• It's a heck of a lot smaller than Qwen-Image 1 (20b parameters) at only 7b, making it one of the smaller open-weight models available (Z-Image Turbo is one of the few that is smaller at 6b) when compared to Ideogram, Krea2, Flux2, etc.
• It supports native transparency (Qwen's team, as far as I know, is the only one attempting to tackle this). Even though it's relatively trivial to set up background removal postprocessors, it's also neat to see it natively supported.
• It's fast using QwenImage2.1 convrot, a 1MP image took around ~5 seconds on an RTX4090.
Negatives
• The license (assuming you respect it) is far more restrictive. The original Qwen Image 1 was released under the standard Apache license; this one explicitly forbids commercial usage without obtaining a separate license. On the other hand, a lot of us didn't expect the Qwen team to ever release "weights-available" ever again.
Qwen-Image 1.0, released about a year ago, only scored 4/15 on my GenAI Showdown Benchmarks. Since that time, they've been upstaged by Krea 2 (6/15) and Ideogram4 (8/15). I'll post the new results once I have some more time to run them.
Well, the results are in, at least for text-to-image (the editing bench will come later).
Qwen-Image 2.1 is definitely a pretty big leap over the last open-weight version, Qwen-Image 1.0, released back in August of last year and managed to score 7 out of 15 as opposed to its predecessor which scored 4 out of 15.
Even though it's significantly smaller, 7b vs 20b, it's multimodal (so you don't need a separate image-to-image model like you did with Qwen-Edit), more coherent, and significantly faster even when outputting at higher 2K resolutions. However, in my testing, I found that I had to play with dialing up the CFG depending on the complexity of the prompt.
I've also added a progress dropdown under Model Performance so you can see how cloud vs. local models have been trending since 2024. Spoiler: June of this year released some of the biggest bangers (Krea 2, Ideogram 4, and the kind of slept-on Boogu-Image 0.1).
Downsides:
- It was clearly trained on at least some level of synthetic training data, and it shows in some of the subpar outputs in terms of fidelity. Some of this you might be able to iron out with a refiner model downstream or a custom LoRA but time will tell.
- They've moved away from the permissive Apache license. Commercial usage is only allowed by request.
This is the fundamental problem with where the AI race is heading, IMHO. Broadly, there are two possible legal interpretations (to my layman's mind):
* A model is derived work of its training data. This seems sane to me. Open, but copyrighted, works (like FOSS) remain protected from abuse. There's some legal moat around AI models. But on the other hand it seems unlikely that there's enough liberally licensed (or public domain) training data to go around. The little guy's status quo remains, the frontier labs' work slows down massively.
* A model is not derived work of its training data. This seems to me insane, but a lot of the world seems to hold this view (including the frontier labs). Stuff like FOSS or indie art is under huge threat of copyrightwashing. But on the other hand, there's also zero legal moat around the models. The little guy is eviscerated, but so are the frontier labs.
Neither interpretation seems, to me, to be capable of sustaining the last couple of years' developments. But what do I know.
> A model is not derived work of its training data. This seems to me insane, but a lot of the world seems to hold this view
Why insane? Models don't take the content as-is, they take measurements. I don't owe you royalties just because I used your photo to get the proportions and coloring of a duck right. Go watch artist streams, you'll often see people to go Google Images for references. I've never seen that result in credit or payments.
The alternative is that we hand out lots of money to a few large companies specializing in content archives, and there's really no benefit to anyone else anyway. On the long term I would expect a few fat cats to get fatter, the small guy to get nothing, and AI still work but get there slowly. I don't see the point or the benefit.
Yeah but here is the thing. Say you train a Lora for a model. You then merge those weights into the open weight model.
Now prompt it for an original image, it will pretty much be able to reproduce that exact image.
You can say it is just measurements but at some point, it can just reproduce with high enough accuracy to just be seen as a copy
Given that corporations buy up any valuable IP, I personally think the answer is to abolish copyright because right now it is really only protecting the rich and corporations . Individuals have the illusion of protection but if Disney steals your shit, good luck with the pain and suffering you experience trying to win a court case against them
> Models don't take the content as-is, they take measurements.
At some point, enough measurements constitute a copy. If I redistribute the average value of all the pixels in your photo, I'm obviously not in violation of your copyright. If I measure and redistribute 90% of its DCT coefficients (i.e. make a slightly compressed JPEG), I am.
The interesting stuff happens between those extremes. We cannot just take as a given that all LLMs always are on the safe side. It is not at all obvious.
If a model is derived work of its training data, surely all the existing frontier models that have been training on copyrighted work have a big legal issue, and therefore so does anything produced with them?
Yes. Of course. They'd have to be rebuilt with acceptably licensed training data. And since there might not be a enough of it, the model owners are screwed.
My point is that they're also screwed in the opposite scenario, because they rely on the same legal protection (against deriving works) as the works they trained on.
That's why I don't understand how any of this can be sustained.
Not slept on at all. It was absolute trash, and I’m super curious why people pretend otherwise. There isn’t a single thing that model did better than any temporal peer.
