23 AI Tools You Won't Believe are Free
For the last eight months I've been experimenting with as many AI tools as possible, every day. It's for this channel but you could also call it an obsession. I've taken my whole list of tools and narrowed it down to the best free option for every use case, and I mean actually free. Let's start with a couple that are just really fun, then we'll get into the more practical. With Luma Labs you're able to take a video on your phone and turn it into a fully interactive highly detailed 3D scene like this That uses Neural Radiance Fields or NeRFs they also released a new feature called Interactive Scenes that uses gaussian splatting. And you don't have to know either of those terms.
The interactive scenes have a small file size so you can embed them anywhere or send it to a friend for them to explore. There's endless uses for this in retail, real estate, travel, we'll start seeing these across tons of Industries but also for creative uses as well as my favorite reliving old memories. Just download the free app on your phone, there's uploads from other users you can check out for inspiration.
I love the way these look when they load. Then you just move your finger around on the screen to navigate. To create your own either start a new capture and follow the steps on the screen or you can take a normal video outside the app. Just walk around a scene or if you're doing an object you'll get the best results if you do multiple circles around it each from different heights.
Just upload and let it render. Once it's finished you can navigate around the scene right on your phone or you can go to the web version on your computer. You can create custom videos using key frames, you can guide the camera movement throughout the scene in any direction at any angle. Here's a really cool one I saw on Twitter from Ruben Fro using Unity, so you can import these into game engines if you want there's a lot of different export formats. One of the ideas I'm working on is using Google Earth Studio. It's another free tool, you can pick anywhere on Earth and it will zoom in or zoom between locations. They have presets to do it really quickly or an area where you can fine-tune
exactly how fast it moves and the angles it takes. It's really fun to play around with and easy to jump in and use. There's a lot of tutorials on YouTube if you want more complex camera movements. This seems like a good spot to test out a NeRF That was my first test combining those tools I don't have the process dialed in yet but I think it has some potential. Either way these are both really fun tools to mess around with each on their own.
Luma Labs has one more recent release called Genie. It's a text-to-3D model that's free right now during the research phase. It's operated in the Luma Discord server in any of these Genie channels using the Genie command. Just add your prompt and it generates incredibly fast and is by far the best version of text-to-3D I've come across in terms of quality and time.
There's people rigging entire scenes with this that look amazing. I don't know how long this will be free but here's a few more examples. I'll go into a lot more creative stuff for music, images, video, and 3D but I'm also very into productivity and learning tools so let's switch to that for a few. One of my favorites that I use all the time is Perplexity. It's a fine-tuned
ChatGPT. Basically an interactive Search Assistant that you can ask anything. It uses up-to-date information and browses the web across multiple sources and comes back with concise answers almost instantly. It shows you what sources it used and you can click through to the article for further reading. It starts responding immediately and sometimes will cite like 10 sources all in one response if needed.
And if you're using the desktop version it can pull up images too. You can also upload PDFs and ask questions about them. I use the mobile app most and find myself using it over Google most of the time for question based searches. There is a pro version but the free version is huge, I've never felt any need to upgrade from the free plan. And I'm only talking about tools that have really robust free plans like this. I want to mention when they have paid upgrades available mainly to give you the full picture of how much comes in the free plan, that way you won't waste your time experimenting with something only to find out you have to pay to do what you actually wanted it for.
SciSpace is the easiest way to find, learn, and understand any research paper, so, similar to Perplexity but this only cites research papers, not websites or the overall internet. So it's geared towards people that are students or researchers, but it's great for anyone that has questions in the realm of science that wants research-backed answers. Ask any question and it will give you insight from the top five papers up at the top with citations you can click through to read the paper and then more relevant papers below. You can also chat with the copilot about the answers you received. Another cool feature is you can upload a PDF of any paper and get a summary and ask questions related to the paper. You could also add the
Chrome extension to use all those features while you're browsing. If you're reading a paper online you can ask it to simplify complex language, ask clarifying questions, clip math equations or tables to find out what they mean, it's a really powerful research tool to get answers in real time. All of that is completely free they do have some additional features for writing that are paid. Here's one more in this realm similar to perplexity but built right into Google, it's their new Search Generative Experience, which also just got the ability for text-to-image This is an experimental feature you can turn on to use AI in search. It
pops up on the top of your search and gives a concise summary of some answers with sources on the side. You can expand it to read more and ask follow-up questions. It can be really helpful in a lot of situations so I have it turned on all the time. To set it up go to labs.google.com
Click get started then switch it on. There's some additional options here for coding and adding AI to sheets. And you can get back to turn the AI on and off at any time with the beaker in the top right. They also very recently added in the ability to generate images.
You just type in the search bar what you want it to generate. From there you can click ask a followup and then you'll have a nice space to generate in. You can add follow-up prompts to change things until you have something you like. I actually liked this one with the guitar. Click export and
download or hit edit to go into a new workspace There's a button to have it help you write a better prompt, let's give that a try. Sometimes it ends up not generating all the images I think it's just censoring itself, but this one is solid. There's more text to image generators that do a better job at various things, I will be covering those a bit later.
Now I wanted to pack this video with as much information as possible, and that is made possible by today's sponsor, Skillshare. Before I started this channel I wanted to get better at video editing, I knew the basics but I wanted to really improve so I went to Skillshare. This is the course I watched by Phil Ebiner, it was great. I used what I learned in that course to make my first few videos on this channel. I've of course learned more since then and am always trying to get better, but that course was extremely helpful. And this channel did grow from 0 to over 100,000 subscribers in 8 months so, not bad.
And I've used Skillshare over the years for all sorts of stuff: photography, Photoshop, illustration, a lot of creative courses, I also took some for coding a while back. And they have a wide range of other courses on like entrepreneurship, social media, marketing, productivity and a ton of other niche topics as well. Honestly, I take a lot of courses. I have time carved out every week that I dedicate to learning
new skills. That involves multiple sources, of course, but one of the main ones is Skillshare. I love the learn by doing approach they take. Each course has something actionable with a project you can share and get feedback on.
I've gotten a lot from Skillshare so it was an easy choice when they reached out about sponsoring a video. And the first 500 people to use my link will get their first month of Skillshare membership free. You can watch multiple entire courses in that first month to make sure you like the platform then continue on if you do. So thank you to Skillshare for sponsoring this video and for helping me create this channel.
Now let's move on to some music tools. This next one is really fun and can get kind of ridiculous. It's called Suno and it can generate you a song with instruments and vocals just from writing a topic. It's operated within Discord so I'll have the link to the Discord server below with all the other links. Go to one of the chirp beta channels, use the chirp command, and either submit your own lyrics or write a topic and have ChatGPT write the lyrics for you. Let's do the greatest movie of all time.
The Big Lebowski Then you can type a genre. If you leave it as random it will choose based off the lyrics. Bluegrass seems like a good fit. And that's all there is to it. The verse and chorus will come back right away. Nailed it. Then usually just another minute or so and it will come back with two different options up to 40 seconds long each let's give it a listen [Music] In a world of bowling balls and weed, there's a man they called The Dude indeed [Music] his robe and is White Russian in hand, living life in his own laid-back land [Music] He's The Big Lebowski, that's his name That's pretty solid, especially considering all I entered was a few words. And there is an option to continue it if you want. Another fun one is to use lyrics from a popular song and see how it interprets it. I'll try Purple Haze with a random genre.
[Music] Purple haze all in my brain, Lately things just don't seem the same [Music] acting funny but I don't know why, 'Scuse me while I kiss the sky [Music] Purple haze all around, don't know if I'm coming up or down [Music] Am I happy or in misery whatever it is, That girl put a spell on me That's actually insane how good that is. I'm going to continue that for a future project. I originally scripted this section when it was 100% free and unlimited, but they have changed that now you get five chirps per day under the free plan. They do have a paid option if you want, but this is mostly just a quick tool I use for fun so the free plan has been plenty.
The next one is in the same realm but completely free and open source. It was released by Meta it's called MusicGen. It generates music under tons of different genres really well, but this time no vocals. There's a couple ways to use it I'll show the easiest one first someone built a better interface called Waveformer.
You just enter a prompt with whatever you want your music to sound like: genre, instruments, tempo, feel. Click submit and let it generate. It takes a minute or two then usually comes back with a great sounding song. There is another feature where you can give it an input Melody to follow but they didn't add that into Waveformer, so if you want to use that, you need to go to the space on Hugging Face. Upload a file that has a melody in it, I'll use the outdated reference of the X Files theme So you describe the music you want then it will use the description and the melody to create a song. Let's see what we got.
When I think of the implications of that combined with Suno and some of the other technologies that are coming out and advancing rapidly, I won't be surprised at all if people start listening to music that's fully generated by AI soon. This last music one is a really cool suite of tools developed by Google in collaboration with Lupe Fiasco. They help with lyric writing but in a collaborative way where the artist is really involved in the process. They're all free and really easy to jump in and mess around with. I like this a lot. Here's a clip from their promo video This bite-sized AI tool that can take in any word as input, explode it into all these different possibilities in the same way that Lupe does and that's just one tool we built nine others as well It's called textfx.
These tools, they generate similes, they create acronyms, they parse words, it's a entire suite of different kinds of possibilities dealing with text and language Now onto the more visual side. Probably the easiest way to animate an image is with LeiaPix Go to convert.leiapix.com click upload and easy as that, you have an immersive depth animation.
There's sliders to change how long the animation cycle is and the way it animates, also an advanced editor where you can get more customizable. Then just select your export format. It's super easy but I love the way these look [Music] You get 60 credits when you join, but to export in full resolution costs 20 credits so it's really only three uses. I have no idea why they call that 60 credits but I included it because you get unlimited free exports in 720p which is still pretty high quality. Also you
can export unlimited depth maps for free which we will use in our next tool, Insta3D This is by Ilumine AI it's kind of like LeiaPix but makes it even more immersive This is the home screen, there's some samples you can test out at the bottom. A bunch of these look really cool and gives you a good idea of what works well in here. Then right here you can upload your own image. Once you've done that you need to upload a depth map, so back to LeiaPix, upload your photo, then export depth map.
Now upload that in Insta3D and it's good to go. You can move all around the scene. They have some different effects down at the bottom to mess around with. Some types of images work really well for this and any of them start to get a little wonky when you go far into them, but it is really amazing that you can generate this so easily from a 2D image. [Music] This one is research from Meta, it's fun and really easy to use. It's designed to
animate children's drawings but you can upload any character you want. It just needs to be on a white background with each limb separate and visible Click upload photo, agree, then adjust the frame here if it doesn't capture the full character. On the next page you can fix the mask if there's any issues then find each of the joints then let it do its thing and there it is we have a bunch of different animations to choose from. Again, it's made to animate children's drawings so a lot of the animations are pretty silly, but it's fun to mess around with.
You could key out the background and let it play over another image, also it looks completely ridiculous if you upload a real person [Music] On to Pika Labs which is amazing and absolutely incredible that it's still 100% free. I'm sure at some point they're going to move to a paid model but it's been free for months and still is currently. It's all within Discord as well. You can operate it in their generate channels, or to generate in private come up to the Pika bot, click this member list button if you don't see it, then right-click the bot, and hit message. Now you'll be able to DM the bot anytime you want to generate.
Use the create command, and for text-to-video just type your prompt. There is a parameter for frame rate if you want -fps and a number between 8 and 24 then aspect ratio is -ar That looks great but I hardly ever use text-to-video, image-to-video is where it gets really powerful. I'll use the create command again, then type what I want to happen in the image, I'll use the capybara playing guitar I generated with Google SGE, and another parameter is camera movement like zoom in, pan, or rotate it's -camera then I'll try to zoom in Click the plus one to add the image, then hit enter That looks awesome. So you can dial in really solid images first then use Pika to animate them and the possibilities are endless CapCut has a whole suite of AI tools that are in beta under the magic tools tab. There's a bunch of different tools here, many of which can be found on other platforms, but the one I want to highlight is this video upscaler. You can upscale videos by 2x which is perfect for videos from Pika those are at 720p and this will convert them to 1080p.
I started using this recently and haven't seen any mention of a limit to the free uses on this or any of the other tools. So currently this is the best free video upscaler I've been able to find Those are some great video tools, but a lot of the time you need to start with an image. And before we jump into the image generators; something that's useful for any images you may have, ai or otherwise, is upscaling and restoration. Most of the upscalers I come across have limited uses, but there's two that are completely free and unlimited. Replicate has a full suite of image restoration and upscalers, upscaling is under Super Resolution. You'll find a bunch of models for different needs, some can 4x the resolutions, some will de-noise and smooth it out, others will add some detail. Pick one for your needs and run it.
The other option I use in Replicate is image restoration There's a ton of cool ones in here for touching up old photos, colorizing black and white photos, night enhancement, or what I find the most mind-blowing: de-blurring [Music] Replicate is great for quick web-based upscaling but if you want something you can run on your own computer that has the option to batch upscale, Upscayl spelled with a y .org is the way to go. It's free, open source software you can download. It has multiple options to choose from for different types of upscaling. Most
of the time I like this real esran 4X you can even upscale it to 16x if you want. I also like the digital art model for some cartoons or vector art styles, that smooths it out a bit. If you regularly need to upscale images then I highly recommend this, and a lot of images that come out of these AI art generators do need upscaling. Now for the cool this name on the list: Illusion Diffusion There's a ton of awesome ways to use this. It was originally trained to
make these really cool QR codes you may have seen, then Mr Ugleh figured out this workflow for these spirals and other illusions. I love these look, it reminds me of... these, by Octavio Ocampo So yeah, I love this style. It's a ton of fun and you can do it all for free. They have some examples to use down here. I like the spiral a lot. Illusion strength is how much the spiral will impact the image, between .8 and 1.2 is a good
starting point. It depends on what you're going for and can take some experimenting. Type in your prompt then click run. Timeframes will vary based on the time of day, it's usually around 30 seconds or less for me but it can be much longer during peak hours.
And that looks awesome. The illusion strength has the biggest impact on how this turns out, that's the main option I usually adjust when I'm fine-tuning these. There are also some advanced options you can mess around with to adjust it further but here's that same prompt run but only changing the illusion strength. Instead of using their images you can upload your own and the possibilities are endless. A couple common ones are text, other cool patterns, logos, and faces and a lot of these you have to shrink down or squint to see the illusion so keep that in mind as you're fine-tuning them. Someone who went all out with this is Herolias on Reddit, they spent 14 hours doing this for every frame of Never Gonna Give You Up.
You pause at any point and it's a normal landscape then hit play and get Rick-Rolled There's so much that can be done, an easier one is to create a brand promo video using a logo. I'll do the YouTube logo. Then add a prompt. That comes back looking great. I did that with a bunch of different prompts and boom [Music] Now for the text to image generators. There's a ton of options out there, each have their pros and cons. The first three are pretty straightforward and easy to use, the last one is the most versatile with the most features so I'll go more in depth.
We'll start with Ideogram. If I click over onto this top tab you'll start to see what it can do well and where it struggles. It's great with text based designs, logos, and illustrations. This isn't going to be as good as the others for things like photo realism or getting really specific with what you want.
When you type your prompt it will pop up with a bunch of different styles, you can choose one if you want, or just direct it with your prompt. Then there's a couple aspect ratios to choose from, and generate. It comes back with four options, if there's one you want to work with, you can click remix and it will generate four more that are similar. You can do that as many times as you want until you have something you like. [Music] Dalle3 was just rolled out to Pro users in ChatGPT so you can use it there if you have ChatGPT Pro. If not, you can use it for free in the Bing Image Creator.
You just type your prompt and hit create. It does a good job with text, the best part is it's prompt understanding. You can get really specific with what you want and it nails it. You get 25 free credits when you create an account and then your boosts refresh each week [Music] And now that we're in the text to image section I'll just revisit Google SGE quickl, that's a really good option that is completely free to generate as much as you want here's a few more examples [Music] Stable Diffusion is open source, uncensored, and amazing.
If you have a capable computer you can install it locally and use it completely free, but that involves some technical steps and is not beginner friendly. Other platforms have been built on top that utilize Stable Diffusion and make it much easier to use. For most of them the free plans are very limited, but there is a platform that's extremely versatile with a ton of features, and the free plan gives you 500 Generations per day which is huge. It's called Playground AI.
It's a canvas-based platform where you have a full workspace to move around and create in. It looks intimidating at first, but it's really powerful. These are all just from today under the free plan. When you're on the canvas you can drag the corners here to change your size and aspect ratio, under the frame is where you can type a prompt: anthropomorphic frog playing a banjo on the right are some options you can learn and mess around with, I'll just use this one for four generations at a time.
On the left is where we can choose the style to apply to the image, it has a huge impact. This whole workspace was basically just me testing out the different styles, I'll demo them in a second. Hit generate and it goes pretty quick. And every one of these looks amazing [Music] Let's say I like this composition but want some variations, the image-to-image button will use this as a reference image then I can adjust the strength over here. You could adjust the prompt or style too
Now I've got four more to choose from. That was all with the Haze style here's some other of my favorite styles using the same prompt [Music] So we've got some cool images but there's a lot more we can do. Up here was with the Neon Mecha style, these turned out awesome so I made a bunch. I also did anthropomorphic frog playing a fiddle. I didn't really prompt for scenery, but these two looked similar and I want to merge them. I can move the generation box over the gap and generate. That combined them well but there is this weird robot-looking thing right here. No worries, I'll just use the eraser. That doesn't look perfect, but I'm going to be bringing in a singer anyways.
This should look good, but I need to remove the background so hit remove background. Then I'll drag it on here and make it look nice, crop off the bottom, and there it is. That's a fast and easy composite shot. So I'll select the whole thing then copy and paste, now it's one image I can export There are tons and tons of other features in here. If you want to learn,
a really good resource is their YouTube channel. Now that I've got this image, let's see how it looks animated with Pika. That's not bad, but I think there's just too much going on with it. I had a ton of other
images so I'm going to animate those then I want to add a song from Suno. I'll use the chirp command, and these frogs have a kind of futuristic style. They look like a prog rock band to me, then they're kind of cyborg-looking so frog cyborg for the topic.
And actually, I think I have to do it... yep, I can't help myself [Music] Alright, this song is so good. I'm about to spend way too much time making a frog cyborg music video, but I'll just use tools I covered in this video: Playground, Pika, and Suno [Music} In the swamp, where the shadows creep Legends rise from the waters deep Frogs transformed by wires and steel Cybernetic warriors ready to reveal Frog cyborgs, leaping strong and free With techno hearts, the future they decree Rocking the lily pads, like never before Frog cyborgs, an amphibious uproar [Guitar Solo] Frog cyborgs An amphibious uproar Amphibious uproar Now that I've got that song stuck in your head, if you want to stay up to date on all of the latest and greatest AI tools go to futurepedia.io AI moves fast so a video like this can get outdated but Futurepedia is constantly updated, and there's various filters to find the best tool for whatever you need.
If you want all of those tools and tips to implement them delivered directly to your inbox, subscribe to the newsletter while you're there. And then like, comment, subscribe, all that stuff too. I've got a lot of great content in the works
2023-11-24 08:10
