Hacker Newsnew | past | comments | ask | show | jobs | submit | BoredPositron's commentslogin

It's a good model sadly the use of the qwen vae is a bit of a downer.


Krea 2 Large (on the website and api) was trained with the FLUX 2 VAE, if you want to test it out and push realism. After working with both I think the flux VAE has a slight edge in learning realistic textures but it's smaller than you might think, the Qwen VAE was overall very good in ablations and good at learning to produce a diverse set of styles.


You can't be serious. What good is a model if the generations look always slightly out of focus? One easy task if it's as close as you say: Produce one sharp image that is not an illustration.


> You can't be serious.

Please edit out swipes from your HN comments, as the guidelines request: https://news.ycombinator.com/newsguidelines.html.

Edit: your account has unfortunately been breaking the site guidelines like this in other places as well (e.g. https://news.ycombinator.com/item?id=48567675). Can you please fix this? I don't want to ban you, but we've already had to ask you this before.


Kill the account dang.


Definitely encourage you to test the models. We tried to optimize for realistic focus and not over-sharpening, which leads to a "hyper" AI-look. It's hard to benchmark because people generally prefer sharp, saturated orangish pictures all else equal, but I believe these are bad shortcuts for the model to learn realism.


Is my taste the problem, or am I simply holding it wrong? The qwen VAE's shortcomings are well documented, and Krea 2 produces the same blurry, airbrushed output as qwen image. Between the chaotic release and every interaction I've had with your team, I've grown to genuinely dislike your platform/company. Good luck.


It's been mentioned by some that using the wan2.1 vae instead solves this. I haven't personally had time to try yet.


There is a lot of discourse about it on Reddit. Check the AMA link I put in the comment above for learning more. The basics is it wasn’t released when we started and we use it for internal models and hope to do further open source releases.


It's engagement bait and marketing blogposts like this should be banned on HN.


Don't use LTT as a source for anything. They are mainly an entertainment channel with a giant track record of fuck ups anything data related.


Even taking into account all their fuckups majority of their content is accurate and well sourced. I dont know why people do this thing where they point to a few instances and then extrapolate that out.


They literally paused uploading videos because they had an insane amount of wrong data in their videos. It got better for a few months when they decided to slow down video releases but now they are back at it. I dont know why people do this thing where they ignore factual data because they like a creator.


You mean when they just opened their new lab and a massive controversy happened?


The lab is 3 years old and it collapsed one time...


so only 99.9% of their output is accurate? I guess that to inaccurate for a video game review to be trustworthy.



How in the world is a guy named Linus using Windows


Mhm... meteorological summer starts 1st of June. Solstice was yesterday as well.


flux klein with LoRa. GPT image and nano often produce high frequency artifacts when editing.


Both can be right.


They could be, but they really aren't.

Name a single aspect of something modern like the Transformer architecture or how it is trained, that is even indirectly attributable to Schmidhuber.

No doubt he'd be jumping up and down wanting to take credit for residual connections, but where was Schmidhuber in the ImageNet era when everyone else was discovering how to build deep neural nets? Why didn't Schmidhuber invent ResNets, but instead waited until someone else (Kaiming He) did, then claim credit for it?

I'll bet Schmidhuber isn't done with yet ... when someone eventually comes up with an architecture for AGI, Schmidhuber will come out of the woodwork and point to a note he made on a napkin in 1800 that predicted it all.


It's annoying but ok in general to verify age of your users. The problem for me is mainly how and what vendor they use. They picked the worst possibilities.


Bad two weeks for aviation.


Hmm? What else did I miss?


Joshua Baer, founder and CEO of Austin's Capital Factory VC firm, killed in plane crash near Laredo, Texas.

https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...


Oliver Tree is a musician who died in a helicopter crash recently too.


The other replies were big stories, as was this B-52 crash:

https://en.wikipedia.org/wiki/2026_United_States_Air_Force_B...