That benchmark might have some issues. You prompted the models to generate an image of striking a ring against a crucible. Then you (presumably, manually?) scored the images that depicted an anvil higher than the ones striking something resembling a crucible.
That’s a good catch. Yes, all scoring is done through manual review since relying on a VL model for these kinds of meta-metrics is a sort of loose equivalent of gödel's second incompleteness theorem.
I’ll have to think about this one. When I crafted the prompt, I wasn’t really thinking about the differences between a crucible and an anvil. It was more the visual of an archangel smelting halos for newly arrived heavenly beings.
I'm not sure why one would even strike metal against a crucible! It's a container for liquid metal. One of the outputs shows it being smashed by the manoeuvre, which is probably the most realistic outcome of all of them.
Sorry, I'm not trying to nitpick. I'm just joining in because I'm interested in how the models dealt with the request.
Well this is HN - original home of the "ummm actually..." - so I appreciate when people pick all the nits. :)
Even though I prompted for a crucible in the prompt, I think the fact that the prompt also contained terms like “blacksmith” and “hammer,” caused it to lean towards anvils over crucibles in some of the pictures (which as you brought up makes more sense anyway).
Perhaps for the particular image you liked. But choosing something “typical” or average over what it was instructed to do is a massively common failure mode for AI. One that makes the difference between a useful model and one that makes you want to throw your laptop out a window.
I agree. In fact the entire reason I initially built GenAI Showdown was because many of the comparative tests on places like Image Arena Leaderboard [1] are not designed to challenge models on prompt adherence. Even when they are, a considerable number of amateur judges tend to prioritize aesthetics over adherence or instruction-following.
I'll likely be redoing that particular bench with added minimum passing criteria of an anvil.
ive been taking a look there. I think for some of the image to image ones you really ought to use real images as the base prompt, there are a lot of weird things happening where the base image has issues and then its hard to say if the model should be correcting the flaws or not. Things like "childrens drawing app where each crayon is clearly sized to be tapped on" but the base image has the rainbow cut off partway across with only half the red and black crayons visible.
Also "vintage" photography being corrected, where the original is clearly ai generated with unrealistic sharp focus everywhere
I promise I searched on my phone before posting that, but it looks like I misinterpreted an hf search result that was actually Qwen-Image-2512 but the url was truncated to Qwen-Image-2 due to my screen width (ugh). For some reason I was thinking Qwen Image 2.1 was the one that had been available for a while but not released, but you're right that was 2.0.
Native transparency isn’t so hard to do by the way, I made an image AE (I don’t say VAE deliberately as none of these are VAEs, I don’t know why they keep being called that since the variational part is completely absent) that supported this about two years ago as a hobby project. I haven’t really been following the space recently, I’m surprised it’s taken so long for this to come out if it’s a first.
It’s not hard architecturally, but it is hard to find or create good datasets of images on the magnitude you want. I suspect the qwen team heavily used synthetic data for this.
Not very, I had a tiny dataset compared to a production run, I used a bunch of 3D renderings to augment the images with transparency. It worked pretty well given the scale.
Can do! GPT-Image-2 already scored unsurprisingly very high: 12 out of 15 on text-to-image, and 10 out of 12 on image-to-image.
The three benchmarks it failed on (D20, Flat Earth, and Banded Snake) are pretty difficult, so I'd be surprised if 2.5 manages to pass them, but I’ll add it for completeness’ sake later this week.
That's because the users don't care about the licence. They did try to take down a porn Lora repo on hugging face. The trick is, don't say you're using minimax, and you can make porn with it.
Calling open-weights as open-source in marketing materials is the usual misrepresentation. But now with the restriction on commercial use (which is against opensource definition) it is not even open-weights, technically it would be more accurate to call it weights-available.
Yes, having no restrictions on use is the first and most critical software freedom. The others are there more to help protect the first (IMO). GPL in particular tries to reach quite far to protect all of them.
It's not going to matter unless you plan to commercially deploy the model, as far as I see.
If you were to generate outputs for commercial use, I think it would still violate this research license, but it's not like they are going to know, are they?
That said, I am disappointed that the model is not actually open-weights as I expected based on the headline.
It’s not entirely clear far part of a model would be copyrightable in the first place - it’s the result of a mechanical process, where the inputs are someone else’s copyrighted works, after all, while copyright requires some form of creative input.
Usually the way the licenses are interpreted, if I run it myself and use it to generate $X million in revenue in my own business, I'm subject to the license. In this case, the license seems to cover any commercial usage at all, which would certainly include self-hosting for internal business purposes.
Why even bother demanding such terms? Release the weights or don't release the weights, but this is silly. No one is going to pay them to run or host a 7B static image model.
You are not wrong, but will a judge and jury be competent enough to understand the difference after you've spent several hundred thousand in litigation?
Why would you even do that? Just... use it? There hasn't been any legal precedent on if models can even be copyright restricted. Labs just keep publishing license documents as if they matter.
Well, it is an indication that it matters to the lab, so if you don't want legal fees to be the first one to set precedent, then it does matter a great deal.
You could say the same thing about "license-washing" the model. It seems like you're just going through a guaranteed expensive process to have roughly the same risk as just using the model and potentially getting hit with legal fees.
