The Free & Uncensored Version of MidJourney! (FLUX.1)

Matt Wolfe
6 Aug 202418:24

TLDRThe video explores Flux.1, a new AI image generation tool by Black Forest Labs, which rivals MidJourney. Developed by the team behind Stable Diffusion, Flux.1 offers three models with varying power and cost. Flux.1 Schnell is open-source, suitable for personal use and local development. The tool is praised for its realism and text generation capabilities but falls short in creating detailed illustrations. The video demonstrates the tool's uncensored nature and potential for prompt adherence, suggesting Flux.1 could become a strong competitor in the AI art world.

Takeaways

  • 😀 Flux.1 is a new AI image generating tool that rivals Midjourney in capabilities and is developed by Black Forest Labs.
  • 🔍 The team behind Flux.1 includes developers of Stable Diffusion, known for innovations like VQ-GAN and Latent Diffusion.
  • 💡 Flux.1 offers three models: Schnell (fastest and open-source), Dev (middle line), and Pro (top line with state-of-the-art performance).
  • 🛠 Flux.1 Schnell is designed for local development and personal use, with images generated being usable for both commercial and non-commercial purposes.
  • 🔑 Flux.1 Dev is more efficient and prompt adherent than Schnell but is limited to non-commercial applications.
  • 🏢 Flux.1 Pro is designed for enterprise solutions and offers the best performance among the models.
  • 🌐 Several websites have integrated Flux.1, allowing free use of the models, with Hugging Face and Glyph being two such platforms.
  • 🎨 Flux.1 is particularly good at generating realistic images and handling text within images well.
  • 🚫 While Flux.1 is uncensored compared to some other models, it does have an NSFW filter and does not currently allow for generating copyrighted images.
  • 📈 Flux.1 is expected to evolve and improve, potentially becoming the best of its kind by combining the strengths of Midjourney, DALL-E, and Stable Diffusion.
  • 📚 The open-source nature of Flux.1 Schnell means it can be downloaded and run locally, which could lead to further development and customization by the community.

Q & A

  • What is the name of the new AI image generating tool discussed in the video?

    -The new AI image generating tool discussed in the video is called Flux.1, developed by Black Forest Labs.

  • Which team members developed Flux.1 and what are some of their notable innovations?

    -Flux.1 was developed by team members who helped build Stable Diffusion. Their notable innovations include creating VQ-GAN, Latent Diffusion, and models for image and AI video generation like Stable Diffusion XL, Stable Video Diffusion, and Rectified Flow Transformers.

  • How many models are there in the Flux suite and what differentiates them?

    -There are three models in the Flux suite: Flux.1 Schnell, Flux.1 Dev, and Flux.1 Pro. Flux.1 Schnell is the fastest and designed for local development and personal use. Flux.1 Dev is more efficient and prompt adherent than Schnell. Flux.1 Pro offers state-of-the-art performance designed for enterprise solutions.

  • What is the licensing of Flux.1 Schnell and what are its usage restrictions?

    -Flux.1 Schnell is available under the Apache 2.0 license, which means it is open source. Tools created using Flux.1 Schnell can be sold, and images generated with it can be used non-commercially or commercially.

  • What platforms have integrated the Flux models and how can they be used for free?

    -There are several websites that have integrated the Flux models. One of the simplest ways to use them for free is through Black Forest Labs on Hugging Face, where you can use the Flux.1 Schnell and Dev models within Hugging Face Spaces. Another platform is Glyph, which allows you to build your own Flux workflows and generate images for free, even using the Pro model.

  • What is the current limitation of Flux.1 in terms of generating illustrations?

    -Flux.1 is not as good at generating illustrations. While it can produce images that resemble illustrations, it may lack the fine details and specific styles like hand-drawn, oil painting, or watercolor that other models like Mid Journey can achieve.

  • What is Flux.1 particularly good at in terms of image generation?

    -Flux.1 is particularly good at generating realistic images and handling text within images. It is designed to produce high-quality outputs right out of the box and is on par with Mid Journey in terms of realism.

