For the comparison to be fair you need to regenerate the runway shots as well. Their showcase shots might have taken 100 generations before they selected the winners
I agree entirely about the showcase shots. I'm sure they were cherrypicked. Did you watch to the end? I did run a comparison with the same prompts typed into my own Runway account just to see what they would look like. Really interesting results.
@@HaydnRushworth-Filmmaker Unfortunately this is how most AI video/image generators do. They generate many videos/images and pick the best one. This might be how Sora AI is doing it so it might not be the "Perfect" AI everyone was expecting.
😁… thanks. I’d like to say it was a cunning plan to get more views, but I figured it would be helpful for me if I were watching the video, so maybe it’ll be helpful for somebody else too 😁🫡
I've tried them all (Krea, Haiper, Pika, Luma and last few days Kling) while the latter is pretty good img2Vid it often does some weird stuff but now Gen 3 with Img2Vid is by far the best I've tried. I'm quite blown away. Not always perfect but coherence and quality output much higher quality. And to think it's only Alpha and getting better from here on! No freebie with Gen3 and nice that Luma and Kling offer some but for pro use the logo is a PITA. So my strategy is to have all my image assets ready for a project and take the unlimited month subscription when needed. (Went through the $35 one in no time)
That sounds like a fair plan, tbh. It's frustrating that we're left needing to pay $10 here, $20 there, and $35 somewhere else, just to cobble together results. I'm looking forward to the day that several, brilliant, all-singing-all-dancing tools become available that really do let you do everything in one place.
Thanks very much indeed :-) Great work on your channel btw. I watched the walking gnome video the other day and it got me wondering what an animated version of my movie could look like.
Thanks, mate. Big fan of your channel btw 🙂, incredibly helpful. That surprises me, actually. I'll definitely be interested in your review on the Gen-3 image to video if you're putting one together.
I have un unlimited account with runway, a basic one with luma and played with kling and with all the other features, Runway is still my choice. No the new image to video is not great yet but the paint tools and camera are coming. I do music videos so for the price compared to a crew etc, really worth it, as a hobby meh
I want to emphasize the idea that these platforms should disclose on their pages and advertisements some metrics, such as the attempt-to-success ratio, so that a prompt’s output quality level can be assessed based on the examples in the ad.
First time I come across your channel. Super interesting comparison and the wholesome moment at the end with the SPF 40 made me laugh! Very nice. You gained a new subscriber. Cheers A fellow director
Thanks for this comparison. Indeed, things are changing so quickly that it's difficult to determine if there is a single company representing the state of the art, but my perception has been that Gen-3 has the edge; but of course that could change on any given day!
Great comparison video, Haydn! I completely agree with your assessment of LUMA - it's indeed a powerful tool, but the cost adds up quickly. I've found myself relying on it less and less due to the expense. KLING seems like a promising alternative, but I'm eager to see more examples of its capabilities. One thing's for sure, the AI video landscape is constantly evolving, and it's exciting to see new players like KLING entering the scene, I use haiper AI for my videos. Keep up the great work, and I look forward to your next video
Heyyy! Thanks very much. I’ve actually started to realise that if I want to get better value for money from any of the new video tools I’m going to need to invest the time into studying prompt engineering, especially from the sources themselves (Runway, Kling, Luma… they often have their own prompt engineering advice).
I love the side by side! I just tried Kling image to video and even on the free plan it is impressive. Good to hear that RW3 has turned their image to video on. Would love to see a paid version image to video comparison similar to this fine video.
Thanks very much, glad you found the examples helpful. Of course, now that Runway has released Image to video, inevitably, I ned to try that out now, especially as I'm working on the promo video for my book and need to use image to video to create it.
If you can afford it (along with the inevitable selection of other AI tools), then some kind of unlimited plan is definitely the way to go, then at least it’s only your time that you’re using up with endless re-rolls. I have a monthly budget set for the entire collection of AI tools I use, and if I spend money on one of them, that’s money that I can’t spend on another, so, for me, each tool really does need to work for its money in order to stay on the books.
