Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

literally insane how good Krea 2 is for outpainting when using this custom LoRA yijunwang2 didn't train this LoRA, cooked it 🧑‍🍳🐐 it's now an app with a cool interactive canvas on Hugging Face spaces ▶️

HuggingApps's profile picture

Hugging Apps

3,687 subscribers

19,232 Aufrufe • vor 2 Monaten •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

🕹️ The way of creating anything in real time with LCM-LoRA technology is so natural that it's here to stay. That's for sure! 💡 This is just a quick test I did using Krea to build an RPG dungeon. Imagine creating a video game in REAL TIME!
0:55

🕹️ The way of creating anything in real time with LCM-LoRA technology is so natural that it's here to stay. That's for sure! 💡 This is just a quick test I did using Krea to build an RPG dungeon. Imagine creating a video game in REAL TIME!

Javi Lopez ⛩️

164,254 Aufrufe • vor 2 Jahren

🎨 I trained 999 Style LoRAs with the fal Krea 2 Trainer. Every single image in this video is a different LoRA. That's 999 unique Style LoRAs, each trained for just 100 steps at 0.00035 LR, taking an average of ~30 seconds to train. As I said before, Krea 2 continues to prove that it's an incredible model for open-source Style LoRAs. Next week I'll be sharing even more LoRAs, along with a few surprises.
1:30

🎨 I trained 999 Style LoRAs with the fal Krea 2 Trainer. Every single image in this video is a different LoRA. That's 999 unique Style LoRAs, each trained for just 100 steps at 0.00035 LR, taking an average of ~30 seconds to train. As I said before, Krea 2 continues to prove that it's an incredible model for open-source Style LoRAs. Next week I'll be sharing even more LoRAs, along with a few surprises.

ilker

12,794 Aufrufe • vor 2 Monaten

🎉 Meet Krea 2 from Krea, an aesthetic open-source image model ranked #1 text-to-image from an independent lab on Artificial Analysis. Day-0 support is now live in SGLang! Krea 2 ships as two models built to work together: 1️⃣ RAW: undistilled base checkpoint: diverse & malleable, made for fine-tuning, post-training & LoRA 2️⃣ Turbo: 8-step distilled checkpoint: fast, high-quality text-to-image Train LoRAs on RAW, run them on Turbo: base for training, Turbo for fast inference on your own hardware Run it now with SGLang!
0:58

🎉 Meet Krea 2 from Krea, an aesthetic open-source image model ranked #1 text-to-image from an independent lab on Artificial Analysis. Day-0 support is now live in SGLang! Krea 2 ships as two models built to work together: 1️⃣ RAW: undistilled base checkpoint: diverse & malleable, made for fine-tuning, post-training & LoRA 2️⃣ Turbo: 8-step distilled checkpoint: fast, high-quality text-to-image Train LoRAs on RAW, run them on Turbo: base for training, Turbo for fast inference on your own hardware Run it now with SGLang!

LMSYS Org

11,216 Aufrufe • vor 3 Monaten

As a photobashing tool this is insane.. Krea AI running as an overlay over my Photoshop canvas as I'm painting. I'm using a tool called Windowtop for that. I'll post a small tutorial on how to do that, in the thread below🧵 #art #ai
0:55

As a photobashing tool this is insane.. Krea AI running as an overlay over my Photoshop canvas as I'm painting. I'm using a tool called Windowtop for that. I'll post a small tutorial on how to do that, in the thread below🧵 #art #ai

Martin Nebelong

121,472 Aufrufe • vor 2 Jahren

Fluxgym: Train FLUX LoRA LOCALLY with LOW VRAM FLUX LoRAs are super exciting, but there has not been an easy way to train them LOCALLY using LOW VRAM machines. Fluxgym is a Gradio app for training your own Flux LoRAs on your own 12G, 16G, 20G+ VRAM computer for free.
1:10

Fluxgym: Train FLUX LoRA LOCALLY with LOW VRAM FLUX LoRAs are super exciting, but there has not been an easy way to train them LOCALLY using LOW VRAM machines. Fluxgym is a Gradio app for training your own Flux LoRAs on your own 12G, 16G, 20G+ VRAM computer for free.

cocktail peanut

237,906 Aufrufe • vor 2 Jahren

🤯🤯 You can now create a chatbot on ANY GitHub repo using the Llama 3.1 405B model with Hugging Face assistants -- FOR FREE! 💰 This is insane! 🚀 Link: clem 🤗 Julien Chaumond
1:22

🤯🤯 You can now create a chatbot on ANY GitHub repo using the Llama 3.1 405B model with Hugging Face assistants -- FOR FREE! 💰 This is insane! 🚀 Link: clem 🤗 Julien Chaumond

