The BEST AI VIDEO Generator is… (Head-to-Head Comparison)

which is the best AI video generator and which one should you use for your project in this video I'm going to compare and rank the best AI video Technologies across a number of different metrics so let's dive in and compare the best AI video generators together we're going to run through a number of tests each round will pit each of these platforms against each other in a unique Challenge and I will give you my opinion on which is the best for each of these specific parameters but I also invite you to offer your own perspective in the comments below so the rounds that we have include user experience where we'll look at the interface and ease of use followed up by realism where we'll look at the quality of the generated content and compare it anatomically and proportionally to that of real life we'll then move on to Res resolution looking at the video Clarity and output options and the ability to upscale these inside of the tool or not the next round is lip syn King which is the ability to take audio dialogue and match it to characters lip movements so this is key if we actually want to move into creating AI movies our next round takes us into the Realms of speed where we'll burn some GPU power and see which of these is the fastest at rendering the clips our penultimate round takes a look at the pricing options of each and how much it costs costs to make a single generation and also what are the unlimited options for these finally we'll take a look at any other features that are worth mentioning and give you a Roundup and a rundown of which is the best AI generator to use for you so we are going to be taking a look at Runway versus pea versus Hyper and occasionally we're going to bring in Sora as a bench mark for what's possible now first up is user experience and this is a balancing act for the product designers Behind these tools because they have to balance on one hand creating a tool that's simple to use and on the other including as much useful functionality as possible and inherently as you add more complexity more features the user experience the user interface becomes more complicated we can take elegant Pia they offer a very simple user interface uh with a simple prompt bar at the bottom and then a few other options to add more insight more Choice more intention more control over your video creation things like uh negative prompts seed numbers consistency as well as frame rate resolution and aspect ratio so essentially they have a a pretty simple user interface uh with a more limited set of features now let's contrast that with Runway now Runway has a huge amount of different features and capabilities it offers more varied options including removing backgrounds using Ai and a full-blown video editor but if we're looking specifically at the video AI generation experience you can see that they've tried very hard to simplify this into a unified simple experience where you can simply add a image or add in your prompt and then they've hidden a lot of the more complex settings uh Behind These tabs on the left hand side it's fairly comparable to peek up in the style however it is actually giving us more features and control which does add some more complexity finally we have hyper which is more like paa add is a very simple prompt entering interface and the ability to select Styles along with that with very limited additional features and functionality but the benefit of that is that the learning curve is very fast so to summarize hyper provides the most accessible user experience with a very simple interface but limited functionality Pika offers slightly more functionality and yet retains much of its Simplicity and finally Runway gives you the most complexity feature Rich environment but it starts to get a little bit more unwieldy so let's move on to realism and take a look at some of the examples that we've got we're going to go through five rounds of five different types of content that you might like to create in your AI video and we're going to compare each of these and see what the strengths are what the weakness Es are how effectively they render out coherently which means that the image makes sense that the the physics that the reality of the image is actually in line with our own understanding and then we're going to take a look at also how they adhere to The Prompt is it actually matching up with what we've asked for and we're also going to talk a little bit about the the style the aesthetic how visually beautiful are they so for this first prompt we're going to be using an old man with gray hair and glasses devours a delicious cheeseburger the bun is speckled with sesame seeds fresh lettuce a slice of cheese and a golden brown beef patty his eyes are closed in enjoyment as he takes a bite he wears a red jacket and appears to be sitting inside a fast food restaurant so with that in mind let's take a closer look at the examples and this first edition is from hyper so the obvious thing that is a a Miss with the hyper rendering is that the burger is floating and he's just eating the bun it also looks a little bit like a cup of tea rather than a burger however we would like to comment on the proportions of the gentleman he looks fairly proportionate