Great aproach! Sometimes we can get a lot of credits failed in a bad video too, I'm thinking to upgrade from Standard plan to Unlimited instead, so I don't have to worry if I waste credits, and it's even cheaper than Kling or Luma for the same price =)
Love your fam! Thank you for doing this vid! Anyway, agree on the cost per clip generation comment. For RunwayML, whether you're there for T2V or V2V, the unlimited plan is a must (if you can fork out the cash). The plan is really to use Kling to generate and chart out your entire story scenes workflow as an affordable first cut, and then subscribe to Runway for a month, and replicate each scene into 2K versions. sounds good?
Hey, thanks very much, really appreciate the moral support :-) Re the cost-per-clip conundrum... BOOOOMMMM!!! I've actually been thinking the same thing... except I'd probably stay submitted to Runway and follow the same pattern on an ongoing basis. That said, if one of the paid plans with LUMA or KLING was worth it, I'd consider that too.
Thanks for that. Of course the companies run as many generations as need be and then cherry-pick the best result. Keep Runway because I need it for lip-syncing. My main go-to is Haiper. $30 for unlimited 4 second generations, but you have credits for 8 second and upscaling. Luma, is, as you said, more bang and better bang for your buck. I also like that their credits are 1 credit = 1 generation, so you really know how much you have left. Now I'm starting to realize which tool is better for which kinds of animations. Considering that it usually takes multiple tries to get a usable result, one really needs an unlimited plan, and they are just ridiculously expensive. So, I rely on Haiper and use the other services for my flourishes. How to be an AI video creator without breaking the bank.
100% I'm not advocating only using free services, but it's irresponsible to "blow the budget" and pay top $ for every AI tool that pops up over the horizon. There's a real balancing act of risk/reward that we have to constantly try to balance.
honestly, as a narrative filmmaker I want videos that specifically adhere to what the scene requires not the wild stuff that everyone's experimenting with. it's confusing which of the subscriptions i should buy at this stage. also , any suggestions for which VIDEO upscaler i should use?
Glad to hear it’s not just me craving a tools that gives me the basic shots really well. Re. which subscription, for me, at this stage, I’m still developing an ever-evolving workflow (which I need to make a video about), and since no, single tool does everything on the workflow, I figure if I can settle on a productive workflow, I can swap tools in and out as they evolve and improve. In terms of video upscaler, I’ll be honest, I’ve never used any. Until I can get results that I’m happy with, I have no reason to upscale. A steaming turd is still a steaming turd, regardless of how high the resolution and quality is. I’d rather have low quality results that tell an emotionally vibrant than a high quality tool with no emotional impact in the story (or editing).
@@shireensingh2834 honestly I think you just have to wait. It seems like every iteration get better and right now, the archaic way of getting shots vs what some of us more traditional types expect just isn’t quite there. I assume it will be within a year, so start getting those shot lists ready lol.
The nerds are back! Missing the point of the story and using a microscope to pick holes, I wonder what Picasso would have answered. "The water on the window is larger than" Oh my. This guy is the Mother of all nits.
You are making very good points about Gen 3, but their $99 unlimited plan is the best compering to cost per generation. It is still cheaper than the cost of the Hollywood movie. I think you would need both Gen3 and Kling if you want to create a movie. They are good at different things and you can use them for different purposes. I do agree that if you are on $30 plan with Gen3 that is very expensive cost per video.
Yeah, cost is a really important issue that everybody in film production needs to keep a tight handle on. Costs can get wildly out of control on a $150m project as well as a $150 project. I think Runway may have just cut their prices though.
Kling just started offering paid subscriptions, and they have a 50% off promotion happening. You might want to get a subscription and try the professional mode. The sunscreen won't smell as bad as the WD-40. 😁 I like what you're doing. If I end up making some kind of creative endeavor with all this AI video stuff, you will have been the major influence behind that. I appreciate your obvious creative talent displayed in your videos!
I think you’re right about Kling. I got interrupted with kiddies bath time mayhem a couple of days ago when I was trying to set it up. Thanks for the reminder :-) Told my wife about her better-smelling oil choice… she laughed 😁😆 Thanks for the encouragement and support too, really appreciate it 😁
Just got into this text to video ai thing. However, unable to test Runway, when try to generate "due to high demand, generations on free plans are not available". How are we supposed to pay for a plan not knowing if it will work for us? Runway = fail. I tried Luma and it works great and will buy a plan soon. I think we already have a clear winner.
