With thousands of new AI tools launching every week it can be really difficult to keep up and find the best tools out there. I've been trying out hundreds of AI tools and I've found some unbelievable tools that can do lots of things from boosting your productivity, to animating images and even cloning yourself and winding back time. Sort of.
ClipDrop is a suite of image tools from Stability.ai who are the team that created Stable Diffusion. If you're not familiar with Stable Diffusion it's an image generation and editing model that can create descriptive images with short prompts similar to MidJourney and Dall-E. ClipDrop features AI-image editing tools including a background remover an image upscaler and more. It's most impressive feature is it's inclusion of Stable DiffusionXL that enables easy image generation and can even generate words within images. The paid version with unlimited runs costs £60/year and the interface is really easy to use and you can even plug the API into your own app. I've been using ClipDrop for a while and it saves me loads of time when it comes to creating images and editing existing images as it's so easy to use. I use MidJourney a lot for creating images but ClipDrop's interface makes accessing the latest Stable Diffusion models much easier and more accessible.
Metaphor is pretty crazy. It's a new kind of search engine designed from scratch using AI to be much more conversational. Metaphor's AI model is trained to do link prediction. This means that given a text prompt, it tries to predict the link that would most likely follow that prompt. For example you might say something like "This is the best tutorial on how to get started with cooking:" the results displayed are quite different from how Google's algorithm searches for results and I've found metaphor takes a bit of getting used to but produces content I wouldn't have discovered via a simple Google search.
If you don't have time to watch long YouTube videos like me and are struggling to find the best AI video summary tool you definitely need to check out Video Highlight which provides a summary of what the video is talking about based on specific time frames.
For example, I can add in a long video I might not have time to watch like an interview with Sam Altman about OpenAI and then easily scroll through the key points. I find video highlight to be the most comprehensive of all the YouTube summary AI tools with it's easy to follow interface and export options.
NVIDIA broadcast is insane as it can transform any room into a home recording studio using AI. Broadcast applies AI to your livestreams, voice chats, and video conference calls with AI-enhanced voice and video. Broadcast features a range of tools. At the most sci-fi end is Eye Contact which uses AI to make it appear as if you’re looking directly at the camera, even when glancing to the side or taking notes. Then there is noise reduction which works really well and also video noise reduction that improves your background. The virtual background works really well without any clipping issues when I tried it and you can also dynamically track your movements in real-time with Auto Frame, automatically cropping and zooming so you remain centered — even as you move around.
This AI tool is mind-blowing not because of it's ability to generate AI art using Stable Diffusion but because it's completely free of charge and runs offline without any limits. DiffusionBee is the easiest way to generate AI art on your computer with Stable Diffusion. All images are generated and stored locally and it's simple interface allows you to generate images from text prompts. DiffusionBee offers easy-to-use tools, powered by AI, to modify your existing images. You can add/remove objects, change the style of the image, or transform the scene using text prompts. If you don't want to work in the cloud or have privacy concerns DiffusionBee is a great place tool to try out.
Ai generated art is insane and I've generated hundreds of images across MidJourney, Stable Diffusion and Dall-E. The problem is that learning how to engineer prompts that produce the precise images you are after is much less accessible than ChatGPT and so people often give up. Step in Lexica. Lexica is a search engine and art gallery for AI artwork that also allows generation of images using Stable Diffusion. The site aims to make image prompting a bit less of a dark art and more of a science. Once you navigate to lexica.art in your browser, you can scroll down to check out recently-uploaded art. Clicking on an image will reveal the entire prompt used to generate the artwork, as well as seed information. The best feature of Lexica is the search bar, which can be used to search for specific prompts or prompt elements which you can then copy and use in your own prompt. If you're just getting started with AI art it's a really fast way to learn.
7. My Heritage
If you're new to AI My Heritage will immediately blow your mind as it uses Deep Fakes to animate and color restore old family photos that you may have. It's very accessible and it's really fun to take old photographs your parents or grandparents may have stored away and blow their minds with what the technology can do.
My Heritage uses the D-ID API and if you want to learn more about Deepfakes you might want to check out Deepface Lab which is an open source project that powers most Deep Fake technology and is pretty crazy.
Here are two very cool Ai tools in one that can save you loads of time when it comes to re-purposing content for socials. It can take my team ages to crop content for shorts, add captions and transitions to make short form videos as engaging as possible. With Vidyo you can upload a video or just add a YouTube link and then the AI will automatically split up your long form content into shorter videos that can be used across socials. You can adjust captions and even style your video all in a matter of minutes. Opus clip offers a similar experience but also adds in a virality score that identifies the most compelling hooks and segments of you long video. I've used both and while their ability to detect cuts isn't perfect they definitely save time and can reduce costs if you don't have an editing team.
Autodraw is insanely fun to use and shows off some of the more creative aspects of AI tools. You can think of Autodraw a bit like old school MS paint but with AI. As you draw a shape Autodraw will complete that shape and interpret your sketch into something more professional looking. Autodraw is part of Google's experiments lab and is completely free to try out. It's also just fun to see if the AI can guess what you're drawing.
10. Restore Photos
Here's a super useful AI tool that will restore any blurry photos. It's called Restore Photos and you can upload any blurry image and instantly improve it's clarity. I like Restore Photo as it's super simple and uses the GFPGAN open source API that is an image restoration machine learning model and shows the power of plugging openly available APIs into good UX to solve a simple problem.
11. Fake You
This AI tool is absolutely insane. It's from the team at StoryTeller AI and it can clone your voice and enable video lip-syncing in just a few clicks.
FakeYou is aimed at creators who want to dub their content rather than re-record. And in addition to the AI tools FakeYou also offers paid voice cloning services, an API with free and paid tiers, and in the future, users will be able to monetize their own voices.
Rewind is one of the most mind blowing uses of AI I've seen so far. It's been created by the founding team at Optimizely and aims to capture anything you've seen, said or heard and make it searchable to augment human memory.
Rewind does this by using optical character recognition and speech recognition on your computer screen. This is then stored locally for privacy and can then be searched using GPT-4 to quickly find that thing you can't quite remember from a zoom call 2 days ago. More than that you can then use GPT-4 to write an email or blog post and it will be able to reference some of this saved data. For example if you write a follow-up email from a sales call using rewind it can automatically reference what was said on the call and in any previous email exchanges to save tonnes of time.
This is probably my favourite tool on this list as it's so helpful for any knowledge worker or anyone looking to boost productivity and it's all built on top of GPT-4 using prompt engineering.