  • How does Flux.1 handle prompts with multiple elements compared to other models like Dolly 3 and Mid Journey?

    -Flux.1 does a decent job with prompt adherence, capturing most elements from the prompt. However, Dolly 3 is still superior in this aspect, as it can capture all elements from complex prompts more effectively than Flux.1 and Mid Journey.

  • What is the uncensored aspect of Flux.1 and how does it compare to other models in terms of generating copyrighted images?

    -Flux.1 is uncensored in the sense that it does not have the same restrictions as some other models when it comes to generating copyrighted images or images of existing IPs. However, it does have an NSFW filter, so it will not generate not safe for work content.

  • What future developments are expected for Flux.1 and how might they impact the AI art generation field?

    -Flux.1 is expected to evolve and potentially incorporate text-to-video capabilities, serving as the foundation for a suite of competitive generative text-to-video systems. This could make Flux.1 an open-source alternative to tools like Lum's Dream Machine and Runway Gen 3 Sora, offering more freedom and options for AI art generation.

Outlines

00:00

📸 Introduction to Flux One and Its Development Team

The video introduces Flux One, a new AI image-generating tool by Black Forest Labs, claiming it rivals MidJourney. Developed by experienced members from Stable Diffusion, including VQ-Gan and latent diffusion creators, Flux One features three models: Schnell, Dev, and Pro, each with varying capabilities and costs. Schnell is open-source and ideal for personal use, Dev is for non-commercial applications, and Pro is designed for enterprise solutions. The video also shows how to use these models via websites like Hugging Face and Glyph.

05:02

🖼️ Exploring Flux One's Capabilities and Comparison with MidJourney

The video continues by demonstrating Flux One's performance in generating various types of images, highlighting its strengths and weaknesses. It shows examples of hand-drawn illustrations, oil paintings, and watercolors of an angry penguin, comparing the results with MidJourney's outputs. Flux One excels in realism and text generation but falls short in creating specific art styles like watercolor or oil painting compared to MidJourney.

10:03

🌟 Realism and Text Generation with Flux One

The focus shifts to Flux One's strong points, such as generating realistic images and handling text well. Examples include realistic photos of people and a polar bear holding a sign. The tool's prompt adherence is tested with complex prompts like a three-headed dragon watching TV. While not perfect, Flux One shows promise in realism and handling detailed prompts, though it doesn't surpass Dolly 3 in prompt adherence.

15:07

🔒 Flexibility and Future Potential of Flux One

Flux One's flexibility is discussed, including its ability to generate copyrighted images and use existing IP, despite not allowing NSFW content. The potential for an open-source text-to-video model is mentioned, suggesting future advancements. The comparison with MidJourney and Dolly 3 concludes that while Flux One is not yet superior, it holds potential to combine the best features of all three tools.

🚀 Exciting Prospects and Final Thoughts on Flux One

The video wraps up with enthusiasm about Flux One's open-source nature and its potential for future improvements. The creator looks forward to seeing developers fine-tune and enhance the model, possibly surpassing existing tools like MidJourney, Dolly 3, and Stable Diffusion. The video ends with a call to follow the creator for more updates and AI tool insights.

Mindmap

Keywords

AI image generating tool

An AI image generating tool refers to software that uses artificial intelligence to create images based on textual descriptions or other inputs. In the video, the tool 'Flux.1' by Black Forest Labs is being discussed, which is positioned as a competitor to 'MidJourney'. The script mentions that Flux.1 has been developed by team members who also contributed to 'Stable Diffusion', indicating its potential to be a powerful and innovative tool in the field of AI-generated imagery.

Black Forest Labs

Black Forest Labs is the company behind the development of Flux.1, an AI image generating tool. The script highlights that this company is staffed by experienced professionals in AI image and video generation, which lends credibility to their new product. The mention of Black Forest Labs in the script establishes the context for the discussion of Flux.1 and its capabilities.