From here https://www.reddit.com/r/aviation/s/cZE1TCwb2B

June 8: A Gulfstream G200 crashed while landing at La Romana International in the DR, it was on the way to pick up former MLB player Yadier Molina, but it had to make an emergency landing. Both pilots, the only occupants, died.

June 10: A Pakistan Army Aviation Mi-17 helicopter crashed near Muzaffarabad. All 22 on board perished.

June 11: Turkish Airlines Flight TK2430, a Boeing 777-300ER, collided with a ground radar antenna while taxiing to its gate after landing at Antalya Airport when the aircraft turned onto a taxiway too small for 777's. Despite the damage to the aircraft, only 1 of 267 souls aboard were injured.

June 11: United Airlines Flight 1275, a Boeing 737, was delayed after a swarm of bees stuck to its right wing. The pilots decided to take off with the insects still on said wing; they all cleared off once the aircraft got airborne. No bees entered the aircraft and no injuries were reported.

June 11: A Piper PA-36 collided with a tower near Barrelman Airport and crashed. The former president of the National Agricultural Aviation Association (NAAA), Rick Boardman, died.

June 11: A Cirrus SR20 crashed into a house after taking off from Portsmouth, Ohio. Only the pilot was killed, and both occupants of the home escaped uninjured.

June 11: A Zhonglian F-27 light helicopter suffered a tail-boom failure and crashed during a test flight in Jiangsu. The pilot and passenger survived with non life-threatening injuries.

June 12: SriLankan Airlines Flight UL-606, an Airbus A330, had its engine struck by lightning just after takeoff from Colombo International; the plane landed safely and no injuries were reported.

June 13: Singapore Airlines Flight SQ114, a Boeing 737-800, suffered burst tires upon landing at Kuala Lumper International; the plane managed to stop on the runway, and no injuries were reported.

June 13: An Indian Air Force An-32 crashed while attempting to land Jorhat Air Force Base in Assam. 5 of the 6 people on board died, with the co-pilot, as of writing, claimed to have been the only survivor.

June 13: A US Marine Corp F/A-18 Hornet Crashed in Washington State near Rimlock Lake. The pilot safely ejected and survived.

June 14: Two helicopters collided mid-air over Brazil. All six people on board both helicopters perished, including American songwriter and rapper, Oliver Tree.

June 14: A Pacific Aerospace 750XL, N221BN, crashed shortly after takeoff from Butler Memorial Airport in Butler, Missouri. The pilot and 11 skydivers on board perished.

June 14: A Piper PA-28, N15564, crashed shortly after takeoff from Key West, Florida. Initially, 5 people were reported to be on board, but it was found out that only 2 people were; both survived.

June 14: Delta Airlines Flight 2905, a Boeing 737-900, collided with a catering truck at Seattle-Tacoma International; no injuries were reported.

June 15: A Russian Tu-22 bomber crashed in Siberia during a training flight. As of writing, everyone on board has been reported to have ejected and survived.

June 15: A Pakistan Air Force trainer aircraft crashed in Pakistan's Mardan district. As of writing, both of the pilots were killed, and at least three motorists were injured.

June 15: A Van's RV-3A crashed into a cow field about 15 miles north of Lakeland, FL. The pilot was killed.

There were some more in the last three days as well...


I'm not sure if the incident with the bees really belongs....


Looks like a lot. Is it always like that, or were the last two weeks really that special?

One thing that sticks out is that GA (small), but also military, planes' accidents usually end with deaths and big planes accidents usually end with everything being fine.


That might be an instance of Berkson‘s paradox.

You’ll hear about deadly small plane accidents, and about accidents of large planes, but are unlikely to hear about harmless accidents with small planes. Just not interesting. But they happen all the time.

https://en.wikipedia.org/wiki/Berkson%27s_paradox


Wow, thanks a lot! I kind of knew things like that happen (one instance of a thing happening brings attention to other similar things), but didn't know it actually had a name.

Thank you. I really like knowing precise names for things, because it gives me a feeling of correctness and satisfaction.


No, we can't all be friends. Never happened, never will be.


It's engaging and I doubt it happened by accident.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: