On the woman in the train shot the outside posts are going backwards relative to the train's motion on the tracks! They should be following the direction of the trees, right to left.
Cinematically speaking, these short clips can be strung together to form a complete movie. Most scenes in movies don't last more than a few seconds before the camera/scene changes.
Open source AI usually perform the best because it has more users from which to train from, from a much greater variety of prompts. If you pay for AI, you have a specific purpose, and the prompts won't differ too much. Also, less users interacting means the AI has less information to determine good vs. bad results.
I find Hailuo Minimax to be difficult. 1st, they are censoring anything that even remotely resembles an erotic image: "DOESN'T MEET COMMUNITY STANDARDS" and yet I have seen tons of videos like that produced from the site (girls in bikinis dancing on the beach) on YT, and I even used the same prompt but got the morality police warning. 2nd, I have struggled getting it to follow a prompt like, "An alien spacecraft blows up an old farmhouse" (I was trying to do a "war of the worlds" trailer). It fired at it, it fired near it, it blew up things around it, but after 10 tries it would not blow up the farmhouse. What part of BLOW UP FARMHOUSE could it not understand? All of this, and 5 seconds at 720 makes me want to look elsewhere.
I tried it with a Godzilla type monster stomping down a city street text prompt. Try #2 was great but subsequent attempts with almost the same text tweaked slightly produced worse and completely unpredictable results . Weird
Tx for sharing. Some good news amongst it all. I think if one focused on this model, a student of this craft could learn to drive it extremely well, given some time at the wheel..
@@VortexBH i think atm most people are willing to pay for better looking output and img to vid. not to mention you cant just run this on any pc. so cloud based gens are more feasible for the average person.
8:29: There is another flaw with the woman on the train: The signal sign passing the window outside is moving into the wrong direction. It looks like if it is passing over the train with higher speed. 🙂
well looks like your video got them enough traction to lower the number of free video generations.....I used it for apx 30 min, had it make 2 videos... using your gpt prompt approach and the results weren't very accurate and then it said i had reached the limit and would have to come back in 5+ hours ...so much for 4 tempts.
@@RDUBTutorial text to vid gives very mixed results when trying to create something consistent. I remember trying it in minimax , it wasnt great. My outputs were all over the place, nothing made sense. Probably useful for 1 offs ...
That white sign was really flying past the train... A few good point, so far each model has it's pluses and minuses. If they beat Runway in image to video...maybe I'll switch.
Yes - I found this and I was searching for another quick mind guy "The model requires at least 4 H100 GPUs to run." I started to download it and saw 40Gb at the file dit.safetensors For now, I will keep my very expensive 15 USD payment to runwayml 😅🤯👽
I watched your video now and headed over there to try this new model, but I got an error message: "Uh oh! Error generating video." We are currently experiencing high demand, so your request was dropped. Please try again later." Edit: It's generating now; thanks for your amazing content, as always
is there AI tool that could help me with making my video enhanced for example changing the background or adding some artistic effect or changing video style I want to edit my dance videos to make it more cinematic and artsy
Yes, I will be able to edit better effects in movies like Terminator, for my own personal use of course. AI Gens are perfect for improving special effects in older movies. Literally modernizes them or even improves them.
Nice click bait but Runway unlimted is a much better deal than this PAID tool plus runway has a lot more features such as image to video , video to video, text to audio etc. Free mode doesnt even work for this trash as the servers are bogged down
But so this means I can do this freely unlimited locally in stead of spending $200 monthly for credits on all the video generate websites and still be limited ? It's about time !
Haha 😂 at the Will/spaghetti video. This is certainly an interesting tool! Image to video will be extremely important, but the realism and the possibilities look great. Also, they’ve got to figure out a way to provide higher quality outputs. Great video overview and comparison as always. See you inside your community.
The eyes of the man in 3:09 are morphing The woman on the train: it disturbs me that the telegraph poles are moving in the opposite direction to the background
It's strange how many things I notice that he never mentions. I know different people will pick up on different things, but he's experienced at this and looking at it with a critical eye, but still glosses past obvious inconsistencies.
Had to skip over Will eating spaghetti segment.Every since Will Smith Punk Slapped Chris Rock, I don't want to even look at him. Chris is a comedian, so jokes are to be expected. If you can't behave like a human, you are not getting me to come to your movies.
What Smith did was absolutely wrong but Rock wasn’t innocent. Just because you call it a ‘joke’ it doesn’t mean you can say anything, especially when it’s about someone’s medical condition. Rock should have behaved like a respectful human too. It didn’t deserve the reaction it got though, loudly calling him out would have been enough.
Will was absolutely right in his behaviour. If you don't understand how he feels about his wife, so you just a child, not man. Not even human, you just artificial intelligence with words instead of real soul and mind.
Ain't free. I used up my free credits super fast without hardly doing anything. Wait. I spoke from experience before watching the video. Now I'm watching. Open source. Hmmmm.
Such pieces of an advanced tech are used mostly for video creation. As always, people want bread and to watch some show, while eating that bread. I am interested in artificial intelligence as a problem solving tool, but video making is not a problem. People are stuck in the unreal world of imagination through videos. Too much of procrastination.
hi thank you my friend we need ai text to video generator that has sfw and nsfw i personally want to generate an nsfw video i need an good ai text to video generator that generate nsfw videos thank you. and also they must increase the length to the at least 5 min
Please keep us updated when Genmo releases *‘Image to Video.’* That’s when things will get more useful & interesting.
ai-videoupscale AI fixes this (AI Video Upscaling (to 4K)). Free alternative to expensive runway.
Link please?
On the woman in the train shot the outside posts are going backwards relative to the train's motion on the tracks! They should be following the direction of the trees, right to left.
Those are robot helper posts that travel along the tracks to check for damaged tracks.
Thank You AI Samson; this channel and 'How to in 5 Minutes' keep me up-to-date on the latest in my Music Video creation venture!
Cinematically speaking, these short clips can be strung together to form a complete movie. Most scenes in movies don't last more than a few seconds before the camera/scene changes.
Open source AI usually perform the best because it has more users from which to train from, from a much greater variety of prompts. If you pay for AI, you have a specific purpose, and the prompts won't differ too much. Also, less users interacting means the AI has less information to determine good vs. bad results.
If it’s a blind comparison, how do they know which is more clear and higher quality?
So if it's 480p, what can I use to upscale it into a much better resolution? I'm having a lot of fun playing with prompts though! Thanks a lot man!
I find Hailuo Minimax to be difficult. 1st, they are censoring anything that even remotely resembles an erotic image: "DOESN'T MEET COMMUNITY STANDARDS" and yet I have seen tons of videos like that produced from the site (girls in bikinis dancing on the beach) on YT, and I even used the same prompt but got the morality police warning. 2nd, I have struggled getting it to follow a prompt like, "An alien spacecraft blows up an old farmhouse" (I was trying to do a "war of the worlds" trailer). It fired at it, it fired near it, it blew up things around it, but after 10 tries it would not blow up the farmhouse. What part of BLOW UP FARMHOUSE could it not understand? All of this, and 5 seconds at 720 makes me want to look elsewhere.
I tried it with a Godzilla type monster stomping down a city street text prompt. Try #2 was great but subsequent attempts with almost the same text tweaked slightly produced worse and completely unpredictable results . Weird
Totally agree after testing like you did
Tx for sharing. Some good news amongst it all. I think if one focused on this model, a student of this craft could learn to drive it extremely well, given some time at the wheel..
Hailuo looks alot better I think, sharper and more natural camera movement, plus it has image to video but yeah nice anyway
Yeah but it's from china and you know what that means...
yeah but the point is that this is open source, unlike minimax
@@VortexBH i think atm most people are willing to pay for better looking output and img to vid. not to mention you cant just run this on any pc. so cloud based gens are more feasible for the average person.
Thanks for video I was wondering do you have a link to that GPT you made so we can all make the most of Genmo
8:29: There is another flaw with the woman on the train: The signal sign passing the window outside is moving into the wrong direction. It looks like if it is passing over the train with higher speed. 🙂
Samson cut his hair :( hope you keep your youtuber's super powers. Love your content.
I hope so too!
well looks like your video got them enough traction to lower the number of free video generations.....I used it for apx 30 min, had it make 2 videos... using your gpt prompt approach and the results weren't very accurate and then it said i had reached the limit and would have to come back in 5+ hours ...so much for 4 tempts.
This is always an issue once a new service gets coverage, It either reduce free credits, increases time to generate or both.
Can u use image to vid?
@@armondtanzI guess I could but it’s not really the same. Isn’t that more like take a photo and animate it?
@@RDUBTutorial text to vid gives very mixed results when trying to create something consistent.
I remember trying it in minimax , it wasnt great.
My outputs were all over the place, nothing made sense.
Probably useful for 1 offs ...
That white sign was really flying past the train... A few good point, so far each model has it's pluses and minuses. If they beat Runway in image to video...maybe I'll switch.
Genmo Mochi 1 requires FOUR H100 GPU's at a cost of £120,000 so good luck running the official model locally.
Yes - I found this and I was searching for another quick mind guy "The model requires at least 4 H100 GPUs to run."
I started to download it and saw 40Gb at the file dit.safetensors
For now, I will keep my very expensive 15 USD payment to runwayml 😅🤯👽
Great points of criticism for sure. A few years and these "Issues" will be long gone. Thanks for the interim analysis.
Please let me know when there's an open source program like this that allows image to video. TY
I think this will have it in less than a week
@@aisamsonreal excited!
Usual AI video, runway "killer" here. Wait, Minimax already killed runway
On the train shot the white sign that flashes past is travelling the wrong way.
Happens a lot
W thumbnail
2:56 Btw, human eyes movement is also difficult.
How long does gnemo take to generate 5 seconds of video ?,
on a pc,
on nVidia high end graphics card ?
Your hair looks amazing!! Thank you for all these tips; you’re always incredible.
Groundbreaking would be a text2video that is over a minute long, installs locally and doesn't crush your machine in the processing of the video
Probably the alternative is to batch image to image frames from an actual video you like.
I watched your video now and headed over there to try this new model, but I got an error message: "Uh oh! Error generating video."
We are currently experiencing high demand, so your request was dropped. Please try again later."
Edit: It's generating now; thanks for your amazing content, as always
unlimited creations and amazing quality and features... I'll pay!
is there AI tool that could help me with making my video enhanced for example changing the background or adding some artistic effect or changing video style I want to edit my dance videos to make it more cinematic and artsy
Yes, I will be able to edit better effects in movies like Terminator, for my own personal use of course. AI Gens are perfect for improving special effects in older movies. Literally modernizes them or even improves them.
It's a shame it requires 4xH100s for "local" generation of 480p 30fps video.
Nice click bait but Runway unlimted is a much better deal than this PAID tool plus runway has a lot more features such as image to video , video to video, text to audio etc. Free mode doesnt even work for this trash as the servers are bogged down
for me minimax.. and a little bit of kling and pika for some effects
Until such things can be run on your home PC with a simple installable GUI, it's all just fairy dust really
The pricing looks good but the 480p output is not really a great news 😕
@8:20 did you not notice the strange white post overtaking the train outside? (ie travelling in the wrong direction) lol
Nice haircut nice new look
It could just be AI :)
I just watched a 30 min vid and still dont know if genmo does image to viddo 😢
5:50 Interesting how the extra pair of chopsticks are phasing through the plate.
But so this means I can do this freely unlimited locally in stead of spending $200 monthly for credits on all the video generate websites and still be limited ? It's about time !
nice hair bro much better than before
Doesn’t this need 4 Nvidia H100’s to run?
It failed 3 times creating a video and then all my credits (why are you saying it's free???) were gone..
Any way to run it with google collabs?
Haha 😂 at the Will/spaghetti video. This is certainly an interesting tool! Image to video will be extremely important, but the realism and the possibilities look great. Also, they’ve got to figure out a way to provide higher quality outputs. Great video overview and comparison as always. See you inside your community.
Appreciate the support! Can't wait to see what crazy things come next
This blind testing site is intriguing... I'll try it, of course.
You didnt include the link to the ChatGPT prompt.
here you go ruclips.net/video/c86ZyqJn6Q8/видео.html&t
Hailuo is the king for the moment.
The eyes of the man in 3:09 are morphing
The woman on the train: it disturbs me that the telegraph poles are moving in the opposite direction to the background
It's strange how many things I notice that he never mentions. I know different people will pick up on different things, but he's experienced at this and looking at it with a critical eye, but still glosses past obvious inconsistencies.
5:52 the boy on the left has an extra digit on his right hand
Had to skip over Will eating spaghetti segment.Every since Will Smith Punk Slapped Chris Rock, I don't want to even look at him. Chris is a comedian, so jokes are to be expected. If you can't behave like a human, you are not getting me to come to your movies.
Will eating pasta is the official AI video benchmark test
What Smith did was absolutely wrong but Rock wasn’t innocent. Just because you call it a ‘joke’ it doesn’t mean you can say anything, especially when it’s about someone’s medical condition. Rock should have behaved like a respectful human too. It didn’t deserve the reaction it got though, loudly calling him out would have been enough.
Will was absolutely right in his behaviour. If you don't understand how he feels about his wife, so you just a child, not man. Not even human, you just artificial intelligence with words instead of real soul and mind.
What it has to do with the Ai video topic?
@@harisbinsaeed It's indirectly related through the video clip of Will Smith.
"Prompt adherence" is such a non-descript metric. There is no way that can be a measurable scale.
We are literally training these models how to create the Matrix.
Your code appears useless. At no point in either website offers to enter it to double their credit.
Ain't free. I used up my free credits super fast without hardly doing anything. Wait. I spoke from experience before watching the video. Now I'm watching. Open source. Hmmmm.
Why is there a board overtaking the train where the woman sits in the train? Lol
11:30 Those aren't tears, they're sweat. See them on his forehead? I swear I think you're trolling us just to get comments! lol
It only does 2 generations a day, and max 30 in a month!
Need img to video with genmo...
Such pieces of an advanced tech are used mostly for video creation. As always, people want bread and to watch some show, while eating that bread. I am interested in artificial intelligence as a problem solving tool, but video making is not a problem. People are stuck in the unreal world of imagination through videos. Too much of procrastination.
Why not skip the fluff and run some prompts?
"The model requires at least 4 H100 GPUs to run." 🤢🤢
Great. But... still waiting for the image to video
I love how the guys eyeball is going full demon mode @3:01 and he doesn't notice.
First of all thanks for keeping us up to date with the new developments on video rendering with AI.
Noticing some weirdness on the eyes at 3:07.
Thanks 🥰
Looks promising I'm tired of paying
Bro you cut your hair!!! SAMPSON!!!
...sounds super!
BTW is that a new haircut or a generated version of you? ...Anyway: Big FANX!
MOP CHOP!
5 seconds long video is just a waste of time... forget it.
Right..these ai apps r real mockery
hi thank you my friend we need ai text to video generator that has sfw and nsfw i personally want to generate an nsfw video i need an good ai text to video generator that generate nsfw videos thank you. and also they must increase the length to the at least 5 min
This!!! If you ever find one can you please let me know🫶🏼tysm ❤
very, very, very, very slow
yes, when realitime on a 5090 :D
ขอบคุณครับ 🥰😍❤
But minimax has unlimited plan unlike genmo.
You need to install it locally on a computer with a high VRAM if you want to use it without any limitations.
@@xenn2996 Genmo lists the hardware needed as 4 H100 GPU's. They are £30,000 each.
Low censorship? 😆 It's the opposite of censorship. It's just making shit up.
hands are all morphy tho
Cool hairstyle
the software is free, the gpu is not.
A LOT of morphing
Hi
niceee
❤❤❤
Well, I hope someone doesn’t create something defamatory with the uncensored version, as it could result in the US restricting AI.
The food gates haven’t opened yet.
give me