8 or FP4 for distribution.

I think they are bfloat16, not FP16, but they are both 16bpw formats, so it doesn't make a size difference.

Wiki article on bfloat16 for reference, since it was new to me: https://en.wikipedia.org/wiki/Bfloat16_floating-point_format

Re: Releasing weights for FLUX.1 Krea

#94

Post by jamespo »

renewiltord wrote: post #88
codedokode wrote: post #86

Earlier quoted context omitted.

Human learning and computer processing millions of works are different things. I don't think any human artists have seen as many images as the developers used for training.

Then let he who hath not sinned cast the first stone.

Yes scale is totally irrelevant, that's what all of FAANG tell us too :/

Re: Releasing weights for FLUX.1 Krea

#95

Post by dragonwriter »

Tokumei-no-hito wrote: post #87
dragonwriter wrote: post #75

Earlier quoted context omitted.

I think they are bfloat16, not FP16, but they are both 16bpw formats, so it doesn't make a size difference.

pardon the ignorance but it's the first time I've heard of bfloat16. i asked chat for an explanation and it said bfloat has a higher range (like fp32) but less precision. what does that mean for image generation and why was bfloat chosen over fp?

My fuzzy understanding, and I'm not at all an expert on this, that the main benefit is that bf16 is less prone to overflow/underflow during calculation, which is a source of bigger problems in both training and inference than the simple loss of precision, so once it became widely supported, it became a commonly-preferred format for models (whether image gen or otherwise) over FP16.

Re: Releasing weights for FLUX.1 Krea

#96

Post by cubefox »

dvrp wrote: post #2

Hello everyone. I’m the Co-founder and CTO of Krea. We’re excited because we wanted to release the weights for our model and share it with the HN community for a long time. My team and I will try to be online and try to answer any questions you may have throughout the day.

Regarding the P(.|photo) vs P(.|minimal) example, how do you actually decide this conflict? It seems to me that photorealism should be a strong default "bias".

My reasoning: If the user types in "a cat reading a book" then it seems obvious that the result should look like a real cat which is actually reading a book. So it obviously shouldn't have an "AI style", but it also shouldn't produce something that looks like an illustration or painting or otherwise unrealistic. Without further context, a "cat" is a photorealistic cat, not an illustration or painting or cartoon of a cat.

In short, it seems that users who want something other than realism should be expected to mention it in the prompt. Or am I missing some other nuances here?

Re: Releasing weights for FLUX.1 Krea

#97

Post by forkeep »

yieldcrv wrote: post #59
cchance wrote: post #57

Earlier quoted context omitted.

People underestimate how much goodwill companies gain from pushing opensource stuff out, not just from word of mouth but even picking up users for their commercial offerings too, while i could run opensource and appreciate it in a lot of cases using API's from the companies that i like (mostly ones that do opensource stuff) tends to be easier for bigger stuff...

(unless the code repository and history is embarrassingly bad, which is most repositories)

[dead]

Re: Releasing weights for FLUX.1 Krea

#99

Post by Western0 »

dvrp wrote: post #2

Hello everyone. I’m the Co-founder and CTO of Krea. We’re excited because we wanted to release the weights for our model and share it with the HN community for a long time. My team and I will try to be online and try to answer any questions you may have throughout the day.

I need model for other language than english

Re: Releasing weights for FLUX.1 Krea

#100

Post by bsenftner »

sangwulee wrote: post #4

Hi! I'm lead researcher on Krea-1. FLUX.1 Krea is a 12B rectified flow model distilled from Krea-1, designed to be compatible with FLUX architecture. Happy to answer any technical questions :)

From a traditional media production background, where media is produced in separate layers, which are then composited together to create a final deliverable still image, motion clip, and/or audio clip - this type of media production through the creation of elements that are then combined is an essential aspect of expense management, and quality control. Current AI image, video and audio generation methods do not support any of that. ForgeUI did briefly, but that went away, which I suspect because few understand large scale media production requirements.

I guess my point being: do you have any (real) experienced media production people working with you? People that have experience working in actual feature film VFX, animated commercial, and multi-million dollar budget productions?

If you really want to make your efforts a wild success, simply support traditional media production. None of the other AI image/video/audio providers seem to understand this, and it is gargantuan: if your tools plugged into traditional media production, it will be adopted immediately. Currently, they are tentatively and not adopted because they do not integrate with production tools or expectations at all.

Post reply on HN