and coherent his chin is maybe a little bit large and this left hand is not rendered particularly accurately it looks like it's missing the ring finger so he does have a ring on the finger which is quite impressive and the other thing I would mention is that the burger looks perhaps just a little bit too large but if you're in the states I know they like things super size over there so let's move on to Pika and you can see that the aesthetic here the style that comes out from exactly the same prompt is very different now P has this this beautiful aesthetic that is I would say that the it's much more of a a darker more mysterious feel to the video it has more cinematic feel certainly the hyper version feels more like a casual advertisement now Pika I would have to say it does not have a lot of movement in the image which limits his ability to start to render out nonsensical movements but the gentleman looks to be enjoying his Burger I would say that the hand once again is not rendered particularly well the little finger is longer than the ring finger which is unusual and then on the other hand the ring the little finger is very small almost as if it's been amputated at the knuckle but the burger does look delicious well sized proportionate and I appreciate the the character in this gentleman I like his large cheeks a bit like mine you know very friendly looking Chap and the style of the glasses as well this uh this red and black tint on the rims of his glasses now it has every other feature that we requested accurately rendered out and finally let's take a look at Runway and here you can see that it's interesting has a very similar composition to hyper and also quite a similar style the the colors the tones the theme of the image is very consistent but if we look at the runway example we can see that his hand is melding into his Burger quite nonsensically which we will have to say knocks a point off but the proportions of the anatomy on the gentleman are very lifelike and realistic his mouth starts to look a little bit asew and not particularly well aligned or balanced but I do like the wrinkles on his face they seem quite lifelike and natural and add a certain character to this man so looking at all three of them together we can say that Pik has the most coherent version of a man eating his Burger I would like to say that Runway has what I would say to be the most realistic looking human and I would like to also comment that P has the most cinematic feel but I would like to comment that hyper has the most realistic hair if you look at the individual strains of hair coming out of hyper they have a real sense of detail compared to Pika where they're more like a block and also with Runway where there is something a little bit too thick about each of the hairs so my opinion for this my favorite would be pea followed by Runway followed by hyper but there are different strengths and weaknesses to eat if you're new here I'm Samson and on this channel we talk about the latest creative opportunities with AI next up we're going to go into something a little bit surreal now this is one of the main advantages of AI video that allows you to create surreal environments in realistic scenarios so you can have very odd things now the next prompt we're going to try out is an elephant made of leaves running in the jungle let's first take a look at hyper and you can see that hyper is outputed an elephant that looks very evidently made of leaves and the tones are restricted and aesthetic now moving on to pea we can see that this elephant perhaps is not made necessarily with leaves but it looks like it has some Festival painting applied to its ears ready to rock it out in the in the jungle get low and dirty on the Jungle the weekend rocking back with your me gone elephan and then runaway has not got the elephant made of leaves at all but the elephant does look very realistic apart from the fact it has a two trunks so in this regard you would say that hyper has adhered to The Prompt most closely that Pik has a nice aesthetic it's done fairly well though it's slightly misunderstood The Prompt and Runway has completely lost all adherence to The Prompt however it is a very realistic looking elephant if that was what you were looking for I would also like to on the the movement of hyper I I quite enjoy how the elephant is moving both like an elephant but also with some semblance of of leaves it's it's slightly floaty as well as Dancy and playful as if that is appreciated so in this round I would certainly say that hyper is number one followed by pea followed by Runway but let's take a look at the Benchmark which is of course exploring how Sora got on with this prompt so here we can see that it's a much more accurately rendered elephant and the lighting reflecting of these leaves is very beautiful it's also beautiful how the leaves move individually independently of the elephant so the leaves maintain their own Sway and glistening in the wind whilst the elephant is walking the reason I'm including Sora in this comparison is because it is The Benchmark of what is possible with AI video however Sora used over 10,000 gpus and it's unlikely that we're going to get our hands on it for a reasonable price any anytime soon and until that happens I