After something of a quiet spell for news regarding AI image and video generation, things have heated up again over the last couple of weeks. Excitingly, some of the developments relate to the release of new open-source models, which means they could potentially be used to create NSFW AI generators.
Stable Diffusion Creatos Unveil Impressive AI Image Generator ‘Flux’
‘Black Forest Labs’ is an AI startup created by the original founders of Stable Diffusion, quickly gaining many millions of dollars in investment. Last week they showed off the results of that investment, by revealing their highly impressive new AI text-to-image generator called Flux.1. Note that Stable Diffusion itself forms the basis of most current AI porn generators, and Flux is likewise open source.
All the tech reviews I’ve seen of Flux have been very positive, and in particular, it has been widely praised for its ability to render hands and feet more realistically than any other AI image generator.
More tests with FLUX.1
It does an excellent job of creating hyperrealistic images. pic.twitter.com/VdTorsJ2Ah
— Halim Alrasihi (@HalimAlrasihi) August 3, 2024
Hands, and perhaps more importantly for some porn fans, feet, have been difficult for existing NSFW image generators to consistently produce with any degree of realism and detail. Flux seems to be another level up when it comes to satisfying foot fetish fans.
Just tested Flux.1 after hearing the news. Huge thanks to @camenduru for creating tost ai, allowing us to generate AI images. Regardless of image quality, its understanding of prompts seems impressive. I'm excited and curious about how the pro version will perform.
Prompt : ALT pic.twitter.com/a7LVWMRCRh— AI opener (@opener_ai) August 2, 2024
Even more excitingly, Black Forest Labs hopes to release an open source AI text-to-video generator soon as well.
Runway releases Gen-3 Alpha Image To Video
A handful of big tech companies, such as Meta, OpenAI, and Runway, are competing to create the first ultrarealistic AI video generators. Although the outputs are usually still limited to seconds of video, with each new release, the realism becomes increasingly mindblowing, and the video lengths a little longer. Last week Runway released the latest version – GEN-3 – of their popular image to video generator. This new version allows users to create short video clips using a still image, and an additional text prompt.
Today we are releasing Gen-3 Alpha Image to Video. This update allows you to use any image as the first frame of your video generation, either on its own or with a text prompt for additional guidance.
Image to Video is major update that greatly improves the artistic control and… pic.twitter.com/OieDwMIspz
— Runway (@runwayml) July 29, 2024
runway gen 3 image-to-video 👀
> midjourney image
> elevenlabs video-to-sound effects pic.twitter.com/ly0wZ4woPy— Anu Aakash (@anukaakash) August 3, 2024
However, the bad news is that Runway Gen-3 is not open-source, making it very difficult for developers to hack it for creating a NSFW video generator.
Open Source AI Video Generator Alternatives
However, there are several exciting open source AI video generator projects underway. One of these is Open Sora, which as its name suggests, is an attempt to make an open-source alternative to OpenAI’s Sora video generator. According to their Hugging Face page :
We present Open-Sora, an initiative dedicated to efficiently produce high-quality video and make the model, tools and contents accessible to all. By embracing open-source principles, Open-Sora not only democratizes access to advanced video generation techniques, but also offers a streamlined and user-friendly platform that simplifies the complexities of video production. With Open-Sora, we aim to inspire innovation, creativity, and inclusivity in the realm of content creation.
You can see some examples of what it is currently capable of at their GitHub page.
Another attempt to make Sora open-source is Open Sora Plan, with some developers on Reddit considering it more advanced than Open Sora.
A third open-source project based on Sora is ‘Mira‘, with the aim of creating ‘Sora like long video generation’. This may be the most advanced of the three at the moment.
As I was drafting this article, Chinese tech giant Alibaba announced that they are working on a text-to-video generator based on Open-Sora above, and which they are calling ‘Tora’. Hopefully, it will be open-source too.
Related posts:
OpenAI Will NOT Allow AI Porn Generation
OpenAI Sora And The Promise And Perils Of AI Porn Video Generators
UK Opposition Party Plans Ban On Nudify Apps And Strict Regulation Of AI Image Generators
Alibaba Make Breakthrough In AI Animated Video Generation
An Interview With The Developer Behind NSFW AI Image Tool PornJoy