Satvik Paramkusham

162,907 Aufrufe • vor 2 Jahren

Anybody know how to fix the lag on the Artist Pro 24 Gen 2? I've used the Gen 1 for years and it didn't have this issue. This is with zero stabilization and windows ink off. It's especially bad when moving around the canvas. I'm not sure why it's so slow when Gen 1 was fine.
0:38

Anybody know how to fix the lag on the Artist Pro 24 Gen 2? I've used the Gen 1 for years and it didn't have this issue. This is with zero stabilization and windows ink off. It's especially bad when moving around the canvas. I'm not sure why it's so slow when Gen 1 was fine.

gooseworx gooseworx gooseworx gooseworx gooseworx

153,000 Aufrufe • vor 1 Jahr

Today's update of Gemini 2.5 Pro Preview is amazing at code web dev, editing, and transformation -- it is now #1 on the WebDev arena, with an ELO score of 1420. That means you can build super cool, visually stunning web apps, with just a single prompt! Check out this example, where we explore the world using this app we created in Gemini's Canvas using the Maps API.
0:30

Today's update of Gemini 2.5 Pro Preview is amazing at code web dev, editing, and transformation -- it is now #1 on the WebDev arena, with an ELO score of 1420. That means you can build super cool, visually stunning web apps, with just a single prompt! Check out this example, where we explore the world using this app we created in Gemini's Canvas using the Maps API.

Tulsee Doshi

115,735 Aufrufe • vor 1 Jahr

Goodbye expensive no-code tools! 👋 DeepSite just changed EVERYTHING for entrepreneurs and marketers. This FREE tool combines Deepseek V3's power with a visual canvas, creating stunning, functional websites and apps instantly. Why DeepSite beats paid alternatives: 💻 More beautiful designs 💻 Better copywriting in templates 💻 Fully functional interactive elements 💻 Simpler deployment process 💻 100% free on Hugging Face Spaces Save this video right now - it might be the most valuable tool you'll discover this month! 🎯 Want the SOP? DM me.
12:36

Goodbye expensive no-code tools! 👋 DeepSite just changed EVERYTHING for entrepreneurs and marketers. This FREE tool combines Deepseek V3's power with a visual canvas, creating stunning, functional websites and apps instantly. Why DeepSite beats paid alternatives: 💻 More beautiful designs 💻 Better copywriting in templates 💻 Fully functional interactive elements 💻 Simpler deployment process 💻 100% free on Hugging Face Spaces Save this video right now - it might be the most valuable tool you'll discover this month! 🎯 Want the SOP? DM me.

Julian Goldie SEO

43,663 Aufrufe • vor 1 Jahr

As promised, here's a tutorial on how to do these special FX insane transitions on your own videos using a Glif agent! i literally use this on my iPhone on the couch, so the demo is done like that as well Link to the agent below, tag me if you're making sth cool with it!
3:25

As promised, here's a tutorial on how to do these special FX insane transitions on your own videos using a Glif agent! i literally use this on my iPhone on the couch, so the demo is done like that as well Link to the agent below, tag me if you're making sth cool with it!

fabian

11,693 Aufrufe • vor 11 Monaten

Another WTF moment. Using Codex as a fully custom game engine With the newly discovered and *insane* img2threejs skill 1) GPT imageGen a 3D item you want 2) ask GPT 5.6 Sol to use img2threejs 3) my harness brings it into the game automatically Just watch how good this is
0:31

Another WTF moment. Using Codex as a fully custom game engine With the newly discovered and *insane* img2threejs skill 1) GPT imageGen a 3D item you want 2) ask GPT 5.6 Sol to use img2threejs 3) my harness brings it into the game automatically Just watch how good this is

Nicolas Zullo

270,037 Aufrufe • vor 2 Monaten

This is crazy 🤯 I built an app in 10 minutes I built the distribution in 5 minutes Literally created a beautiful APP w functional navigation in 2-3 prompts And then Cooked up and AI reel to promote my new app in literally 5 minutes, Using Arcads I used to think AI UGC was cooked but.. The quality of these AI reels is insane. You can literally mass produce the reel format and post 30-60 reels/mo (then if any video gets above 5k views organically , throw it on ads) and you are on the money. im 100% serious Me and Mau have an app doing $1,000/day with just 4 winning UGC videos running on spark ads But now you can save time and money by having your UGC generated with AI We are living in the ERA of prompts a few prompts, can change your life
0:35