I love the non-commercial clauses because of how many people are using these for deceptive ads and “virtual staging” and fake social media accounts. Anything that makes those guys lives harder while still letting me make silly pictures for my kids and tapestries for my D&D campaign feel fine by me.
I have seen comments on X that they will consider a revenue cap for the non-commercial restriction.
A researcher commented among the lines that they have no interest in restricting creators from using it for monetized YT content.
The way he put it, it suggested they are at this time looking to understand use cases rather than necessarily limit or charge for commercial use, and he recommended reaching out to the commercial team.
So it sounds like one could likely receive a free commercial license if needed.
Boogu-Image has the Apache 2.0 License [1] (good coherence, but outputs can look synthetic).
And Krea 2 has a community license [2] that is fairly permissive - I think commercial usage is allowed under $1 million.
Boogu-Image scored 6/15 and Krea 2 scored 7/15 on my GenAI Showdown benchmark [3] - only Ideogram4 eclipses them in terms of local models, but its got a far more restrictive license and the JSON structured inputs can be a pain to work with.
I run a prompt-to-ui design site that uses image models for the design process[1]. The text rendering especially makes this model deeply interesting to me, despite the license. Here are some tests using my harness comparing the outputs of gpt-image-2 and qwen 2.1:
The text rendering definitely is much, much better than anything else on the open weights market right now. Small text fidelity is quite good. It seems like the text encoder however gets a little bit overloaded with larger prompts - note the presence of hex codes in the design output, those were inputs from the expanded prompt.
I'll be trying a post-training run on this for web design, it has some serious potential.
My harness expands the prompt into a json representation that specifies layout much more rigorously, which is why you see such that amount of alignment between the two.
That internal json backing helps significantly when you want to maintain consistent design system components/patterns across multiple pages. The aligned layout is it working as intended.
Totally normal for modern models due to training on the same datasets supplied by third parties, dataset contamination, and mode collapse, especially for simple prompts that don't have enough semantic capacity. -isms are often very similar even without distillation, and tend to come and go in waves along with model generations.
Agreed. There's also a lot of bad tinging/yellow saturation that very much reminds me of early gpt-image outputs on a lot of the non-cherry picked stuff I've been seeing on Twitter/Reddit.
A lot of people were putting ZiT as a refiner downstream in early Qwen-Image 1.0 workflows, so I'm wondering if we're going to see something similar with 2.1.
> The text rendering definitely is much, much better than anything else on the open weights market right now
Really? Because basically everything in those screenshots is completely garbled. I didn't follow it super closely but I thought Ideogram or whatever was really good for this particular use, with actual clear text.
This is my experience as well. Ideogram4 (assuming you are willing to put in the work to use the proper structured JSON input) is very accurate when it comes to text rendering in an image.
The capabilities of local LLM text-to-image is honestly pretty damn impressive. IMO, I think local image generation is currently ahead of local code generation. I can get an image in seconds locally with the quality being way higher than what I'd expect from a local model. However with coding it's much slower and much less impressive. I'm sure there's a reason for this and I'm not an AI expert so I'll let the smarter folks tell me why, but that's just been my observation thus far.
I've played with diffusion models on and off since the first release of Stable Diffusion - just for amusement, without a particular goal.
Recently, I've been helping a friend's wife with some basic vector images for her sewing hobby (she has what is essentially a CNC sewing machine) and have been super-impressed with FLUX.1-Kontext, which I've been running on my Macbook Pro with mflux. Its ability to (for example) take a photo of a human or an animal and return a line drawing which is recognisably them (rather than just a generic similarish image as I've experienced with other models) is excellent.
It's an older model now, but (AIUI) has the text-handling features baked in, and in my various testing is very reliable at giving me the outputs that I want, without the randomness I've experienced previously. It's big and relatively slow (~3 mins per 512x512 image edit on my M1 Max Mac) but excellent to work with. It's also very straightforward to set up, without the harness complexity of e.g. comfyui.
This is currently the weakest part of the workflow. I'm tracing the output images with VTracer [0] and then finishing them manually.
Through a combination of careful initial prompting (e.g. requesting line drawings, requesting limited colours ["posterization" works] if you need colour at all, requesting background removal) and then using the settings in VTracer to limit the colours used for the resulting SVG to the minimum needed, you can end up with an SVG that doesn't take too long to clean up in Inkscape. It usually takes a few minutes of deletions and path unions to produce something usable.
Diffusion models outputting SVGs would obviously be very interesting for this, but this is probably a small niche compared to everything that diffusion models can do with raster images, and I've not found anything usable online so far. LLMs (understandably, given their core technology) are still pretty bad at creating SVG art.
Claude and I have spent most of the summer on exactly this problem for Skechworks (skechworks.com), a free little Mac design app I made to help my own workflow that is very similar to yours. I use it for my custom coin shop The Achieve Mint (theachievemint.com).
It runs VTracer under the hood too. Most of the cleanup you're doing in Inkscape could possibly be pushed to before and after the trace?
Skechworks picks the palette before the trace and forces it again after. I quantize to 8 colors with pngquant, trace, then snap every fill back to the nearest palette color. VTracer fills each region with its own average color, otherwise you get hundreds of nearly identical grays. One design went from 279 fill colors to 3.
