Just a heads up that it will be a slow couple weeks here for the channel, and I appreciate your patience and understanding. My entire family including myself ended up with pneumonia. I will be back in action shortly, but did want to give you all a heads up.
2024-03-12 19:27:08 +0000 UTC View Post
I started using Claude 3 to write code yesterday to compare it against GPT-4 and Gemini 1.5 pro... Wow, impressed to say the least. Not only is the context window large enough to fit a large chunk of my code base, it's also very good at writing complex code that touches multiple files across a complex system. As a result, thanks to Claude 3, https://pixeldojo.ai should no longer have any image generation timeouts when creating an up...
2024-03-08 22:34:55 +0000 UTC View Post
I started testing Claude 3 from Anthropic. It’s scores very high on all benchmarks, but that doesn’t always translate into real world performance. So, I’ve spent the day using it to write code. In this case to add webhooks to my application which requires the model to understand a significant amount of context while making changes to multiple files across the code base. So far…. I’m pretty impressed. If this continues I may not keep my OpenAI subscription, but then again I imagine t...
2024-03-07 20:45:55 +0000 UTC View Post
I'm working on another new feature for https://PixelDojo.ai that uses InstantID to allow you to create consistent characters using a single source image. Really powerful stuff, and I should have it ready for everyone to start testing out shortly so stay tuned!
2024-03-05 15:41:44 +0000 UTC View Post
Here is the video as promised along with all of the images generated by each model.
https://youtu.be/QVJrXGBrfpo
A few days ago I put together a video showcasing the differences in output from 10 different stable diffusion models. My idea was to keep the settings the same between image generations. This was actually flawed because the guidance scale and inference steps need to be optimized for each model in order to get the highest quality out of them.
So, I spent several hours today tuning the guidance scale, inference steps and scheduler for each of the 10 models I tested. I'll put together a v...
Hey gang, i'm a big fan of building in public and gathering feedback from customers as often as possible. As you know, I started PixelDojo based on the idea that Open Source AI Art and Language models should be accessible to as many people as possible, and not everyone can afford the computer hardware needed to run these resource intensive models. https://pixeldojo.ai runs on top of Nvidia A100 GPU's, and provides access to over 10 ...
2024-03-03 13:18:27 +0000 UTC View Post
Vote on your favorite Stable Diffusion model from my comparison video: https://youtu.be/XsG_-UlS_NI
Test the models at https://pixeldojo.ai
or
Download each model here:
Proteus V2: https://models.invoke.ai/dataautogpt3/ProteusV0.2 SSD-1B: 2024-03-01 17:54:52 +0000 UTC
View Post
Today I updated https://pixeldojo.ai with examples in the Stable Diffusion section. Simply click on any of the images and the prompt, negative prompt, and all related settings are automatically applied. This makes it super easy to get started creating higher quality images.
Don't forget about the new Image upscaler as well! (examples coming soon).
I've also been busy working on a face swap feature! As always, let me k...
Discover the ultimate training presets for your LORA model with Kohya SS right here.
What can you expect?
3 Custom Local Training Presets: Perfect for those who prefer training on their personal systems.
Why opt f...
2024-02-27 15:30:27 +0000 UTC View PostI had a couple people reach out asking for access to the private discord. You should have access when you join my Patreon, but just in case, here it is! https://discord.gg/XxYdQJJB9m
2024-02-27 15:19:47 +0000 UTC View Post
Over the past few days i've been working on adding a new AI image upscaler to https://pixeldojo.ai, and i'm happy to announce that the first version is now live! This upscaler relies on controlnet to identify and retain the key details in your images. It upscales images to 2048x2048, and can be tuned to add more detail and creativity to your images. Check it out and let me know what you think.
The dragon image above was...
2024-02-26 12:39:32 +0000 UTC View Post
I was excited that I got this working in my development environment and wanted to share. When released, this new feature will allow you to upscale images using a controlnet that upscales to 2048x2048 and refines the image using stable diffusion. You will be able to control how much it adheres to the original image, or go wild and allow it to make some big changes.
More to come soon. I need to go take my dog for a walk, i've been up coding and creating videos for over 24 hours lol. Thi...
Stable Diffusion 3: https://stability.ai/news/stable-diffusion-3
Stable Cascade Installer: https://www.patreon.com/posts/stable-cascade-98993458
Pixel Dojo AI Playground: https://pixeldojo.ai
Stability AI has done ...
2024-02-23 11:06:14 +0000 UTC View Post
Today I was able to add a new section for Large Language Models to https://pixeldojo.ai . The first 3 models I added are Google's new Gemma 7B, Mistral Mixtral 8x7b, and Meta LLaMA 2 70B. The 70B model would require 3 RTX 3090's to run locally, so it's a nice addition. Enjoy!
2024-02-22 16:37:10 +0000 UTC View Post
Fresh on the heels of the Stable Cascade announcement is news that stability AI is previewing stable diffusion 3. Below is the announcement along with a link to sign up for the preview. I will of course do an in-depth review once it launches, and get the model added to https://pixeldojo.ai as soon as possible!
From the announcement:
Announcing Stable Diffusion 3 in early preview, our most capable text-to-image model wit...