That sounds like solid reasoning to me. Minimise risk wherever you can whilst still learning (and beyond). We already carry a lot of the risk of failure for these tools, so I don’t think it’s unreasonable to find the least risky option.
If you make money with it, 100$ for unlimited is simply mind blowing and also the cheapest option. If not and it's just a hobby at the moment, I can understand.
Yeah, we're still firmly in the R&D phase with generative AI, and with so many AI tools charging for monthly usage, we all need to be really selective about what we pay for. The money can trickle away without us even realising.
Really like this channel. so much i subbed on my non trash account 🤣 Hope you more attention, this is really interesting for those of us who arent scared of AI
I know, it really did literally become available as I was putting the finishing touches to this video. Funnily enough, I'm about to re-subscribe because I need to use image to video more to finish off the advert for my book that I'm getting ready to promote.
Compare the best AI video tools: KLING, LUMA, and Runway Gen 3! Each offers unique features for video creation, from advanced animations to powerful editing capabilities. See how they stack up and find the perfect tool for your needs. #KLING #LUMA #RunwayGen3 #AIVideoTools #TechComparison
I’m inclined to agree. LUMA is great, but for the free plan you really do wait a looooong time for results. Kling seems to give the best all-round results at the moment.
Hi. Sorry for the late reply. Do you mean that you want to create a still / static image in Kling? They do have a Text-to-Image option, is that what you mean?
Thats how i feel about AI subscriptions right now. You are also paying for AI failures too. I know its worst it will be today but i dont see a point in wasting my money.
Sure, no problem: "a large robotic King Kong wearing square diamond encrusted sunglasses, kings robes and a very large crown stands menacingly and confidently with his arms folded facing the viewer. Angry, menacing, threatening. Dark grey background. High contrast." As with so many AI images, the end result, whilst cool, didn't exactly match the prompt. I struggled and struggled and STRUGGLED to find a prompt that would make the gorilla robotic. Also, I expanded the original image to enlarge the grey space for text. The KLING KONG was a photoshopped version of the KLING logo that I screen-grabbed and then edited :-)
@@HaydnRushworth-Filmmakerthank you very much, usually for robotic you can use, cyberpunk, robotic enhancements, anthropomorphic, humanoid cyborg and so on. This I saw with good results. Again thank you, far too kind 🙏🏻❤
I am try to to just Generate a video sequence of a green frog jumping across to the right of the screen. The frog should start from a relaxed pose, then transition into a getting ready to jump pose, followed by a mid-air jump. Ensure the frog maintains the same color, size, and markings throughout the sequence. The frog should appear to be in motion, moving smoothly from the left side to the middle of the screen in a fluid jump but I have problems any ideas Thanks
It’s a good question. I’d break this into smaller, manageable steps. First, I’d create three still images in something like Midjourney. One starting image, then the mid-point image, and finally the end image. For this step is keep the patterns on the frog as simple as possible so that it’s easier for the AI to get it right. Eg. A single, brown mottled stripe along the full length of the frog… something like that. Midjourney has a great repaint option to repaint specific areas of the image, so if the original prompt doesn’t deliver the right results (texture, colour, pattern etc) you can repaint a specific area and specifically state that part of the prompt again. Once you have three images that you’re happy with, I’d use one of the video tools with start and end image options, and break the move into two parts. Video with shot 1 to 2, then the video with shot 2 to 3. Once you have two videos that you’re happy with (one for each half of the overall move), I’d stitch them together in video editing software. Love to know if it works 😁
cancel runway yall. its VERY, VERY.... restrictive about photos size and limitiations of prompt usage and won't refund crredits after already in process of making it. If i had to advise luma ai labs has been great and kling its somewhat around the ball park alright. I think for kling its model has only been trained to capture highlights of human movement and food interactions.
All these “token” based subscription pay models are complete scams, I don’t understand why no one seems to want to point this out. Consider how much money is being literally just burned simply thru the process of learning how to game the AI model to get it to spit out something that is even halfway useable. It’s f’n ridiculous and everyone is just eating it up like any of this has any value.
