[HN Gopher] FLUX.1 Kontext [Dev] - Open Weights for Image Editing
___________________________________________________________________
FLUX.1 Kontext [Dev] - Open Weights for Image Editing
Author : minimaxir
Score : 117 points
Date : 2025-06-26 15:31 UTC (7 hours ago)
(HTM) web link (bfl.ai)
(TXT) w3m dump (bfl.ai)
| minimaxir wrote:
| The new non-commercial license is a bit of a doozy:
| https://huggingface.co/black-forest-labs/FLUX.1-Kontext-dev/...
| samtheprogram wrote:
| If I'm understanding this correctly, you can't run this in a
| commercial setting, even if you're not creating a derivative
| but simply generating outputs?
| smerrill25 wrote:
| I believe you can buy licensing? But def not the same as
| 'Open Weights..'
| StevenWaterman wrote:
| "Weights available" perhaps
| stefan_ wrote:
| The same people that claim using all of humanities creation
| is fair use want you to pay for a bunch of MatMul inputs that
| are unrecognizable to anyone after quantizing them yourself.
| cchance wrote:
| Stupid question, whats to stop someone from quantizing it,
| shit even just barely finetuning it for 1 step and calling
| it something different, no ones actually checking WTF these
| models are based on when they're released, especially for
| the source models, especially if the release is not around
| the same time of release as the base, i'm 99% sure someone
| could fine tune SD3.5 a bit and release it today as Frizz
| 1.0 and people would just take it as a new model using the
| same layer structure as SD3.5 lol
| elpocko wrote:
| Not impossible but you'd gonna have to do a bit more than
| that. Most people are ignorant, but not all of them. An
| experienced user can tell what model family was used from
| a bunch of generated images. Also, no one would believe a
| nobody who just showed up claiming to have trained a
| brand new diffusion model.
| SV_BubbleTime wrote:
| I forget which, but some HiDream maybe was called out for
| this when it happened to generate basically the same dude
| in front of the same archway when compared against flux.
| liuliu wrote:
| HiDream is a separate architecture. OTOH, it _might_ be
| finetuned on FLUX generated data, we will never know.
| doctorpangloss wrote:
| HiDream is trained on AI generated outputs.
| liuliu wrote:
| There is a simple method to detect this: taking a model
| "claimed" to be trained scratch, taking the model you
| suspected is the original, generate a new model =
| claimed_model * 0.5 + suspected_model * 0.5.
|
| If the claimed_model is trained from scratch, the new
| model will have 0 capability (basically generate
| gibberish words or noise). If it is a derivative of the
| suspected model, it will do something sensible.
|
| It is a bit more interesting for diffusion model because
| you can fine-tune to a different objective, making this
| investigation harder to do, but not impossible.
| doctorpangloss wrote:
| FLUX watermarks its outputs.
|
| Additionally, certain prompts will produce nonsensical
| but specific outputs known only to BFL.
| whywhywhywhy wrote:
| The double standard is frankly disgusting.
|
| I'm actually all for open training but I think it's only
| fair you treat the model as your treated the life's work of
| others.
| ronsor wrote:
| Quite frankly, I still believe that these model licenses
| are dubiously enforceable at best, and I'm skeptical that
| models are copyrightable at all.
| kristopolous wrote:
| I was at a hackathon with this thing last weekend in SF at bfl.
| It's a pretty good system.
| HanClinto wrote:
| What sorts of things were built with it?
| kristopolous wrote:
| I think this should work: https://docs.google.com/spreadsheet
| s/d/1cxh9oA1ZHkzGRMKutVNb...
|
| I was at the top of the list ... pitched it poorly. That
| night I made a party game to practice:
| https://pitchanary.com/
|
| The rules might need some work.
| HanClinto wrote:
| Wow, this is a seriously good turnout for the hackathon.
| Thank you for posting this list, it's fun to look through
| these!
| whatevsmate wrote:
| Neat, I plan to check this out.
|
| I really want an AI to jam with on a canvas rather than to just
| have it generate the final results.
|
| I have been hoping someone would pick up on the time series
| forecasting innovations in the LLM space, combine them with data
| from e.g. the Google quick draw dataset, and turn that into a
| real-time "painting partner" experience, kind of like chatting
| with an LLM through brush strokes.
| vunderba wrote:
| Using the kontext models in Fal.ai shows you a nice slider of
| the before and after edits and has a button that lets you set
| the edited image as the new source so you can continue to make
| changes.
|
| Now that BFL has released a dev model, I'd love to see a
| Kontext plugin for Krita given that it already has one for
| Stable Diffusion though!
|
| https://github.com/Acly/krita-ai-diffusion
| dragonwriter wrote:
| The Krita plugin is a bridge to ComfyUI which can already run
| Flux and presumably will have native support for Kontext
| (dev) within a week or so, and the plugin already has limited
| support for using Flux, so Kontext in the existing plugin
| (rather than requiring a new one) seems a fairly reasonable
| expectation.
| dragonwriter wrote:
| > ComfyUI which can already run Flux and presumably will
| have native support for Kontext (dev) within a week or so
|
| This was pessimistic, native support today, with workflow
| and pointer to an alternate fp8 model download for people
| that can't run the full fp16 checkpoint.
|
| https://comfyanonymous.github.io/ComfyUI_examples/flux/#flu
| x...
| b0a04gl wrote:
| this intent in licenses might lowkey prescreens who even gets to
| build. this creates a soft perimeter = technically open,
| operationally narrow. it's shaping who can even try, plus cuts
| out misuse but also cuts out maybe-use. over time, that subtly
| redefines what counts as valid experimentation
| rushingcreek wrote:
| This is awesome, and kudos to BFL for releasing the weights. The
| financial sustainability of open-source is hard to get right, and
| giving academics this model for free while charging a reasonable
| licensing fee for startups is something I think makes sense if it
| allows BFL and others to continue releasing open-weight models.
| doctorpangloss wrote:
| Would it be financially sustainable if BFL had to pay for
| express permission for all the image and derived-from-video
| content it uses? (No)
| vunderba wrote:
| Here's hoping the distilled [Dev] model can hold up reasonably
| well against the larger pro/max models which in a lot of ways can
| completely replace the relatively old-school inpainting
| techniques of Stable Diffusion.
|
| Some _before /after_ experiments with editing images using
| Kontext:
|
| https://specularrealms.com/ai-transcripts/experiments-with-f...
| thetoon wrote:
| What amount of VRAM is this supposed to work with?
| SV_BubbleTime wrote:
| Today... about 18-20GB.
|
| Tomorrow... like 4GB if you have an hour.
| dragonwriter wrote:
| > Today... about 18-20GB.
|
| There's an FP8 version that's the default for the ComfyUI
| template that's in the release that just came out with
| Kontext support that I've seen reports of running in 12GB or
| less, and which I'm running at this moment in 16GB.
| treesciencebot wrote:
| One interesting feature that gets enabled with open weights is
| adding new capabilities (tasks) to these editing models. They
| generalize quite well with low samples (30 ish). We talk about it
| here https://blog.fal.ai/announcing-flux-1-kontext-dev-
| inference-...
| qingcharles wrote:
| Absolutely. This is the version of Kontext that everyone has
| been waiting for. It's far more useful now. This is the first
| of the new generation of imagegens that allows training. Can't
| do that with Gemini, GPT, MJ etc.
| oTsanony wrote:
| Yo guys, I think I might've found a chill and straightforward way
| to openly generate NSFW stuff using flux1-context on ComfyUI.
| popalchemist wrote:
| License is a major bummer.
___________________________________________________________________
(page generated 2025-06-26 23:01 UTC)