Stable Diffusion

Stable Diffusion is an AI model for image and video generation that has been influential in the field. The script notes that the team behind Flux.1 also developed Stable Diffusion, which includes models like 'Stable Diffusion XL' and 'Stable Video Diffusion'. This reference situates Flux.1 within a lineage of advanced AI tools and suggests that it may offer similar or superior capabilities.

Flux one Schnell

Flux one Schnell is described in the script as the fastest model of Flux.1, designed for local development and personal use. It is significant because it is openly available under the Apache 2.0 license, making it an open-source option for users. The script also mentions its intended use for non-commercial applications and its availability for free use on certain platforms.

Flux one Dev

Flux one Dev is positioned as a middle-tier model in the Flux.1 lineup, offering slightly more efficiency and prompt adherence than Flux one Schnell. The script specifies that this model can be used for non-commercial applications, which means it cannot be used to create tools for commercial sale but can be used freely for other purposes.

Flux one Pro

Flux one Pro is the top-tier model of Flux.1, offering state-of-the-art performance designed for enterprise solutions. The script suggests that this model is the most powerful and is intended for professional use, indicating a higher level of sophistication and capability compared to the other Flux models.

Hugging Face

Hugging Face is mentioned in the script as a platform that has integrated the Flux models, allowing users to try them out for free. The platform is significant as it provides a user-friendly interface for experimenting with AI image generation without the need for extensive technical knowledge or resources.

Glyph

Glyph is an AI workflow builder platform that the script describes as allowing users to build their own workflows with Flux models, including the Pro model, for free. This platform is highlighted for its potential to enable users to customize and optimize their AI image generation processes.

Prompt adherence

Prompt adherence refers to the ability of an AI tool to accurately incorporate all elements of a given textual description into the generated image. The script compares Flux.1's prompt adherence to that of MidJourney and Dolly 3, noting that while Flux.1 performs well, Dolly 3 excels in this area. The concept is crucial for understanding the capabilities and limitations of AI image generation tools.

Uncensored

The term 'uncensored' in the script refers to the flexibility of Flux.1 to generate a wide range of images without built-in restrictions. While the tool currently has an NSFW filter, the open-source nature of Flux one Schnell suggests that it could potentially be adapted to generate a broader spectrum of content, including copyrighted images or other IP, as long as it adheres to legal standards.

Text to video model

A text to video model is an AI system that generates video content from textual descriptions. The script mentions that Flux.1 will serve as the foundation for an upcoming suite of text to video systems, indicating a significant expansion of AI creative capabilities. This development is noteworthy as it suggests that Flux.1 will not only be limited to image generation but will also extend to video production.

Highlights

Introduction of a new AI image generating tool called Flux.1, developed by Black Forest Labs.

Flux.1 is comparable to MidJourney and may outperform it in some aspects.

The development team behind Flux.1 includes creators of Stable Diffusion and other AI models.

Three models of Flux.1 are available, each with varying power and cost.

Flux.1 Schnell is the fastest model, open source, and suitable for local development and personal use.

Flux.1 Dev is a middle-tier model with higher efficiency and prompt adherence than Schnell.

Flux.1 Pro is the top-tier model designed for enterprise solutions and state-of-the-art performance.

Flux.1 models can be used for free on certain websites, including Hugging Face and Glyph.

Glyph is an AI workflow builder that allows the use of Flux.1 Pro for free.

Flux.1 is capable of generating high-quality images with optimized prompts.

Flux.1 may not be as effective for generating illustrations compared to other models.

Flux.1 excels in generating realistic images and handling text within images.

Flux.1's uncensored nature allows for a wider range of image generation, including copyrighted IPs.

Flux.1 is good at prompt adherence, capturing many elements from complex prompts.

Flux.1 is expected to evolve and improve with community contributions and fine-tuning.

Flux.1 has the potential to become a superior tool by combining the strengths of MidJourney, DALL-E 3, and Stable Diffusion.

Flux.1's open-source nature means it will be adaptable and customizable by developers.

Upcoming text-to-video capabilities for Flux.1 will expand its generative capabilities.

Tips for improving prompts with Flux.1 include using AI prompt tuners and being detailed in descriptions.