This is crazy 🤯 I built an app in 10 minutes I built the distribution in 5 minutes Literally created a beautiful APP w functional navigation in 2-3 prompts And then Cooked up and AI reel to promote my new app in literally 5 minutes, Using Arcads I used to think AI UGC was cooked but.. The quality of these AI reels is insane. You can literally mass produce the reel format and post 30-60 reels/mo (then if any video gets above 5k views organically , throw it on ads) and you are on the money. im 100% serious Me and Mau have an app doing $1,000/day with just 4 winning UGC videos running on spark ads But now you can save time and money by having your UGC generated with AI We are living in the ERA of prompts a few prompts, can change your life

Ernesto Lopez

50,060 Aufrufe • vor 8 Monaten

📡 LoRa (MeshCore) module for Flipper One — an off-grid radio messenger The video shows Flipper One sending a public message that's relayed by repeater nodes, reaching users over a long distance. The MeshCore network is built on LoRa technology (868 MHz in the UK) and uses repeaters maintained by radio enthusiasts to extend its coverage. In our London tests, messages traveled more than 150 km and reached hundreds of clients. The network operates independently of the internet, requires no radio transmission licence, and is free to use. A good option for keeping text messaging available when the internet goes down. Our LoRa module prototype is based on the Semtech SX1262 LoRa transceiver and u-blox NEO-6M GNSS receiver. The module can optionally broadcast your location, which is useful for coordinating outdoor activities. It's hardware-compatible with both MeshCore and Meshtastic mesh networks.
0:25

📡 LoRa (MeshCore) module for Flipper One — an off-grid radio messenger The video shows Flipper One sending a public message that's relayed by repeater nodes, reaching users over a long distance. The MeshCore network is built on LoRa technology (868 MHz in the UK) and uses repeaters maintained by radio enthusiasts to extend its coverage. In our London tests, messages traveled more than 150 km and reached hundreds of clients. The network operates independently of the internet, requires no radio transmission licence, and is free to use. A good option for keeping text messaging available when the internet goes down. Our LoRa module prototype is based on the Semtech SX1262 LoRa transceiver and u-blox NEO-6M GNSS receiver. The module can optionally broadcast your location, which is useful for coordinating outdoor activities. It's hardware-compatible with both MeshCore and Meshtastic mesh networks.

Flipper R&D

277,726 Aufrufe • vor 7 Tagen

AI Agents Just Made me this Video "The Honey Chase" 🍯 🐻 This new AI video tool, vid.fun, is incredible and I'm so addicted to it! With zero editing skills, you can go from idea to full video in minutes using AI agents. Here’s how it works: - Start with an idea: Drop a prompt, link, or even a PDF into the reasoning box (powered by o3-mini) for research & inspiration. - AI agents handle everything: Choose from 1,000+ styles or use your custom LoRA for storyboarding. - Pick the best video model: Minimax, Luma Veo2, and more. - Final Touches: Viddotfun stitches it all together, adds an editable Udio soundtrack, and you’re DONE! It's never been this easy, try it you'll love it 🤩 🔥
0:29

AI Agents Just Made me this Video "The Honey Chase" 🍯 🐻 This new AI video tool, vid.fun, is incredible and I'm so addicted to it! With zero editing skills, you can go from idea to full video in minutes using AI agents. Here’s how it works: - Start with an idea: Drop a prompt, link, or even a PDF into the reasoning box (powered by o3-mini) for research & inspiration. - AI agents handle everything: Choose from 1,000+ styles or use your custom LoRA for storyboarding. - Pick the best video model: Minimax, Luma Veo2, and more. - Final Touches: Viddotfun stitches it all together, adds an editable Udio soundtrack, and you’re DONE! It's never been this easy, try it you'll love it 🤩 🔥