think it's very valuable to see what is possible with the available tools we've looked at the importance of ux in each of these products and this is something close to my heart because I spent many years working as a ux UI designer from some of the most remote places on Earth in the Himalayas I battled Avalanches take come up in a desert I overcame dehydration and in tropical rainforests I battled cobras having the skills of a designer allowed me to travel to over 50 countries in 5 years all while developing a lucrative career but do you dream of remote work long to work when you want where you want and how you want but fear the risk of changing careers or have concerns about not having a college degree well careerist offer live lessons internships and onetoone coaching allowing you to start a new career as a UI ux designer investing in their course might cost you upfront but think of it as a seed for your Financial Freedom so if you want to learn how to design AI tools check out my link below the video to get $600 off thank you to careerist for sponsoring this video so next up we're going to take a look at Urban realistic setting with a female character and the prompt I used for this is a stylish woman walks down a Tokyo Street field with warm glowing neon and animated City signage she wears a black leather jacket a long red dress and black boots and carries a black pearls feel free to pause the video and read the whole prompt if you like but I will move on to the demonstration so first up we have hyper and you can see the woman is anatomically correct her hair is well rendered she is moving through a Tokyo like setting and the only thing I would say is that it's it's quite strange that it has rendered out the video with her moving backwards it's almost in Reverse time but we could reverse that and it feels more natural as you can see here so that's quite interesting I particularly like the the way the hair is rendered here floating gracefully in the breeze and everyone looks fairly consistent there is some Mis rendering going on here with this object possibly it's a bicycle or some sort of Street Furniture but it's effective I would say her nostrils are a little bit large as the other thing now Pika has given out quite a dystopian looking rendering it certainly has less realism to it her lips are insanely big they look like a little boat and her skin is very flat it does not look particularly realistic but it does have a stylism to it I do appreciate that it also understands that she's wearing a black leather jacket but it does not include the red dress whereas hyper really identified this part of the prompt and rendered it well finally we have Runway now the biggest issue here is that she has a very deformed face though the background is rendered with a lot more realism and detail compared to the other two which have have used a very shallow depth of field to allow them to blur out the background and remove the need to put in so much detail which does give it a cinematic feel but it lacks some depth of realism certainly hyper has slightly more detail than Pika and Runway has the most detail in the surroundings but it it falls down in the in the human anatomy so for this round I would say that hyper is the most realistic the main issue is that it's playing in Reverse followed by a close call between PE and Runway now Sora once again is in a league of its own however it's not perfect I think there are a couple of things we'd want to notice is that the floor moves at a different speed to which the woman is walking that it's not completely it's almost as if she's sliding over the ground rather than actually stepping on it which is one thing to note also if you look at the pedestrians in the background uh it is almost like they are floating a bit like Death Eaters from Harry Potter but the proportions even the way that her bag is swaying in the same Cadence as her movements is very beautiful and the rendering of the background is beautiful as well particularly I'm enjoying the stunning Reflections on the wet floor of the neon lights next up is animal realism and we're going to look at a golden retriever in sunglasses sunbathing on a beach in Hawaii sounds delightful I will shy with this little golden retriever so first up it's hyper the dog looks realistic there's beautiful Reflections in the sunglasses of our little dog enjoying his holiday it looks like Hawaii there's nothing wrong with the background and yeah we got a sense that he's panting and the whiskers are moving next up we have Pik it's certainly more stylized almost graphic interpretation of this the dog doesn't look particularly happy to be on holiday he looks like God it's hot here in Hawaii but it adds a little bit of humor to the image the colors are a lot more saturated and less natural uh but the dog is anatomically correct and yeah the the composition is quite interesting of this it's extremely symmetrical and balanced I mean even the symmetry of these two trees uh lends to it some sort of slightly more orchestrated feel to the video whereas this certainly feels a bit more like there's another golden retriever there with a camera saying hey let's take a photo you look so happy Lola I have a a golden retriever called Lola that's why oh yes now Runway