I upscale 2x before I quantize. The anti-aliasing is what tells the tracer where an edge really sits. Quantize at 1x and you throw that away, and small text comes out lumpy. Related gotcha: filter_speckle is an area, so don't double it when you double the image.
I also flatten transparency onto white first. VTracer counts every alpha value as a different color. I had a 5 color picture that it saw as 688 colors, and the trace took 6 GB of memory. Flattened, the same picture took 300 MB.
For the background I sample the whole edge of the picture. If 85% of it is one color, I delete every path in that color that touches the edge. White inside the drawing never touches the edge, so it stays.
For black and white I threshold with Otsu, then trace in color mode with hierarchical stacked instead of bw mode. bw mode hands you one giant black path full of holes. Stacked gives you every region as its own shape, which is much easier to edit. You have to pad the image with the background color first, or VTracer decides your subject is the canvas.
Your FLUX step is the same trick I use for photos and messy art. I have an image model redraw the picture as flat art, then trace the redraw. That came out about a quarter the file size of the best direct trace I could get. The catch is that the model moves and resizes the subject a little every run. I measure the box around the ink in both pictures and fit the trace back onto the original.
What I haven't solved is curve fitting. vectorizer.ai gives you real arcs and straight lines. VTracer gives you splines, so circles are never quite circles.
Off the shelf - it’s a Brother. It prints via a proprietary file format (.PES) but there’s an extension for Inkscape that supports creation and export.
Remember that quality output is a necessary but insufficient property of a generative model.
Prompt-adherence is really hit-or-miss—especially if one lacks the visual vocabulary. Likewise with coding, I find junior devs don't think to prompt re: respecting this-or-that interface, or refactoring to point-free style, etc.
I think there was a lot more brainpower invested in the media generation side of things. The noise-based diffusion technique is further developed. It had a discovery of applying a physics-based understanding of Brownian motion to guide it. Image generation has comparatively simple training process - this is an image with dog, and without dog (contrastive learning).
Text is in a sense way harder to do than images because of radical nonlocality. A word at the start of one paragraph can directly influence the meaning of a word five paragraphs away. Whereas images typically represent the real world, or at least a spatial domain, which gives you a lot of structure 'for free'. If you are drawing a human, you can make a reasonable guess where their hands go in relation to their face. If someone hands you the first half of an essay, finishing it is not trivial.
I've set it up on my local machine just now, as my first local image diffuser. I can confirm it's very easy.
I tried stable-diffusion.cpp, following its compile guide here[0], and its Qwen Image-2.1 specific instructions here[1]. It works out of the box. I made a test pelican[2]. It took 3 minutes on a CPU.
Additional question is what kind of local hardware would be required for this? 7B parameters sounds very light weight, but I'm not sure. (Edit: The download is 33 GB).
Edit x2: As usual I'm in a twisty maze of pip packages that don't work together, with obscure errors about missing modules, even though I followed the instructions on the page to the letter. I really wish people didn't use Python for this stuff. A simple C/C++ program would be so much better.
It's about 16 GiB at Q8 quants (combining both the image and language parts). (Meaning, community quantized models from HuggingFace).
I think it will technically run on anything that has enough memory. I just tried it on a standard laptop (dual-channel DDR5), and it took about 3 minutes for a 512x512. If you'd want to run it at interactive speeds, you would want a GPU (one which fits this in VRAM).
> "I really wish people didn't use Python for this stuff. A simple C/C++ program would be so much better."
There is difussion.cpp which is intended for those types of models. I set up krea-2-turbo with the help of ChatGPT 2 months ago, if you have a capable computer that's what I would suggest once it becomes supported.
I am on AI max 395, comfyUI+qwen models is all you technically need. With today's release, I just built a quick and dirty html that allow simpler prompt use and edits ( via headless comfyui ).. its not bad for a day's work, but a little too unpolished to publish. I would say, try comfyUI first ( complex, but it worked OOTB ).
> Diffusers, ComfyUI, vLLM-Omni, SGLang, and LightX2V
I think that's all Python (not a direct executable).
You could just do (see the "Quick Start") four `pip install` and have a dozen lines script to generate the image. But `llama.cpp` and similar do not require e.g. installing Torch (or PyTorch) - you can use `llama.cpp` on a non-specialized machine.
I use opencode + <a decent saas llm> to set up all this new ai generation stuff. GLM-5.3 is my current gun. Safely inside podman containers too because I dont trust this fast moving python eco system at all. Never do I want this running on my main OS.
I have FLUX.2 klein and dev, Ideogram, LaDA-Image and SenseNova locally. Works great. Ive never touched a file.
The days of making container yamls myself is over. I read them but I dont edit anymore.
Probably ComfyUI is one of the easiest way to get started with local image/video models. Or perhaps vLLM, if they have support for it already, would be something like `vllm serve <model> --omni --port 9080`
Seems I'm missing something. Does this model support other inputs?
Image outputs are supported, videos I'm not sure but I don't think that's an output, just a preview of the equirectangular example, so, same question here, what does this model outputs that isn't supported?