Amira Zairi

11,039 Aufrufe • vor 1 Jahr

As I promised yesterday, I'll briefly explain LoRA training and share a workflow I made so you can do it quickly. First, let me answer a very common question: 'Why train LoRAs when we have such advanced models?' Even though we have incredibly advanced models now (like NBP), we still can't always get them to do specific things we want. Simplest example: the spritesheet LoRA I made the other day. I generated 1000 images with Nano Banana and only 100 were what I wanted. The LoRA I trained using those 100 images gives me nearly 100% consistent results. Second point is cost and speed. With LoRA, we can cut costs by 4-5x. And while doing that, we're generating 4-5x faster. How many images do you need for a good LoRA? This depends on your LoRA's complexity. For example, when I training the spritesheet LoRA, even though I used 100 images, I didn't include buildings in the training data, so this LoRA doesn't work for buildings. So think about your LoRA's use cases and add examples for as many use cases as possible to improve quality. What are paired images and how to train LoRAs for image-editing? When training LoRAs for image editing on fal, we call each edit example paired images - one with _start suffix, one with _end suffix. For example, if you're training a background remove LoRA, the unedited original photo will be your '_start' image. The image with background removed will be the '_end' image. Simply put: images we want to edit or use as reference get _start, target images we want to achieve get '_end'. Important: save both images with the same name. Like image332_start.jpg and image332_end.jpg. This way the system knows which images pair together. What about training LoRAs for models with multiple image inputs? Same logic. We still use _start and _end suffixes, but with one difference. Since there are multiple input images, we can number them: _start, _start1, _start2. Example: start images, 1st image = Woman portrait (image35_start.jpg) 2nd image = Glasses photo (image35_start1.jpg) 3rd image = Hat photo (image35_start2.jpg) Output image = portrait of woman wearing glasses and hat (image35_end.jpg) Can we do more detailed captioning? Yes. Similarly, you can improve training quality by creating a txt file for each set with the caption inside. Example: create image35.txt and write: 'Recreate the image by putting the glasses from the second image and the hat from the third image on the woman in the first image.' What are Steps? How many should I use? What's Learning Rate? Steps determines how many times the model sees and processes your training data (your images). Each step, the model learns a bit more. But as steps increase, so does the risk of overfitting. So there's no real default. But for a simpler LoRA with 20 paired images, 1000 steps is ideal. Here's a metaphor for the Steps and Learning Rate relationship: Imagine you have a balloon. Our goal is to inflate it to the optimal size. Steps = How many times we blow into the balloon Learning rate = How hard we blow each time If we blow too softly, we need to blow many more times. If we blow too hard, we risk popping it quickly and can't reach optimal size. Of course training won't explode, but it won't work as intended because it wasn't trained optimally. Training's done, now what? Once training's complete, you'll have a safetensors file. Every model you train on fal has a LoRA inference endpoint. In that inference, add your safetensors file link to the LoRA url input, and you can use your LoRA. Thanks for the read! The workflow in the video: If I forgot anything, let me know in the replies.
1:06

As I promised yesterday, I'll briefly explain LoRA training and share a workflow I made so you can do it quickly. First, let me answer a very common question: 'Why train LoRAs when we have such advanced models?' Even though we have incredibly advanced models now (like NBP), we still can't always get them to do specific things we want. Simplest example: the spritesheet LoRA I made the other day. I generated 1000 images with Nano Banana and only 100 were what I wanted. The LoRA I trained using those 100 images gives me nearly 100% consistent results. Second point is cost and speed. With LoRA, we can cut costs by 4-5x. And while doing that, we're generating 4-5x faster. How many images do you need for a good LoRA? This depends on your LoRA's complexity. For example, when I training the spritesheet LoRA, even though I used 100 images, I didn't include buildings in the training data, so this LoRA doesn't work for buildings. So think about your LoRA's use cases and add examples for as many use cases as possible to improve quality. What are paired images and how to train LoRAs for image-editing? When training LoRAs for image editing on fal, we call each edit example paired images - one with _start suffix, one with _end suffix. For example, if you're training a background remove LoRA, the unedited original photo will be your '_start' image. The image with background removed will be the '_end' image. Simply put: images we want to edit or use as reference get _start, target images we want to achieve get '_end'. Important: save both images with the same name. Like image332_start.jpg and image332_end.jpg. This way the system knows which images pair together. What about training LoRAs for models with multiple image inputs? Same logic. We still use _start and _end suffixes, but with one difference. Since there are multiple input images, we can number them: _start, _start1, _start2. Example: start images, 1st image = Woman portrait (image35_start.jpg) 2nd image = Glasses photo (image35_start1.jpg) 3rd image = Hat photo (image35_start2.jpg) Output image = portrait of woman wearing glasses and hat (image35_end.jpg) Can we do more detailed captioning? Yes. Similarly, you can improve training quality by creating a txt file for each set with the caption inside. Example: create image35.txt and write: 'Recreate the image by putting the glasses from the second image and the hat from the third image on the woman in the first image.' What are Steps? How many should I use? What's Learning Rate? Steps determines how many times the model sees and processes your training data (your images). Each step, the model learns a bit more. But as steps increase, so does the risk of overfitting. So there's no real default. But for a simpler LoRA with 20 paired images, 1000 steps is ideal. Here's a metaphor for the Steps and Learning Rate relationship: Imagine you have a balloon. Our goal is to inflate it to the optimal size. Steps = How many times we blow into the balloon Learning rate = How hard we blow each time If we blow too softly, we need to blow many more times. If we blow too hard, we risk popping it quickly and can't reach optimal size. Of course training won't explode, but it won't work as intended because it wasn't trained optimally. Training's done, now what? Once training's complete, you'll have a safetensors file. Every model you train on fal has a LoRA inference endpoint. In that inference, add your safetensors file link to the LoRA url input, and you can use your LoRA. Thanks for the read! The workflow in the video: If I forgot anything, let me know in the replies.

ilker

15,192 Aufrufe • vor 8 Monaten