See? You learn stuff here." I always feel like you are just chatting with us extemp about our favorite subject, even bringing up in-jokes like the man in the blue suit and his wife in the red dress divorcing him for too many jaywalking tickets.. I really appreciate the way you don't overhype stuff like Runway or Adobe or even the latest new editing feature of Midjourney. Your ai updates are the most fun and helpful I've found and I watch 'em all.
As someone who has written five scripts using Claude 3.5, I'm pretty certain script A is human, and script B is AI. A has a distinct voice and pacing that B doesn't. A uses non-standard slugline annotations (I./E., rather than INT./EXT.) that break the rules in a way the AI doesn't. B abuses "wrylys" and puts longer character descriptions in parens which is typical of AI output. The dialog in A is sharper and edgier than B, which is competent but not especially imaginative. The action lines in A use sentence fragments and caps for punch and to draw the spec readers attention like a pro writer would do. B uses more complete, grammatical sentences which slow the read and lack punch. These are some of things I have to correct when using Claude by specifying detailed style prompts and exemplar's of my own writing style to add voice to the generated output. LLMs are like a reverse Turing test. If you ask it unintelligent questions, you get unintelligent answers. If you know what a good spec script requires and ask Claude nicely, it can surprise you with its creativity at least for a first draft.
Agree. I read the first line of Script A and instantly thought that there is no way an AI wrote that line. On the other hand, Script B's first line sounded exactly like AI. That said if they tried hard enough for enough hours I'm sure one of them could have gotten Claude (not chatGPT) to write about a nip slip.
I didn't have the patience to read the full scripts. I started with A and read the first page and a half, then went to B. Honestly, neither of them was good, but I got through the first two pages of script B a lot easier than with A. I can't say which is more likely to be human vs AI. They seemed pretty comparable to me, which doesn't bode well for the human writer I would say going forward. Lol. Part of the problem for me though is that reading a screenplay vs a regular story feels very unnatural. I would much prefer an AI that could write competent sci-fi or fantasy stories.
I love to discuss the question which could be ai and which could be human. After reading both scripts completely I came to the opposite conclusion that A must be ai and a little bit edited and B must be human. A is going straight forward to the task, while B started to introduce both characters in their own way in non boring scenes first. A has too much dialogue, while B has more „show, don‘t tell“ aspects and subtext. For example the „chinese security lady“ had the opinion that the popstar‘s security precautions are too lax. It‘s also shown by her reaction, after she noticed the lame ‚6969‘-security pin. In script A both characters starts with a relationship to each other and the girl is good with her bodyguard. This means there are less potential for conflicts between both. I also find B much more exciting. I want to know what the reason is, why someone try to shoot the popstar. In A they just run in a mafia situation by accident and there is nothing different, if two other random people(no popstar) run into this situation. I am not interested to find out, why yakuza killed a random cashier. B doesn‘t include a murder yet. I think, the author prioritized his own ideas first than the task. And in his idea, the two characters need to meet each other first.
@@musicinthemachine B goes into the absurd almost from the start (the bodyguard beating up fans for no reason) and ends up with a random casher in a random store being murdered by Yakuza of all people. If a person wrote it, they should be ashamed to post this publicly. If a person was assisted by AI, they should also be ashamed to post this publicly. B isn't spectacular, it's kinda safe and boring, but it works as a movie. Both characters are introduced properly, the main plot line is introduced properly, and the tension between the characters is set up properly (the bodyguard was a fan of the star). With A, the age difference is too big for a (non icky) romance, not right for a father-daughter type relationship. A is just nonsense.
Did you EVER think you would hear: "We just ended up with an upscaled version of Ghosting" and it would actually make sense? Life in an AI world has stuff so weird you couldn't even dream it up LOL Great job as always Tim!