Not all architectures are supported by llama.cpp . The GGUF format encodes the NN in a standardized way, but then you need code that can use that NN structure.
I understand that llama.cpp could only output text, last time I checked (I do not know how to find a good source for that though).
I haven't used llama.cpp for image generation either but I recall an issue about it. Unfortunately I can't pinpoint it now and there is the older closed issue https://github.com/ggml-org/llama.cpp/issues/4408 so unless mtmd supports also multimodal outputs out of the box safe to assume output is still limited to text.
I am really grateful to the Chinese Labs for open sourcing their best models. If it was left to the Americans, we would be forced to pay obscene API fees to use them.
> You shall not use the Materials for any commercial purpose without obtaining a separate commercial license from us.
It probably will be much cheaper to use than other image models, but it seems that will be up to the whims of Qwen/Alibaba rather than just being the cost of putting it in a cloud provider.
Honestly, that's fine. The commercial license isn't that bad, and cloud providers selling API access to this can afford it.
I am just happy I can run these models on my own hardware. Hopefully in 10 years, self-hosted models far exceeding what's currently available will run comfortable on commodity hardware.
Just think about how recently we got that feature in the official ChatGPT image gen. And now we have that running locally — assuming that is, I can figure out how to get this running on my Mac — blows my mind.
I want to say that they released a version since then before 2.5, but I'm not entirely sure. I should also note that I do not know a single Chinese character, so it's possible they are broken in such a way that I wouldn't necessarily notice. I do remember, however, that English text used to be broken; now it's completely readable.
The minute you start looking at other non-frontier models, you understand why Dario, Altman, and Musk are coming together to say we need regulation.
None are as good yet, but what everyone said is coming true - models are not moats. And these folks need an exit (even Msuk whose shares are still locked)
latents go from 16ch @ 8x compression to 64ch @ 16x, so roughly the same total latent budget but much more channel heavy. It’s also deeper/wider, and the old 2x2 transformer patching is gone.
On some images it still produces artifacts but can't say if it's the transformer or the VAE yet.
Its happy to see a new open image model from qwen. But the license is a let down. And it dosent even beat their closed qwen3 image wich is already a bit old.
Anyone get success editing videos, frame by frame, using image editing AI? Which model works well?
In my experience, video models generate videos pretty well but are mid at editing. They actually regenerate the entire video along with the edit. So these models being non-deterministic tweak the rest of the video as well, the parts you hoped would be left not edited. It gets exponentially worse when there are humans in the videos, annoying face distortions and for some reason these models just don't understand fingers.
My first impression is that it's not so good at following prompt directions. I asked it to place a 3D text made of glass in a particular city. It instead gave me a broken 3D text on a white background. Maybe with different seeds it gets better, but it's more of a trial and error process than reliable results.
You could try attaching other images as references (I think you can attach a maximum of 10 images). If the attachments can be blurred or sketchy or generic enough, they could be used for generalization.
Qwen and Alibaba are the biggest competitor for basically every model out there. They're beating the benchmarks like top-frontier models, focused on open-source and much cheaper than the competitors.
I'm actually looking for a model that is able to create old school pixel art (like 90s style, games like Sierra and so on). But to this date, I haven't seen anything yet. It all reeks of "AI slop". Maybe this is a good thing, I don't know. But if anyone has any tips, I'd appreciate. It's for my own personal use.
Not sure there's a better avatar for the absurdity of AI slop imagery than the "cowboy on horseback". That's a pony with a child's saddle on it, and they've composited a grown man on top of it.
People say ChatGPT generates images with a yellow tint. The person you are replying to is suggesting that these images have a yellow tint and therefore this model is distilled from ChatGPT.
Positives
• It's a heck of a lot smaller than Qwen-Image 1 (20b parameters) at only 7b, making it one of the smaller open-weight models available (Z-Image Turbo is one of the few that is smaller at 6b) when compared to Ideogram, Krea2, Flux2, etc.
• It supports native transparency (Qwen's team, as far as I know, is the only one attempting to tackle this). Even though it's relatively trivial to set up background removal postprocessors, it's also neat to see it natively supported.
• It's fast using QwenImage2.1 convrot, a 1MP image took around ~5 seconds on an RTX4090.
Negatives
• The license (assuming you respect it) is far more restrictive. The original Qwen Image 1 was released under the standard Apache license; this one explicitly forbids commercial usage without obtaining a separate license. On the other hand, a lot of us didn't expect the Qwen team to ever release "weights-available" ever again.
Qwen-Image 1.0, released about a year ago, only scored 4/15 on my GenAI Showdown Benchmarks. Since that time, they've been upstaged by Krea 2 (6/15) and Ideogram4 (8/15). I'll post the new results once I have some more time to run them.
https://genai-showdown.specr.net
Qwen-Image 2.1 is definitely a pretty big leap over the last open-weight version, Qwen-Image 1.0, released back in August of last year and managed to score 7 out of 15 as opposed to its predecessor which scored 4 out of 15.
Even though it's significantly smaller, 7b vs 20b, it's multimodal (so you don't need a separate image-to-image model like you did with Qwen-Edit), more coherent, and significantly faster even when outputting at higher 2K resolutions. However, in my testing, I found that I had to play with dialing up the CFG depending on the complexity of the prompt.