Hum... did you seriously took the videos from the demo to compare Gen-3 ? This is absolutly not representative of the quality of the model... People say they often get ugly results unrelated to the prompt with Gen-3.
You're right, of course, and that's what I was hoping to gauge when I did the "Runway vs Runway" tests towards the end of the video, entering the same prompts into my own Runway account.
@@HaydnRushworth-Filmmaker I understand, but the model is still better at generating videos with prompts like the ones in the demo. For example, for the ants, the zoom out and the village, it looks like they forced it to learn the different steps of the prompt. But it cannot understand similar complex scenarios.
I think you might be right, mate. I’ve been reconsidering that cancellation ever since I spotted the image-to-video upgrade. In reality, though, I’m really waiting for a next-gen video-to-video service to finally emerge.
Checked it out. Looks like you put a lot of work into that. Definitely one of the most surreal AI clips I've seen this week, but that's the kind of work that makes for great music videos.
Thanks for the comparisons really helps me understand where I am at. Yes I use LUMA and have posted some "micro movies" using LUMA and Nightcafe's suite of AI graphics algorithms. ruclips.net/video/VRwS9PQPIEY/видео.html What I would really like to see is an Image to Video comparison of these three and others.
It’s a good point, and probably for the same reason AI can’t handle accuracy in all kinds of things. AI can “view” mountains of images, but unless unless AI is “told” what each element of each image means, it has no understanding of WHAT it is looking at. If the information isn’t there, it will never know. This is doubly problematic if the human explanations of the meaning of the images are flawed by inaccurate best-guesses or misunderstanding. For this to be fixed, an AI company would need to work with reputable military historians who can help to train meaning and context into the AI.
![ARGENTINA vs. CHILE [3-0] | RESUMEN | ELIMINATORIAS SUDAMERICANAS | FECHA 7](http://i.ytimg.com/vi/De3AWmXzhuA/mqdefault.jpg)
For the comparison to be fair you need to regenerate the runway shots as well. Their showcase shots might have taken 100 generations before they selected the winners
I agree entirely about the showcase shots. I'm sure they were cherrypicked. Did you watch to the end? I did run a comparison with the same prompts typed into my own Runway account just to see what they would look like. Really interesting results.
@@HaydnRushworth-Filmmaker Unfortunately this is how most AI video/image generators do. They generate many videos/images and pick the best one. This might be how Sora AI is doing it so it might not be the "Perfect" AI everyone was expecting.
Love that you gave us the option to skip the video. That made me want to watch the whole thing!
😁… thanks. I’d like to say it was a cunning plan to get more views, but I figured it would be helpful for me if I were watching the video, so maybe it’ll be helpful for somebody else too 😁🫡
I've tried them all (Krea, Haiper, Pika, Luma and last few days Kling) while the latter is pretty good img2Vid it often does some weird stuff but now Gen 3 with Img2Vid is by far the best I've tried. I'm quite blown away. Not always perfect but coherence and quality output much higher quality. And to think it's only Alpha and getting better from here on! No freebie with Gen3 and nice that Luma and Kling offer some but for pro use the logo is a PITA. So my strategy is to have all my image assets ready for a project and take the unlimited month subscription when needed. (Went through the $35 one in no time)
That sounds like a fair plan, tbh. It's frustrating that we're left needing to pay $10 here, $20 there, and $35 somewhere else, just to cobble together results. I'm looking forward to the day that several, brilliant, all-singing-all-dancing tools become available that really do let you do everything in one place.
mate - you're delightful and it was a fun video. left a like and subscribed.
Heyyyy! Thanks very much, great to have you 😁😁😁
Great work exploring all the outputs with some nice comparison discussion. 👍
Thanks very much indeed :-)
Great work on your channel btw. I watched the walking gnome video the other day and it got me wondering what an animated version of my movie could look like.
Good comparisons man, Gen has strong text-to-video, but I was disappointed when trying out image-to-video today.
Does gen 3 even have Image to video feature?
Tao I am waitnig for you video abouy image to video - I do not know if it is worth to buy Gen-3 subscription:)
@@SNP2082 Yes,since 1 day! ...but unfortunately no end-image yet...
Thanks, mate. Big fan of your channel btw 🙂, incredibly helpful.