I still am amazed at how good midjourney is at cinematic. I've been working on an overlook hotel and the ones in flux look like the Ritz or the plaza. MJ just captures the 1920s aesthetics, the dramatic light, shadows, lock stock the lot. I did notice it absolutely struggles with mazes, where the flux model can produce detailed hedge mazes easily.
Oh, I know-- actually, the image UNDER the prompt that I typed out is actually MJ. Haha, it really did deliver on a straight up Overlook Hotel. Which...if I remember, I was being super lazy, and that was basically the prompt.
So much going on! It was actually nice to slip away to the land of Danny Torrance for 3 hours to catch a (super tense) break! Man, that movie was good.
Jay walking is no longer illegal in NYC as of I think just this week... so maybe the man in the blue business suit and the woman in the red dress can get back together? Reunited and it feels so good.
oh, man-- I'm so bummed I couldn't make the live premiere-- since oddly, I was working on this video! But yeah-- that was a TON of fun. SOOOO much good stuff there! That was probably the toughest one of those things I've done yet!
You left out Topaz Gigapixel 8 diffuser model. It crushes Magnific and Krea just because you can run it locally, and up to 16X, which none of the other models can do. Give it a try and you’ll be amazed by it. But wait, but there’s more! It’s coming to Topaz video as well!!! Without limits of the infamous 10 seconds.
@ Cool beans! Honestly, I tried it out this past weekend and is really better than Krea and Magnific. You also missed the new Video to 3D Scene from Wonder Dynamics. This is perhaps bigger than anything you showcased in the last 6 months.
@@MabelYolanda-c9i Video to 3D Scene if i understood it correctly is just a Camera Tracker + Motion Capture in one go. Which is convenient but not revolutionary. We've been doing this for years)
Easily, script A is human, and script B is AI because at least the closed models don't allow cursing and generally won't produce any violence if it can get away with it. Script B seems to pretty low stakes stories in general, which are generally the stories that AI make.
As someone who regularly has chatgpt scripts written, I can tell you that it has recently been allowing much more violent text. It does say somewhere that the generated text or request violates the guidelines, but they still leave it up for some reason.
@@sasbe1852 Well, I was assuming that the entire script itself was written by AI, and had little to no human input versus a script that was completely written by a human (so no restrictions). Yes if a human prompts it, it will allow violence and such.
Thank you for the shoutout, my friend. Based on my experience, it appears that SellerPic AI may have downgraded their model. The output no longer aligns closely with the prompts, and face distortions seem more frequent compared to before when the results were more realistic.
I welcome you. The video is very good, it helped a lot to navigate. I would like to ask how it can be set in the text description so that the video does not move in several directions? I mean that it should be in a side view and in a fixed position.
that's a solid go-to right there. I also (oddly) recommend trying v6 as opposed to v6.1. Six, for some reason or another, tends to behave a lot better on that front.
Books of Magic. Kind of a Sandman offshoot Vertigo did in the 90s. Totally random page. I've always just wanted to own a little piece from the analog era of comic book making. It's kinda cool, you can see tape marks and the blue pencils-- I got it off Ebay years ago, but only just got around to framing it! hah
@@TheoreticallyMedia Nice! And yeah, actual comic book pages will soon be hard to find. I've got art from the first bunch of comics I wrote on, but that's not possible anymore. That said, I do appreciate the speed of digitally drawn comics, and the ability to change anything on the page without much effort. Makes the art directing/publishing process so much easier.
to be honest, I think I visually process while I'm talking-- but, usually only on camera? I think it's because I feel like a psychopath staring directly into the lens.
I must say your reviews are quite inclusive and I look forward to new episodes. Side note" free King is now not performing in the free version. They claim over 700 hours to create a video so they have it on hold unless of course you are a paying customer.
Hey, have you tried Freepik and their integration of Magnific? I used you prompt for that lady in armor, I added a huge bear to and chose their model Mystic and Bam! Damn... :O
Just the fact that the A.I script at this point competes, spells DOOM, now studios have decades of bought scripts, feed them into a specialized A.I dedicated program, GAME OVER, output.