I've also added a progress dropdown under Model Performance so you can see how cloud vs. local models have been trending since 2024. Spoiler: June of this year released some of the biggest bangers (Krea 2, Ideogram 4, and the kind of slept-on Boogu-Image 0.1).
Downsides:
- It was clearly trained on at least some level of synthetic training data, and it shows in some of the subpar outputs in terms of fidelity. Some of this you might be able to iron out with a refiner model downstream or a custom LoRA but time will tell.
- They've moved away from the permissive Apache license. Commercial usage is only allowed by request.
Comparisons:
https://genai-showdown.specr.net
If you just want to compare local models only:
http://genai-showdown.specr.net/?models=local
* A model is derived work of its training data. This seems sane to me. Open, but copyrighted, works (like FOSS) remain protected from abuse. There's some legal moat around AI models. But on the other hand it seems unlikely that there's enough liberally licensed (or public domain) training data to go around. The little guy's status quo remains, the frontier labs' work slows down massively.
* A model is not derived work of its training data. This seems to me insane, but a lot of the world seems to hold this view (including the frontier labs). Stuff like FOSS or indie art is under huge threat of copyrightwashing. But on the other hand, there's also zero legal moat around the models. The little guy is eviscerated, but so are the frontier labs.
Neither interpretation seems, to me, to be capable of sustaining the last couple of years' developments. But what do I know.
Why insane? Models don't take the content as-is, they take measurements. I don't owe you royalties just because I used your photo to get the proportions and coloring of a duck right. Go watch artist streams, you'll often see people to go Google Images for references. I've never seen that result in credit or payments.
The alternative is that we hand out lots of money to a few large companies specializing in content archives, and there's really no benefit to anyone else anyway. On the long term I would expect a few fat cats to get fatter, the small guy to get nothing, and AI still work but get there slowly. I don't see the point or the benefit.
Now prompt it for an original image, it will pretty much be able to reproduce that exact image.
You can say it is just measurements but at some point, it can just reproduce with high enough accuracy to just be seen as a copy
Given that corporations buy up any valuable IP, I personally think the answer is to abolish copyright because right now it is really only protecting the rich and corporations . Individuals have the illusion of protection but if Disney steals your shit, good luck with the pain and suffering you experience trying to win a court case against them
At some point, enough measurements constitute a copy. If I redistribute the average value of all the pixels in your photo, I'm obviously not in violation of your copyright. If I measure and redistribute 90% of its DCT coefficients (i.e. make a slightly compressed JPEG), I am.
The interesting stuff happens between those extremes. We cannot just take as a given that all LLMs always are on the safe side. It is not at all obvious.
My point is that they're also screwed in the opposite scenario, because they rely on the same legal protection (against deriving works) as the works they trained on.
That's why I don't understand how any of this can be sustained.
Not slept on at all. It was absolute trash, and I’m super curious why people pretend otherwise. There isn’t a single thing that model did better than any temporal peer.
I’ll have to think about this one. When I crafted the prompt, I wasn’t really thinking about the differences between a crucible and an anvil. It was more the visual of an archangel smelting halos for newly arrived heavenly beings.
Sorry, I'm not trying to nitpick. I'm just joining in because I'm interested in how the models dealt with the request.
Even though I prompted for a crucible in the prompt, I think the fact that the prompt also contained terms like “blacksmith” and “hammer,” caused it to lean towards anvils over crucibles in some of the pictures (which as you brought up makes more sense anyway).
I'll likely be redoing that particular bench with added minimum passing criteria of an anvil.
[1] - https://huggingface.co/spaces/ArtificialAnalysis/Text-to-Ima...
Also "vintage" photography being corrected, where the original is clearly ai generated with unrealistic sharp focus everywhere
Edit: This is wrong.
The three benchmarks it failed on (D20, Flat Earth, and Banded Snake) are pretty difficult, so I'd be surprised if 2.5 manages to pass them, but I’ll add it for completeness’ sake later this week.
There is no single opinion, and clearly no single Chinese approach.
Also Chinese labs are in particular very careful about anything which be used to create pornographic content, which is highly illegal in the PRC.
MiniMax H3 is dominating AI porn right now, and they are Chinese.
https://en.wikipedia.org/wiki/Qwen#List_of_models
Unfortunately, it looks like this model is using a much more restrictive license:
https://github.com/QwenLM/Qwen-Image-2.1/blob/main/LICENSE
It's just freeware.
If AI labs get to ignore licenses, so do we.
Image data?
> GPL licensed instead and use it however the fuck I want
GPL is not a "use it however the fuck I want" license. Maybe you're thinking of the WTFPL?
No. GPL is a "use it however the fuck you want" license. The GPL has no restrictions on *use*, only on redistribution.
https://en.wikipedia.org/wiki/The_Free_Software_Definition
If you were to generate outputs for commercial use, I think it would still violate this research license, but it's not like they are going to know, are they?
That said, I am disappointed that the model is not actually open-weights as I expected based on the headline.
It's not going to matter then, either. What are they going to do, sue me for copyright infringement?