That surprises me, actually. I'll definitely be interested in your review on the Gen-3 image to video if you're putting one together.
I have un unlimited account with runway, a basic one with luma and played with kling and with all the other features, Runway is still my choice. No the new image to video is not great yet but the paint tools and camera are coming. I do music videos so for the price compared to a crew etc, really worth it, as a hobby meh
I want to emphasize the idea that these platforms should disclose on their pages and advertisements some metrics, such as the attempt-to-success ratio, so that a prompt’s output quality level can be assessed based on the examples in the ad.
Wooaaaahhhhh, now that’s a REALLY good idea.
@@RenatoFlorencia if only!
First time I come across your channel. Super interesting comparison and the wholesome moment at the end with the SPF 40 made me laugh! Very nice. You gained a new subscriber.
Cheers
A fellow director
Heyyyy, thanks very much, I really appreciate the encouragement. Always glad to have fellow directors on board too :-)
Thanks for this comparison. Indeed, things are changing so quickly that it's difficult to determine if there is a single company representing the state of the art, but my perception has been that Gen-3 has the edge; but of course that could change on any given day!
😁😆 it really does change day by day. As I was putting the finishing touches on this video, Runway literally JUST opened up text-to-video 😆
great video! sure runway is expensive, but it wins by a landslide in these tests. no competition at all. for now at least…
Thanks very much. I agree entirely that the quality award goes to Gen-3... like you said, for now.
Great comparison video, Haydn! I completely agree with your assessment of LUMA - it's indeed a powerful tool, but the cost adds up quickly. I've found myself relying on it less and less due to the expense. KLING seems like a promising alternative, but I'm eager to see more examples of its capabilities. One thing's for sure, the AI video landscape is constantly evolving, and it's exciting to see new players like KLING entering the scene, I use haiper AI for my videos. Keep up the great work, and I look forward to your next video
Heyyy! Thanks very much. I’ve actually started to realise that if I want to get better value for money from any of the new video tools I’m going to need to invest the time into studying prompt engineering, especially from the sources themselves (Runway, Kling, Luma… they often have their own prompt engineering advice).
I love the side by side! I just tried Kling image to video and even on the free plan it is impressive. Good to hear that RW3 has turned their image to video on. Would love to see a paid version image to video comparison similar to this fine video.
Thanks very much, glad you found the examples helpful. Of course, now that Runway has released Image to video, inevitably, I ned to try that out now, especially as I'm working on the promo video for my book and need to use image to video to create it.
That kid at the end looks so real!
😆😆😆
Thanks, the prompt was “cute toddler splashing about in the bath being bathed by his mother”
😁
if it good, I probobaly should get that Runway unlimited video generation plan
If you can afford it (along with the inevitable selection of other AI tools), then some kind of unlimited plan is definitely the way to go, then at least it’s only your time that you’re using up with endless re-rolls. I have a monthly budget set for the entire collection of AI tools I use, and if I spend money on one of them, that’s money that I can’t spend on another, so, for me, each tool really does need to work for its money in order to stay on the books.
Great aproach! Sometimes we can get a lot of credits failed in a bad video too, I'm thinking to upgrade from Standard plan to Unlimited instead, so I don't have to worry if I waste credits, and it's even cheaper than Kling or Luma for the same price =)
Yeah, that seems to be the case. It’s an individual judgement call, but if you have the budget it certainly gives you the playtime you need.
Love your fam! Thank you for doing this vid!
Anyway, agree on the cost per clip generation comment. For RunwayML, whether you're there for T2V or V2V, the unlimited plan is a must (if you can fork out the cash). The plan is really to use Kling to generate and chart out your entire story scenes workflow as an affordable first cut, and then subscribe to Runway for a month, and replicate each scene into 2K versions. sounds good?
Hey, thanks very much, really appreciate the moral support :-)
Re the cost-per-clip conundrum... BOOOOMMMM!!! I've actually been thinking the same thing... except I'd probably stay submitted to Runway and follow the same pattern on an ongoing basis. That said, if one of the paid plans with LUMA or KLING was worth it, I'd consider that too.
If you thought this was a lot, Vidu has entered the chat.
I find all of the video generators hitting market at the same time very odd.
Yeah, and I suspect the development teams are very similar with very similar outlooks.
Thanks for that. Of course the companies run as many generations as need be and then cherry-pick the best result. Keep Runway because I need it for lip-syncing. My main go-to is Haiper. $30 for unlimited 4 second generations, but you have credits for 8 second and upscaling. Luma, is, as you said, more bang and better bang for your buck. I also like that their credits are 1 credit = 1 generation, so you really know how much you have left. Now I'm starting to realize which tool is better for which kinds of animations.
Considering that it usually takes multiple tries to get a usable result, one really needs an unlimited plan, and they are just ridiculously expensive. So, I rely on Haiper and use the other services for my flourishes. How to be an AI video creator without breaking the bank.
100%
I'm not advocating only using free services, but it's irresponsible to "blow the budget" and pay top $ for every AI tool that pops up over the horizon. There's a real balancing act of risk/reward that we have to constantly try to balance.
Great comparison video!... And you and your wife are too cute 🥰🥰🥰
Hey Tasha, thanks so much :-) :-) :-)
honestly, as a narrative filmmaker I want videos that specifically adhere to what the scene requires not the wild stuff that everyone's experimenting with. it's confusing which of the subscriptions i should buy at this stage. also , any suggestions for which VIDEO upscaler i should use?
Glad to hear it’s not just me craving a tools that gives me the basic shots really well.
Re. which subscription, for me, at this stage, I’m still developing an ever-evolving workflow (which I need to make a video about), and since no, single tool does everything on the workflow, I figure if I can settle on a productive workflow, I can swap tools in and out as they evolve and improve.
In terms of video upscaler, I’ll be honest, I’ve never used any. Until I can get results that I’m happy with, I have no reason to upscale. A steaming turd is still a steaming turd, regardless of how high the resolution and quality is. I’d rather have low quality results that tell an emotionally vibrant than a high quality tool with no emotional impact in the story (or editing).
@@shireensingh2834 honestly I think you just have to wait. It seems like every iteration get better and right now, the archaic way of getting shots vs what some of us more traditional types expect just isn’t quite there. I assume it will be within a year, so start getting those shot lists ready lol.
Kling and Luma.. Rock head to head I use them every day
I think they’re fast becoming staple workhorse tools for me too 😁
The nerds are back! Missing the point of the story and using a microscope to pick holes, I wonder what Picasso would have answered. "The water on the window is larger than" Oh my. This guy is the Mother of all nits.
“The Mother of all nits”… I think I’ll print that on a t-shirt for myself, my wife would definitely agree 😁
You are making very good points about Gen 3, but their $99 unlimited plan is the best compering to cost per generation. It is still cheaper than the cost of the Hollywood movie. I think you would need both Gen3 and Kling if you want to create a movie. They are good at different things and you can use them for different purposes. I do agree that if you are on $30 plan with Gen3 that is very expensive cost per video.
Yeah, cost is a really important issue that everybody in film production needs to keep a tight handle on. Costs can get wildly out of control on a $150m project as well as a $150 project. I think Runway may have just cut their prices though.
Kling just started offering paid subscriptions, and they have a 50% off promotion happening. You might want to get a subscription and try the professional mode.
The sunscreen won't smell as bad as the WD-40. 😁
I like what you're doing. If I end up making some kind of creative endeavor with all this AI video stuff, you will have been the major influence behind that. I appreciate your obvious creative talent displayed in your videos!
I think you’re right about Kling. I got interrupted with kiddies bath time mayhem a couple of days ago when I was trying to set it up. Thanks for the reminder :-)
Told my wife about her better-smelling oil choice… she laughed 😁😆
Thanks for the encouragement and support too, really appreciate it 😁
@@HaydnRushworth-Filmmaker 😁
Just got into this text to video ai thing. However, unable to test Runway, when try to generate "due to high demand, generations on free plans are not available". How are we supposed to pay for a plan not knowing if it will work for us? Runway = fail. I tried Luma and it works great and will buy a plan soon. I think we already have a clear winner.
That sounds like solid reasoning to me. Minimise risk wherever you can whilst still learning (and beyond). We already carry a lot of the risk of failure for these tools, so I don’t think it’s unreasonable to find the least risky option.
If you make money with it, 100$ for unlimited is simply mind blowing and also the cheapest option. If not and it's just a hobby at the moment, I can understand.
Yeah, we're still firmly in the R&D phase with generative AI, and with so many AI tools charging for monthly usage, we all need to be really selective about what we pay for. The money can trickle away without us even realising.
Really like this channel. so much i subbed on my non trash account 🤣
Hope you more attention, this is really interesting for those of us who arent scared of AI
Well, since I’m subbed on your non-trash account I feel extra special, warm and fuzzy 😁😁😁😁
Thanks for the encouragement 👍🏻🫡
Runway Gen-3 image to video is here!
I know, it really did literally become available as I was putting the finishing touches to this video. Funnily enough, I'm about to re-subscribe because I need to use image to video more to finish off the advert for my book that I'm getting ready to promote.
I ran through my Runway account just doing lip sync
I can believe that, and I feel (and share) your pain 🤦♂️
@@ai-aniverse definitely making me reevaluate other techniques
Compare the best AI video tools: KLING, LUMA, and Runway Gen 3! Each offers unique features for video creation, from advanced animations to powerful editing capabilities. See how they stack up and find the perfect tool for your needs. #KLING #LUMA #RunwayGen3 #AIVideoTools #TechComparison
Yeah I been trying to figure this stuff a well confusing
Completely agree.
I have try all , I like Runway quality but overall performance of Kling is good
I’m inclined to agree. LUMA is great, but for the free plan you really do wait a looooong time for results. Kling seems to give the best all-round results at the moment.
Hello. I need help. How to remove the animation camera in Kling (make it static)? Just like in your video.
Hi. Sorry for the late reply. Do you mean that you want to create a still / static image in Kling? They do have a Text-to-Image option, is that what you mean?
Thats how i feel about AI subscriptions right now. You are also paying for AI failures too. I know its worst it will be today but i dont see a point in wasting my money.
Agree entirely, we really are paying for the AI shortfalls at the same time.
4:38 actually your footage is even better than the promo one
😆😆😆 thanks very much indeed 😁😁😁
can you share the prompt of your thumbnail ai image, thats look awsome 👏
Sure, no problem:
"a large robotic King Kong wearing square diamond encrusted sunglasses, kings robes and a very large crown stands menacingly and confidently with his arms folded facing the viewer. Angry, menacing, threatening. Dark grey background. High contrast."
As with so many AI images, the end result, whilst cool, didn't exactly match the prompt. I struggled and struggled and STRUGGLED to find a prompt that would make the gorilla robotic.
Also, I expanded the original image to enlarge the grey space for text.
The KLING KONG was a photoshopped version of the KLING logo that I screen-grabbed and then edited :-)
@@HaydnRushworth-Filmmakerthank you very much, usually for robotic you can use, cyberpunk, robotic enhancements, anthropomorphic, humanoid cyborg and so on. This I saw with good results. Again thank you, far too kind 🙏🏻❤
I am try to to just Generate a video sequence of a green frog jumping across to the right of the screen. The frog should start from a relaxed pose, then transition into a getting ready to jump pose, followed by a mid-air jump. Ensure the frog maintains the same color, size, and markings throughout the sequence. The frog should appear to be in motion, moving smoothly from the left side to the middle of the screen in a fluid jump but I have problems any ideas Thanks
It’s a good question. I’d break this into smaller, manageable steps.
First, I’d create three still images in something like Midjourney. One starting image, then the mid-point image, and finally the end image. For this step is keep the patterns on the frog as simple as possible so that it’s easier for the AI to get it right. Eg. A single, brown mottled stripe along the full length of the frog… something like that.
Midjourney has a great repaint option to repaint specific areas of the image, so if the original prompt doesn’t deliver the right results (texture, colour, pattern etc) you can repaint a specific area and specifically state that part of the prompt again.
Once you have three images that you’re happy with, I’d use one of the video tools with start and end image options, and break the move into two parts. Video with shot 1 to 2, then the video with shot 2 to 3.
Once you have two videos that you’re happy with (one for each half of the overall move), I’d stitch them together in video editing software.
Love to know if it works 😁
@@HaydnRushworth-Filmmaker thanks a lot for help I will try this God Bless
Not a fair comparison.
Do it again by generating the videos from the start.
It would be cool to compare img2vid of Runway, Kling and Luma
Image 2 Vid, Runway vs Kling vs Luma? Your wish, sir, is my command 🫡
(Actually I’ve just been planning it 😁😁😁)
Kling is king
Price and performance balanced
It's professional mode is exceptionally good for film making.
100%
On balance, Kling definitely offers the best overall value (at the moment).
cancel runway yall. its VERY, VERY.... restrictive about photos size and limitiations of prompt usage and won't refund crredits after already in process of making it. If i had to advise luma ai labs has been great and kling its somewhat around the ball park alright. I think for kling its model has only been trained to capture highlights of human movement and food interactions.
I have to agree about Runway. I'm torn, actually, because I'm obviously getting some great results, but the pricing... *OUCH!*
All these “token” based subscription pay models are complete scams, I don’t understand why no one seems to want to point this out. Consider how much money is being literally just burned simply thru the process of learning how to game the AI model to get it to spit out something that is even halfway useable. It’s f’n ridiculous and everyone is just eating it up like any of this has any value.
Interesting, really
Thanks very much :-)
you have to pay for gen 3 kiling is free for now
It really does make a difference when you’re still learning the ropes.
Hum... did you seriously took the videos from the demo to compare Gen-3 ? This is absolutly not representative of the quality of the model... People say they often get ugly results unrelated to the prompt with Gen-3.
You're right, of course, and that's what I was hoping to gauge when I did the "Runway vs Runway" tests towards the end of the video, entering the same prompts into my own Runway account.
@@HaydnRushworth-Filmmaker I understand, but the model is still better at generating videos with prompts like the ones in the demo.
For example, for the ants, the zoom out and the village, it looks like they forced it to learn the different steps of the prompt. But it cannot understand similar complex scenarios.
can you extend the video to 2 minutes in kling?
Doesn't look like it. Just 10 seconds as far as I can see.
@@HaydnRushworth-Filmmakeryes you can, paid tier
luma ai♥
I’ve had some really great results with Luma 😁
Which one is free?
Klin gall the way baby. I'm done with Runway and their expensive tier prices.
👍🏻👍🏻👏🏻👏🏻🫡
nice - I waiting for some video about gen3, image to video:) - you need to resubscribe:)
I think you might be right, mate. I’ve been reconsidering that cancellation ever since I spotted the image-to-video upgrade. In reality, though, I’m really waiting for a next-gen video-to-video service to finally emerge.
vidu is a good option for animated image to video generations and they're really fast at generating too.
Great tip, I’ll check Vidu out 😁
short AI video clip: 🎥 Run and Drive
ruclips.net/video/-BQreD5TsWA/видео.html
Checked it out. Looks like you put a lot of work into that. Definitely one of the most surreal AI clips I've seen this week, but that's the kind of work that makes for great music videos.
@@HaydnRushworth-Filmmaker I have 138 AI music videos and only 2 short AI movies.
ruclips.net/video/ZQYVJ3Ho5gM/видео.html
runway = money money!
kling= free free
fxk
😂😂😂 fantastic summary!!
Thanks for the comparisons really helps me understand where I am at. Yes I use LUMA and have posted some "micro movies" using LUMA and Nightcafe's suite of AI graphics algorithms. ruclips.net/video/VRwS9PQPIEY/видео.html What I would really like to see is an Image to Video comparison of these three and others.
🫡 Coming right up, sir 🫡
😁😁😁
Image to video:
Gen-3 vs LUMA vs KLING!!
Working on it 😁
Why the HELL can't ANY of the AI imagers or video do realistic military history correctly?
It’s a good point, and probably for the same reason AI can’t handle accuracy in all kinds of things. AI can “view” mountains of images, but unless unless AI is “told” what each element of each image means, it has no understanding of WHAT it is looking at. If the information isn’t there, it will never know. This is doubly problematic if the human explanations of the meaning of the images are flawed by inaccurate best-guesses or misunderstanding. For this to be fixed, an AI company would need to work with reputable military historians who can help to train meaning and context into the AI.