https://www.xatakaon.com/robotics-and-ai/the-compact-version-of-stable-diffusion-3-is-here-and-its-generating-monstrous-human-bodies Xataka On * Menu * Latest * Tech * Reviews * Buying Guides * Science * Space * Entertainment * Twitter * Facebook TRENDING * iOS * Google * AI * Elon Musk * Amazon * ChatGPT * Windows * Boeing The 'Compact' Version of Stable Diffusion 3 Is Here. And It's Generating Monstrous Human Bodies * The most recent version of this generative AI model looked promising, but people are mocking its newest results. * The company's situation has recently become complicated , paving the way for DALL-3 3 and Midjourney. Monstrous human bodies generated by the 'compact' version of Stable Diffusion 3 No comments Twitter Flipboard E-mail 2024-06-13T21:28:02Z javier-pastor Javier Pastor 34 articles by Javier Pastor Stable Diffusion is Midjourney's great rival in AI image generators, mainly because it allows anyone to use it on their PC and power it with all kinds of external components. Months ago, its developers released Stable Diffusion 3, and now they're introducing a new compact version called Stable Diffusion Medium. But there's a problem: It creates monsters. What happened? Although its developers released Stable Diffusion 3 on February 22, and the public API has been available since April 17, they've just introduced Stable Diffusion Medium. This more "compact" version can run smoothly on any PC with a powerful GPU. OpenAI Has Just Undertaken a Challenging and Ambitious Journey: Understanding AI's 'Black Box' More from Xataka On OpenAI Has Just Undertaken a Challenging and Ambitious Journey: Understanding AI's 'Black Box' All you need is a GPU with at least 5GB of memory. While SD3 Large (the original) has 8 billion parameters, SD3 Medium has 2 billion. Christian Laforte, co-CEO of Stabilit, said, "Unlike SD3 Large, SD3 Medium is smaller and can run efficiently on commodity hardware." The developers explain that the minimum requirement is a GPU with 5GB of graphics memory. However, they recommend one with 16GB for best results and performance. Stable Diffusion 3 Medium is also available for free via Hugging Face. A disturbing image of a woman with four fingers on each hand. A disturbing image of a woman with four fingers on each hand. Not those hands, please. | f1-f2-f3-f4-/Reddit The model is promising. This version benefits from all the improvements of the large model. Thus, it offers a higher degree of photorealism in the generated images, much better typography supports due to the Diffusion Transformer architecture, a better understanding of complex prompts, and perfect efficiency when running on "consumer" GPUs. Image generated by Stable Diffusion 3 Image generated by Stable Diffusion 3 Image | HornyMetalBeing/Reddit But it's generating abhorrent bodies. However, the model's limitations are apparent, as evidenced by some images users share publicly. As explained by Ars Technica, the model's work has popped up on Reddit's threads, where users ridicule SD3 Medium and criticize the monstrous pictures of human bodies. Images generated by Stable Diffusion 3 Images generated by Stable Diffusion 3 Image | Bryandroid98/Reddit The hand issues are the least of the worries. The problems appear, for example, in images created by users with simple prompts of women lying on the grass or in the water. This model also has hand problems that seemed to be a thing of the past. But again, generally, fantastic images end up horrible because of how this AI represents these extremities. Image generated by Stable Diffusion 3 Image generated by Stable Diffusion 3 Image | ThereforeGames/Reddit A setback in the fight against Midjourney. These problems are a setback for Stable Diffusion, which users saw as Midjourney's main competitor alongside DALL-E 3. One Reddit user joked, "At least our [training] datasets are safe and ethical!" pointing out that Midjourney's training is unknown and allegedly uses copyrighted images. Censorship is a possible cause. The creation of such anomalous images may be due to Stability AI's insistence on censoring adult content from SD3's training data. This data teaches the model how to generate pictures and it's an essential source of information for the AI to learn about human anatomy. However, withholding this data makes the model fail to understand these requests and generate absurd and disturbing images. Something similar happened with Stable Diffusion 2.0 in 2022. The company fixed the problem with SD 2.1 and SD XL. Internal problems at Stability AI. The company's situation hasn't been the best lately. A year ago, Getty sued it, which probably affected its development. CEO and founder Emad Mostaque resigned in March, followed by three of Stability AI's most influential engineers. The company laid off 10% of its workforce in April. Its financial situation hasn't looked good in recent months, complicating its future. Images | Reddit Related | AI Services Are Everywhere, Ushering in the Return of Subscription Fatigue Twitter Flipboard E-mail Topics * Robotics and AI * Artificial Intelligence * Stable Diffusion * Midjourney Los mejores comentarios: View 0 comments Popular Topics * Best Wireless Headphones * Best PC Games * Best Sci-fi Books * Best e-Readers * Best VR Headsets * IPad Pro (2024) Review * Google Pixel 8 Pro Review Follow us * Twitter * Facebook * RSS At Xataka On, we cover: * Reviews * Energy * Space * Smartphones * Aeronautics * Basics * Apple * Android * AI * Samsung * China * Work See more topics Return to top Webedia Technology * Xataka * Xataka Movil * Xataka Android * Xataka Smart Home * Xataka On * Applesfera LATAM * Xataka Mexico * Xataka Colombia Home o Index * Reviews * Energy * Space * Smartphones * Aeronautics * Basics * Apple * Android * AI * Samsung * China * Work See more topics Follow us * Twitter * Facebook * RSS Receive "Xatakaletter", our weekly newsletter [ ] Subscribe By subscribing, you agree to our privacy policy Error: Email is not in the correct format * Editorial team * Contact us * * Legal Notice * Terms of use * Cookie Policy * Advertise with us Home o Index Latest * Microsoft's Agreement With OpenAI Was Only the First Step. What It (Probably) Wants Is to Get Rid of the AI Company * The ISS Said an Astronaut Was in 'Critical Condition' on Its Radio. It Forgot to Say It Was a Drill * This Is How One of the Most Famous Windows Screensavers, the Hypnotic Network of Moving 3D Pipes, Was Born * When Using ChatGPT at Work, We May Be Exposing Confidential Information: Here's What You Can Do to Avoid It * Boeing Stated That Starliner Could Safely Withstand Up to Five Helium Leaks. It Just Found the Fifth One * Google Pixels Have Received One of the Best Updates Ever, Which Also Fixes Critical Bugs * An Amazon Tribe That Recently Gained Access to the Internet Has Been Accused Internationally of Being Addicted to Porn. It's Not True * After the Latest Launch, Elon Musk Is Convinced That Starship Will Reach Mars in 3 Years. How Realistic Is It? * 'I Have 3 Years Left of Work': More AI Managers Believe AI Will Eventually Take Their Jobs * From Startup to French National Gem in 12 Months: This Is the Meteoric Rise of Mistral, Europe's Biggest Hope to Break Into AI * The First Diamond Solar Panels Have Introduced a Material That Surpasses the Limitations of Photovoltaic Systems * Starship Has 9 Months to Complete NASA's Lunar Contract. Against All Odds, SpaceX Believes It's Doable * South Korea Thought Its 'Jeonse' Were the Magic Formula for the Rental Market, Until the Scams Began * HP CEO Provides Details on the Significant Decline in Printing Since the Pandemic: It's Bad News for the Company's Business * Understanding Apple Intelligence, Apple's Proprietary AI Model: Here's How It Compares to Other Models and How It Will Use ChatGPT * I've Tried iOS 18. On One Hand, I'm Very Excited. On the Other, I'm Very Confused See more articles