Why even bother demanding such terms? Release the weights or don't release the weights, but this is silly. No one is going to pay them to run or host a 7B static image model.
It’s more of something to scare companies with legal teams. If you’re an individual or hobbyist doing a side project the risk is essentially zero.
How difficult would it be to use this model to create a second model without licensing issues?
People can continue to use closed SOTA models to generate outputs for commercial or malicious purposes.
What this research license achieves is that we cannot use this model in applications we publish.
A researcher commented among the lines that they have no interest in restricting creators from using it for monetized YT content.
The way he put it, it suggested they are at this time looking to understand use cases rather than necessarily limit or charge for commercial use, and he recommended reaching out to the commercial team.
So it sounds like one could likely receive a free commercial license if needed.
And Krea 2 has a community license [2] that is fairly permissive - I think commercial usage is allowed under $1 million.
Boogu-Image scored 6/15 and Krea 2 scored 7/15 on my GenAI Showdown benchmark [3] - only Ideogram4 eclipses them in terms of local models, but its got a far more restrictive license and the JSON structured inputs can be a pain to work with.
[1] - https://github.com/Boogu-Project/Boogu-Image
[2] - https://www.krea.ai/krea-2-licensing
[3] - https://genai-showdown.specr.net/?models=fd,hd,kd,qi,f2d,zt,...
https://html.non.io/qwen-comparison/
The text rendering definitely is much, much better than anything else on the open weights market right now. Small text fidelity is quite good. It seems like the text encoder however gets a little bit overloaded with larger prompts - note the presence of hex codes in the design output, those were inputs from the expanded prompt.
I'll be trying a post-training run on this for web design, it has some serious potential.
[1] diffui.ai
That internal json backing helps significantly when you want to maintain consistent design system components/patterns across multiple pages. The aligned layout is it working as intended.
Even the artifacts are getting picked up.
A lot of people were putting ZiT as a refiner downstream in early Qwen-Image 1.0 workflows, so I'm wondering if we're going to see something similar with 2.1.
Really? Because basically everything in those screenshots is completely garbled. I didn't follow it super closely but I thought Ideogram or whatever was really good for this particular use, with actual clear text.
Recently, I've been helping a friend's wife with some basic vector images for her sewing hobby (she has what is essentially a CNC sewing machine) and have been super-impressed with FLUX.1-Kontext, which I've been running on my Macbook Pro with mflux. Its ability to (for example) take a photo of a human or an animal and return a line drawing which is recognisably them (rather than just a generic similarish image as I've experienced with other models) is excellent.
It's an older model now, but (AIUI) has the text-handling features baked in, and in my various testing is very reliable at giving me the outputs that I want, without the randomness I've experienced previously. It's big and relatively slow (~3 mins per 512x512 image edit on my M1 Max Mac) but excellent to work with. It's also very straightforward to set up, without the harness complexity of e.g. comfyui.
Through a combination of careful initial prompting (e.g. requesting line drawings, requesting limited colours ["posterization" works] if you need colour at all, requesting background removal) and then using the settings in VTracer to limit the colours used for the resulting SVG to the minimum needed, you can end up with an SVG that doesn't take too long to clean up in Inkscape. It usually takes a few minutes of deletions and path unions to produce something usable.
Diffusion models outputting SVGs would obviously be very interesting for this, but this is probably a small niche compared to everything that diffusion models can do with raster images, and I've not found anything usable online so far. LLMs (understandably, given their core technology) are still pretty bad at creating SVG art.
[0] https://github.com/visioncortex/vtracer
It runs VTracer under the hood too. Most of the cleanup you're doing in Inkscape could possibly be pushed to before and after the trace?
Skechworks picks the palette before the trace and forces it again after. I quantize to 8 colors with pngquant, trace, then snap every fill back to the nearest palette color. VTracer fills each region with its own average color, otherwise you get hundreds of nearly identical grays. One design went from 279 fill colors to 3.
I upscale 2x before I quantize. The anti-aliasing is what tells the tracer where an edge really sits. Quantize at 1x and you throw that away, and small text comes out lumpy. Related gotcha: filter_speckle is an area, so don't double it when you double the image.
I also flatten transparency onto white first. VTracer counts every alpha value as a different color. I had a 5 color picture that it saw as 688 colors, and the trace took 6 GB of memory. Flattened, the same picture took 300 MB.
For the background I sample the whole edge of the picture. If 85% of it is one color, I delete every path in that color that touches the edge. White inside the drawing never touches the edge, so it stays.
For black and white I threshold with Otsu, then trace in color mode with hierarchical stacked instead of bw mode. bw mode hands you one giant black path full of holes. Stacked gives you every region as its own shape, which is much easier to edit. You have to pad the image with the background color first, or VTracer decides your subject is the canvas.
Your FLUX step is the same trick I use for photos and messy art. I have an image model redraw the picture as flat art, then trace the redraw. That came out about a quarter the file size of the best direct trace I could get. The catch is that the model moves and resizes the subject a little every run. I measure the box around the ink in both pictures and fit the trace back onto the original.
What I haven't solved is curve fitting. vectorizer.ai gives you real arcs and straight lines. VTracer gives you splines, so circles are never quite circles.
Prompt-adherence is really hit-or-miss—especially if one lacks the visual vocabulary. Likewise with coding, I find junior devs don't think to prompt re: respecting this-or-that interface, or refactoring to point-free style, etc.
So, as others have said, the artist knows better.
Might be worth to watch the diffusion based LLMs.
(I mean: outside direct or substantial use of Python, and running the Neural Network in the most efficient way.)
I tried stable-diffusion.cpp, following its compile guide here[0], and its Qwen Image-2.1 specific instructions here[1]. It works out of the box. I made a test pelican[2]. It took 3 minutes on a CPU.
[0] https://github.com/leejet/stable-diffusion.cpp/blob/master/d...
[1] https://github.com/leejet/stable-diffusion.cpp/blob/master/d...
[2] https://i.ibb.co/yMknC2K/output.png
In fact, like it appears in the reports above, it is "7b" as in
> 7B parameters in its visual generation component
It seems they calibrated the size to fill a 16GB VRAM near the limit. RAM requirements will vary.
Edit x2: As usual I'm in a twisty maze of pip packages that don't work together, with obscure errors about missing modules, even though I followed the instructions on the page to the letter. I really wish people didn't use Python for this stuff. A simple C/C++ program would be so much better.
I think it will technically run on anything that has enough memory. I just tried it on a standard laptop (dual-channel DDR5), and it took about 3 minutes for a 512x512. If you'd want to run it at interactive speeds, you would want a GPU (one which fits this in VRAM).
> "I really wish people didn't use Python for this stuff. A simple C/C++ program would be so much better."
You mean besides stable-diffusion.cpp ?
Yes, thanks, I didn't know about that. Will try it.
it already has day-0 qwen image 2.1 support!
I think that's all Python (not a direct executable).
You could just do (see the "Quick Start") four `pip install` and have a dozen lines script to generate the image. But `llama.cpp` and similar do not require e.g. installing Torch (or PyTorch) - you can use `llama.cpp` on a non-specialized machine.
I don't think I have ever once run "pip install transformers" and had it work without three rounds of fiddling
Yep, that's (also) what I meant ;)
Lean, efficient... Also sensible and trouble-less.
I have FLUX.2 klein and dev, Ideogram, LaDA-Image and SenseNova locally. Works great. Ive never touched a file.
The days of making container yamls myself is over. I read them but I dont edit anymore.
> Currently, we support image, audio and video input.
Image outputs are supported, videos I'm not sure but I don't think that's an output, just a preview of the equirectangular example, so, same question here, what does this model outputs that isn't supported?
I understand that llama.cpp could only output text, last time I checked (I do not know how to find a good source for that though).
See https://github.com/ggml-org/llama.cpp/blob/master/src/llama-... , the
...At the risk of stating the obvious llama.cpp isn't just about LLaMa as https://github.com/ggml-org/llama.cpp/blob/master/src/llama-... someone else pointed out.
> You shall not use the Materials for any commercial purpose without obtaining a separate commercial license from us.
It probably will be much cheaper to use than other image models, but it seems that will be up to the whims of Qwen/Alibaba rather than just being the cost of putting it in a cloud provider.
https://github.com/QwenLM/Qwen-Image-2.1/blob/main/LICENSE
I am just happy I can run these models on my own hardware. Hopefully in 10 years, self-hosted models far exceeding what's currently available will run comfortable on commodity hardware.
Back in Apr, ChatGPT Images 2.0 has some broken Chinese texts in its featured examples, and they later removed that from blog post. Is 2.5 better now?
https://formulae.brew.sh/cask/comfy
None are as good yet, but what everyone said is coming true - models are not moats. And these folks need an exit (even Msuk whose shares are still locked)
EDIT: It still produces artifacts it's better but unusable for production work. In midvalues you will see a slight dot pattern.
Is this not simply some sort of watermark instead of an artifact?
Can you share the sources?
latents go from 16ch @ 8x compression to 64ch @ 16x, so roughly the same total latent budget but much more channel heavy. It’s also deeper/wider, and the old 2x2 transformer patching is gone.
On some images it still produces artifacts but can't say if it's the transformer or the VAE yet.
In my experience, video models generate videos pretty well but are mid at editing. They actually regenerate the entire video along with the edit. So these models being non-deterministic tweak the rest of the video as well, the parts you hoped would be left not edited. It gets exponentially worse when there are humans in the videos, annoying face distortions and for some reason these models just don't understand fingers.
Excited to see what the future holds for them!
(https://www.shutterstock.com/blog/shirley-card-racial-photog...).
(Or does this model has those capabilities natively ?)
I know a few friends of mine who are running models and are ignoring the licence.
Whether it is AGPL 3.0, or a completely restrictive license, it is going to get broken anyway and be used for commercial purposes.
I don't know anyone who looks at the licenses of the OSS software they are using.
In today’s world OSS is synonymous with "Free" and the AI model providers are proof of that with their training of code, datasets, etc.
So it begs the question, why should we abide by their licenses of their models?
Code on github, models on huggingface, nice intro text: "We are excited to open-source Qwen-Image-2.1 [...]".
meh...