I would I'm I'm enjoying this wow look at the Poise the posture the intelligence of this wonderful golden retriever makes you makes you feel uh patriotic almost but what I appreciate about this is yeah the the feel that we get from the beautiful garden retriever that there is both character and Poise the waves look very natural in the background the dog doesn't seem to move a tall which is one thing I would mention and there is some strange morphing of the fur here which doesn't look particularly natural realistic but the overall quality of the image the aesthetic is very beautiful and to compare them all together I would say that Runway has the most realistic interpretation of this prompt there's something a little bit strange going on with the keychain on the collar but apart from that the actual rendering of the image is the most realistic however there is very little movement in this one P gives us a real graphic interpretation I mean this almost seems like a postcard a comical postcard but there's not so much animation going on and there's something that's a little bit overly stylized for my taste and Hyper I think does a very good job here so I would call this one a tie hyper and Runway first with Pica up next but I think they've all performed fairly well in this scenario but let's take a little look ourselves at of course the leader in the field right now and we have oh yes a very Regal Happy Dog who's even found found himself a beach towel to rest upon off for his holiday you can see that both the camera movement and the subject movement is more developed from Sora but actually if we if we compare perhaps the detail and the realism to to Runway I would say there's not so much to separate these two I would actually like to say that the composition of Runway is more visually engaging there is certainly a feel of the Sora version to be very much of a a very amateur video being taken by a holiday maker it doesn't have the feel or the intention of a more thoughtful cinematographer so one thing that's particularly interesting to reflect upon is the cinemagraphic intelligence of each of these AI tools that is what I would describe as their basic understanding of cinematography and whether they render out something that would have the eye and approach of Christopher Nolan or it simply looks more like a a dodgy iPhone video that your auntie took at the beach now our next round is an action shot round and we're going to be looking at a rally car driving through a forest and the prompt for this version was a wide shot video of a r car driving through a redwood forest and showing up behind the trees spraying dirt into the air as it drifts around a sharp turn cinematic film shot in 35 mm depth of field so let's go ahead and take a look at how each of our AI video generators did with this prompt so first up we have hyper it's got a very slow motion short clip the dust is coming out it looks like a rally car and we've got a forest it doesn't feel quite so cinematic and we don't have any feel of it going around the corner next up we have Pika it has more of a feeling of a video game especially with the intense saturation and there is something a little bit a Miss with the car being so heavily weighted down on one side Perhaps it is really drifting around a corner but I don't really get a sense that that is happening also the road does not seem quite so sensical it seems like a an unusual landscaping for an actual dirt road whereas Runway I would say has a very beautiful setting and scene the the road looks much more natural than in P there is a nice element of color grading and it does have a feeling of being cinematic especially the smoke and the dirt which is billowing out behind the car however it does look like it's missing a front wheel and the movement of the car is not coherent it looks more like a a toy car that a small child is pushing along and going also it's seems to be floating so it looks like all three of these have struggled with this route that action shots of cars is something that it does not have the current capability to effectively render but if I were to give my review on my favorites I would go hyper Runway peer and to take a look at how Sora does on this front you can see that it it does exceptionally well I would say that this is one of the most profound and beautiful videos I've seen rendered out of Sora just to compare how much more advanced it is compared to the existing AI video generators that the car is moving pretty realistically I would say there is a feeling just here as it drifts around the corner that it doesn't feel quite realistic there's something about the tires that do not look quite wide enough and seem to vary in width there's also not spraying so much dirt behind the car but the dirt it does spray is realistic the forest is particularly beautiful and the occlusions that happen work very seamlessly and realistically which means as trees pass between the car and us the viewer they obscure the car very naturally and this gives us a real sense of dynamism and also immersiveness it really feels very immersive like look at this tree here and the accuracy of the car being rendered but also the ability to keep the focus on the car Works particularly well so it's very interesting to see that I would say that this is where the biggest discrepancy is between sorum and our existing AI videos so to summarize with the realism I would say that hyper does a fairly good job at rendering out realistic scenarios however the movement lacks Clarity Pik does a better job at rendering out something that's has a little bit more of a comical feel to it slightly stylized uh iteration on version so for realism it is slightly lacking in that department and finally if we talk a little bit about Runway I would say that it often renders things effectively but it does have the tendency to completely disfigure elements inside of the piece I would also say that hyper and Runway do a better job at adhering directly to the prompt so for me on realism I would put uh I would say just hyper takes it followed by very closely second with Runway and in third I would say peer but feel free to disagree with me in the comments now of course it's very important to discuss the resolution of these images now the standard resolution that comes out of both Pica and Hyper is 1280 by 720 however Runway output a standard resolution of 148 by 768 this is for 16x9 aspect ratio which is essentially what this YouTube video is right here now for other aspect ratios essentially it will be reconfigurations of this amount of resolution so you will still get roughly the same pixels for each aspect ratio that you output now what is interesting is that both pea and Runway offer the option to upscale your videos inside their tools and P gives us an upscaling to 2560 by 1440 and Runway gives us 2,816 by 1,536 so you would say that Runway offers the highest resolution followed by pea followed by hyper now a little caveat on that is that of course you can take a hyper video or any other video and use an external video upscaling tool something like topan video but what I would say is the real key thing actually is not necessarily the amount of resolution that get out but actually the detail inside of these videos yes because what is really important is actually the detail in images that you can upscale because if it has low detail it doesn't make so much of a difference upscaling it anyway and if we look at the finer details on each of these you would say that Runway and Hyper have slightly more specific details so next up we have speed which is how long it takes to render out a video in each one of these so for this we're going to run the same prompt in each of the generators and take a look at how long it takes to get out a video The Prompt we're going to be using is one I've been playing with in mid Journey this was the mid Journey image that you got out and now we're going to pop it in to all of these so this is the hyper version I would say it's a very beautiful aesthetic image that's come out I like the the color tones and the proportions of her face it's a little bit flat the face though with the tonality the definition there's no sign of pause or imperfections on her face either which would be desired now the WoW The Runway video is very aesthetic the character is extremely uh beautiful profound and engaging I am very curious about her I notice is the immediate visceral response where this one feels a little bit Ah I'm like you don't seem so real and finally the P version is slightly shocking it's an odd composition to have and she has extremely intense cheek B in this regard I would say that I prefer Runway hyper peer but let's move on to length so both Runway and Hyper allow you to generate 4 second videos initially and Runway allows you to extend these so then you can take your 4C video and keep extending it so you have the potential to create an unlimited length of video now hyper is going to allow us to extend our videos soon but it's not available just yet whereas P gives you 3 seconds straight away and also the ability to extend the winner is Runway followed by Pika followed by hyper now the next round is lip syncing and lip syncing is a key feature for being able to take dialogue and match it to the lip movements in our AI video so if you are looking to create any sort of real narrative this is a key feature So currently only p and Runway offer this technique so first of all let's take a look at how Pik does feels like you and I are headed somewhere magical not bad followed by Runway is this the Hogwarts Express because it feels are somewhere better in Runway that the realism of the lips moving is better in Runway than Pika well I expect more than just a disappearing act if we're going magical I expect more than just a disappearing act and again in this second version with a woman you can see I would say that again Runway is the winner but let's take a look at these both side by side and you can certainly see that the rendering of the lips and teeth is more realistic in Runway however using both of these I have found that the P Pika version is versatile fast and easy for lip syncing however Runway may give you a better output it's more pedantic and it can be really challenging to actually get it to work because you need a very specific shot type to get it to work I was playing around with using some historical footage of famous people and lipsyncing their voices to new languages and I could not get it to work at all in Runway however I was able to do it in Pika so I would say Pika is versatile fast and easy however Runway gives you a better output but it can be more pedantic with its requirements and Hyper does not offer lip syncing so for this I would say Runway is the winner followed by P followed by hyper it's also important to note that we have not seen anything related to this in inside of Sora so it does not like that is a feature that they have built out or started to create inside of Sora yet now let's take a look at pricing so the great thing about hyper at the moment is it's entirely free and you don't have to pay anything to make use of it which is a huge bonus next up uh pea it gives you a basic free option to try it out as well as an entry level option of $8 a month which gives you 700 credits their unlimited option is $28 a month now Runway gives you a free trial uh but the once the credits run out you don't get any more and for their standard option it's $12 a month and this gives you 600 credits now their unlimited version is $76 so it is the most expensive to get the unlimited version so in times of pricing I would say that hyper is the runaway winner here being completely free and unlimited however Pika gives you reasonable option of $28 a month for unlimited and Runway is very expensive for $78 a month now let's have a quick chat about the unique selling points of each any additional features that are important to know about I really like that P has a new feature called AI audio which allows you to generate realistic sound effects with AI for your Clips so let's just take a look at a few examples from that back need it [Music] immediately now Runway has a whole sweet of additional tools that are useful in video making for example it has an AI green screen remover and a whole host of other tools that you can make use of so it's really a comprehensive video editing software as well whereas hyper has limited extra functionality it is offering you a repainting feature which allows you to change objects or specific parts of the videos that you have rendered so to summarize I would say that pea focuses on an entrylevel AI video generator it allows you to quickly create fun quirky almost comical videos without much knowledge or need for learning a complex tool it's price is in the mid-range and the quality of the videos that it outputs are reasonable but not particularly realistic or aesthetic Runway is an advanced AI generation for more serious AI video creators it is the most expensive and it has the most features but it is the most complex and sometimes the videos that it renders are disfigured and not particularly coherent so it can be a little bit hit and miss with the quality however hyper is a new player on the game it's free it offers high quality videos with an easy learning curve and a simple user interface however it has limited functionality and it doesn't have the ability to extend videos which is quite a key feature that you would like to have access to now what I think is the best approach is to all use all of these in combination I think there are situations where each one of these is the best and which one you use really depends on what it is you're trying to make I use all of these in my workflow I mean I particularly like to use hyper for quick easy videos to generate as well as animating my AI art I like to use runway for something a little bit more serious if I'm really trying to get something very specific done and I like to use Pika especially for the lipsyncing feature now let's comment a little bit on how these all match up to Sora I would say that Sora is in a league of its own in terms of realism coherence and prompt adherence however sorum is unavailable it uses vastly more computing power than any of these and was trained on more than 10,000 gpus where these were likely trained on tens or possibly hundreds at most of gpus obviously these are all accessible you can actually use them whereas Sora it's very unlikely we'll get our hands on it anytime soon the other consideration is that Sora is built with open AI open AI created this and they have extremely strict ideas around what you allowed to make with its tools so it's very unlikely we'll be able to use celebrities use art Styles inspired by art artists or directors or anything that has even a hint of sensuality or violence now Runway is also quite pedantic around its content controls however I found both Pica and Hyper to be a little bit more lenient with having a bit of fun with these Technologies so let me know what you think in the comments which of these is your favorite and if you have any thoughts on this video on how to make it better anything that you liked anything that you learned or anything that you felt was missing I'm very open to feedback and I would love to hear what you thought it would be very much appreciated and if you're new here why not subscribe and join the fun I'm Samson inves and I hope you have a delightful day

Channel: AI Samson

Views: 73,284

Keywords: ai, samson vowles, pika, runway, leonardo, pika vs runway, vs, review, comparison, best ai video generator, ai video generator, ai sora, haiper, haiper ai, ai video comparison, ai video review

Id: uoWOzWzWC7U

Length: 33min 41sec (2021 seconds)

Published: Mon Apr 15 2024