# Stable Diffusion Is Getting Outrageously Good! ## Метаданные - **Канал:** Two Minute Papers - **YouTube:** https://www.youtube.com/watch?v=rzKVInXep5M - **Дата:** 27.04.2023 - **Длительность:** 6:22 - **Просмотры:** 190,877 ## Описание ❤️ Check out Fully Connected by Weights & Biases: https://wandb.me/papers W&B+Stable Diffusion: https://wandb.ai/capecape/stable_diffusions/reports/Speed-Up-Stable-Diffusion-on-Your-M1Pro-Macbook-Pro--VmlldzoyNjY0ODYz 📝 The paper "High-Resolution Image Synthesis with Latent Diffusion Models" is available here: https://arxiv.org/abs/2112.10752 Try it: Web 1: https://huggingface.co/spaces/stabilityai/stable-diffusion Web 2: https://beta.dreamstudio.ai/generate Web 3 (also Stable Diffusion XL!): https://clipdrop.co/stable-diffusion Web 4 (notebooks): https://github.com/TheLastBen/fast-stable-diffusion Guide: https://stable-diffusion-art.com/know-these-important-parameters-for-stunning-ai-images/#Sampling_methods Draw Things app: https://drawthings.ai/ Stable Diffusion Web UI: https://github.com/AUTOMATIC1111/stable-diffusion-webui Photoshop integration: http://stable.art Sources: Video https://twitter.com/dreamwieber/status/1618453304970997762 Photorealistic image: https://twitter.com/DiffusionPics/status/1619444407937241089 Realistic vision: https://civitai.com/models/4201?modelVersionId=29461 Infinite zoom: https://twitter.com/hardmaru/status/1612134809924685825 Tiled texture: https://stackoverflow.com/questions/24319825/texture-tiling-with-continuous-random-offset Stable.art (Photoshop): https://github.com/isekaidev/stable.art Wand - drawing: https://twitter.com/wand_app/status/1604186054923210752 Texturing: https://twitter.com/CarsonKatri/status/1600248599254007810 + https://twitter.com/CarsonKatri/status/1603419328019169280 AR + assistant: https://twitter.com/StrangeNative/status/1569700294673702912 Metahumans: https://twitter.com/CoffeeVectors/status/1569416470332858372 My latest paper on simulations that look almost like reality is available for free here: https://rdcu.be/cWPfD Or this is the orig. Nature Physics link with clickable citations: https://www.nature.com/articles/s41567-022-01788-5 🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible: Aleksandr Mashrabov, Alex Balfanz, Alex Haro, Andrew Melnychuk, Benji Rabhan, Bryan Learn, B Shang, Christian Ahlin, Eric Martel, Geronimo Moralez, Gordon Child, Jace O'Brien, Jack Lukic, John Le, Jonathan, Kenneth Davis, Klaus Busse, Kyle Davis, Lorin Atzberger, Lukas Biewald, Martin, Matthew Valle, Michael Albrecht, Michael Tedder, Nikhil Velpanur, Owen Campbell-Moore, Owen Skarpness, Rajarshi Nigam, Ramsey Elbasheer, Richard Sundvall, Steef, Taras Bobrovytsky, Ted Johnson, Thomas Krcmar, Timothy Sum Hon Mun, Torsten Reil, Tybie Fitzhugh, Ueli Gallizzi. If you wish to appear here or pick up other perks, click here: https://www.patreon.com/TwoMinutePapers Thumbnail background design: Felícia Zsolnai-Fehér - http://felicia.hu Károly Zsolnai-Fehér's links: Twitter: https://twitter.com/twominutepapers Web: https://cg.tuwien.ac.at/~zsolnai/ ## Содержание ### [0:00](https://www.youtube.com/watch?v=rzKVInXep5M) Segment 1 (00:00 - 05:00) Dear Fellow Scholars, this is Two Minute  Papers with Dr. Károly Zsolnai-Fehér. Today we are going to marvel at the amazing things  that happen when a paper like Stable Diffusion   comes alive as a free and open text to image AI.   To say that it is improving at a rapid pace would   be an understatement, and it gets even better,  it also gives you the warm and fuzzy feeling of   running it yourself on your own hardware. Here  are 8 amazing, non-trivial applications of it. One, it can be combined with several other  tools to create something magical. Here,   you see a ChatGPT prompt, then, a request to a  different text to image AI to create the images,   finally, get this, these were combined into a  video by Stable Diffusion. Lovely work. So cool! Two, this incredible photorealistic  image was created by a custom model.    These are typically specialized models  that are fine-tuned to do particularly   well at a given topic. This one is  for photorealism, but there are tons   of other ones out there as well that make  Stable Diffusion an absolute powerhouse. Three, we can also create a set of images,   and weave them together into one of those  famous zooming videos. Depending on how we   choose the start and endpoints, we can even make  gorgeous infinitely zooming videos. Loving it. Four, I hear you asking, okay Károly, but  I already work with existing tools like   Photoshop. Can I make Stable Diffusion work with  it seamlessly? And the answer is yes! Finally!    Look at that! You can now write your prompts,  and even remix your real or generated images   by inpainting them. This combines the powerful  built-in tools of Photoshop with text to image.    I absolutely love this. Note that this requires a  little fiddling, but all the information you need   is there in the video description if you wish  to give it a try. This is going to change how   artists use Photoshop and make them a lot  more productive. What a time to be alive! And five, if you do not wish to use it in  photoshop, it can still supercharge your   creativity if you are looking to draw only  a rough outline for your ideas and fill them   in through text to image. This gives you  even more granular artistic control over   the generated images. I think applications  like this are fantastic because they put the   human in charge much more than just generating an  image from thin air using a text prompt. So good! Six, okay, but all of these images are 2D. What  about 3D artists, can they also benefit from   Stable Diffusion somehow? Does it help them in  any way? Well, look at that! Wow! It just textured   these 3D objects, and not one by one. No sir. It  textured the whole scene in one go. My goodness. Seven, it can also create tileable  textures. Why does that matter? Well,   if we are asking for a big patch of material, we  need to make sure that it is not just one image   copied and pasted all over again, but, it truly  gives us a seamless texture. That is excellent,   especially if we combine it together with an  alternative reality app where we look at our room,   and just ask a voice assistant to recolor the  wallpaper to be a little more corn-themed. Wow,   that is a little too much corn.   What about making it a little less   busy and more stylized? Now we see  exactly what that would look like. And finally, eight. This is going to be  incredible. If we have an animation of   a digital human and we wish to give  its appearance a complete makeover,   look no more. Stable Diffusion can take this  video, and just one image of the new test subject,   and, bam, there we go. We get a completely new  video with a much higher quality character. And before we go, I have even better news.   Yes, hold on to your papers Fellow Scholars,   because Stable Diffusion even runs on your  phone too. Just imagine having this kind   of power in your pocket, wherever you go. And  don’t forget, this is all out there in the open,   so you can contribute too, whether it might be  code, prompts, experiments papers, anything. We   are building this together. Oh, and you can also  track your experiments with Weights & Biases, ### [5:00](https://www.youtube.com/watch?v=rzKVInXep5M&t=300s) Segment 2 (05:00 - 06:00) and it can even help you optimize the efficiency  of your own image generation process. It is super   useful, that’s why I always talk about it  here on Two Minute Papers. So I hope this   gave you some ideas to experiment at home or on  the go, and for now, let the experiments begin! Thanks for watching and for your generous  support, and I'll see you next time! --- *Источник: https://ekstraktznaniy.ru/video/13195*