2024-02-22 15:49:55 +0000 UTC View Post
I'm creating a video on the subject, but wanted to share this to give you a head start. This is an auto installer for stable cascade that will help you get started running it on your PC with almost zero effort.
Launcher: https://www.pixeldojo.ai/installers/stable-cascade/stable-cascade-launcher.bat
Installer: 2024-02-22 13:57:27 +0000 UTC
View Post
Yesterday I shared a special discount code for https://pixeldojo.ai but there were some issues applying it at checkout. The issue has been corrected, so once again, for a limited time, here is a special promo code you can use to get 50% off for life. This is a special thank you for supporting my youtube channel and my content in general.
Use the following at checkout: Patreon50
A subscription gains you access to u...
2024-02-22 09:48:52 +0000 UTC View Post
This morning I pushed several updates to the Stable Diffusion Image generator.
You can now use SDXL, Juggernaut XL, and an XL Anime model.
In addition, you now have control over the sampler, inference steps, and guidance scale.
Finally, I added the ability to enable and disable NSFW filters meaning you can generate completely uncensored images if you so wish.
Enjoy and keep the feedback coming!
I've been working on a side project for a couple weeks, and I wanted to share it with all of you. It's https://pixeldojo.ai and the goal is to build the ultimate playground for Generative AI tools.
Right now it has the following features
* Chat with GPT-4
* Generate Music with Meta MusicGen
* Dall-E 3 Image Generations
* Stable Diffusion Image Generation - JuggernautXL, SDXL, SD 2.1
* Stable Diffusion V...
More news that none of us saw coming. OpenAI dropped a new stable diffusion video model called SORA, but it's more than it seems. This isn't just a video diffusion model, it's a reality simulator, and it paves the way for AGI or artificial general intelligence.
2024-02-16 14:57:18 +0000 UTC View Post
It has been an insanely busy week for the OpenAI team. We now have a preview release of ChatGPT memory, a new feature that helps ChaGPT gain more context when answering your questions. At the same time we have heard rumors that autonomous AI agents are on the way which will pave the way for ChatGPT to access and interact with web pages on our behalf.
Check the video for all of the details, and an announcement of my own :)
2024-02-15 17:20:52 +0000 UTC View Post
I haven't posted any videos this week, but that's because i've been busy creating a new project for everyone to check out. It isn't quite ready yet, but stay tuned :)
2024-02-14 16:17:09 +0000 UTC View Post
In my last video I showed you how to install, configure, and use ComfyUI, now we dive into one of the new workflows you can test out which installs the new Stable Video Diffusion XT 1.1 model that was released last week. This will get you everything you need to setup the workflow and animate your own images.
💻My Stable Diffusion PC: https://kit.co/AllYourTech/stable-diffusion-build <...
2024-02-11 17:00:07 +0000 UTC View Post
This week I cover everything from humanoid robots to 3d printable AI-generated models and everything in-between. Check it out and view all of the news at the links below:
Meshy Text-to-3d object: https://app.meshy.ai/workspace/text-to-3d
OpenAI AI Agents: https://twitter.com/blovereviews/statu...
2024-02-10 17:21:53 +0000 UTC View Post
ComfyUI can be overwhelming when you first start using it. This guide will take you through installing the software on your PC along with Comfy Manager to help add missing nodes to your Comfy workflows. By the end of teh video you should be able to start using workflows to create your own generative AI art.
Patreon Installer: https://www.patreon.com/posts/one-click-for-ui-97567214 Pytho...
2024-02-10 17:15:44 +0000 UTC View Post
Go Follow AlexandrIA for some amazing prompt ideas: https://twitter.com/AleRVG
This was done with MidJourney V6, but can easily be used with SDXL or any other diffusion model.
Prompt:
[𝚜𝚞𝚋𝚓𝚎𝚌𝚝]wearing an [𝚘𝚞𝚝𝚏𝚒𝚝] in [𝚜𝚌𝚎𝚗𝚎], in the style of vibrant postmodernism, made of crystals, Daria Endresen, renaissance, Joana Vasconcelos, [𝚌𝚘𝚕𝚘𝚛𝚜] ...
2024-02-08 10:11:14 +0000 UTC View Post
It's no secret that Google has been lagging behind in the world of generative art, so I was excited to test out Imagen 2 and see how the latest offering from Google stacks up to DallE-3 and others. Spoiler.... I wasn't super impressed. There are SOME things it does well, but overall there are far more capable models out there. Try it out and let me know what you think!
2024-02-07 15:00:06 +0000 UTC View Post
Nvidia showcased some pretty mind-blowing advancements in the field of AI simulations. Combing through the details makes me think we're edging closer to AGI and intelligent/capable machines. That's both exciting and a little terrifying. Have a watch and let me know what you think!
2024-02-06 15:00:14 +0000 UTC View Post
Lots of AI News this week from Google showing off a new Mobile Diffusion image generator, Neuralink implants going into their first humans, Meta releasing Code Llama, Mistral AI facing a leak, Google changing bard to Gemini and announcing Ultra, and Alibaba giving GPT-4 Vision a run for its money. Check the video for all of the juicy details!
Huggingface chat assistant: https://huggingface.co/chat/assistan...
2024-02-05 15:44:50 +0000 UTC View Post