How are the pricing model at maginific these days? I was so impressed by the upscale when i tried it for trial version but stopped using it because of pricing. But if i can now cancel my other subscriptions and generate directly in magnific then this might just be worth it.
I think the monthly price is still the same, but the overall cost of runs is cheaper. So, you get more for your money now. I’ve had a few talks w/ the big guys at the various AI platforms, and I guess that’s the way it works with most of those cloud GPUs. The more you run it, the cheaper it gets. So, there’s this double edged sword for them: it costs a lot for earlier adopters, but can get cheaper as more users start running it. I presume that’s why a lot of them end up folding 3 or 4 months in. If they don’t get that audience wave, the cost of GPUs eventually crushes them.
I do write quite a bit, and I do use ai aswell. but I can never give ai the credit that it can create a good story by itself. it needs a massive amount of involvement and direction, which I dont mind.
Yup I second that. I'm great at coming up with ideas , but the absolute noob at sentence structure. This is a god send, I just throw all my bullet point ideas in and it can smooth out the edges. It's great for cheesy puns which are hard to come up with. 99% are garbage but every now and then you'll get a gem. Asking it for variations or give me 10 examples is my number 1 goto.
@@TheoreticallyMedia and has sugar related mind fog where they forget plot points a few hours long. honestly thats fairly accurate to having an actual assistant, except a little more enthusiastic.
@@armondtanz oh yeah!, developing ideas is definitely it's strong side. though it really got no chance of handling an actual full story script. personally I dont mind it as a "very eager intern" who also has special-needs. carrying more weight is good for me. it should only iron out the kinks. otherwise I'll lose my muscles.
@steve-g3j6b it is getting better. Maybe one day we can move onto asking for advanced ideas. Be great if they could introduce movie catalogue , like a folder with all your fave movies. It could then analyse everything from cinematic, costume, dialogue, camera, coloring. Then you could have sliders saying how much of each you want. So if u wanted a sci fi meets old skool disney , you could tweak and lean into many of the aspects that makes that genre work.
as long as you add "8k, ultra realistic, IMAX, Ray Trace, Octane, RTX ON" to your prompt, I think either will work. haha...that of course, is not true!
Magnific not being honest is normal. Upgraded UI just got grid view which is a simple integration for a dev. They need to take some of that money and build out a better UI. Mystic is yet anothe rmoney grab as the wonkey eyes shows the quality of the model.
Well, they've integrated w/ Freepik at this point, so I'm tending to think that Magnific might be the test bed for what we'll eventually see in Freepik? I do give them a break on the gird view thing-- mainly because Magnific is actually just 2 people. So, one REALLY overworked Dev!
@@TheoreticallyMedia recently had a little opinion exchange about the perspectives of AI visual generation and realistically it doesn't seem to be that bright. It feels like we're somewhere near of hitting a plateau. And then i stumbled across a "Has Generative AI Already Peaked?" video which has some serious research paper to support the point.
@AarreLisakki First of all, i was talking about visuals not ChatGPT-like text models. Secondly, we don't see major improvements in - hands tracking, object tracking, prompt understanding, consistency, massive undertrained areas, etc. Lastly, if someone was shouting wolf before it was there it doesn't make current concerns less plausible. Especially when it has a scientific base not just someone's premonition. Majority doesn't understand neither technical side deep enough nor the current state of the models to take their personal opinions into account anyway which you cannot say about researchers and people who work with multiple models intensively.
Nailed the script detection in under half a page. Basically, the AI copies how most scripts are written. Most published scripts are shooting scripts, ie, scene numbers. Additionally, AI tends to tell, not show. And dialogue tends to be extremely on the nose. AI's don't really understand subtext. The AI script might be better but it would need serious work to fix the issues and add in all the subtext. But it is getting there. Once AI's understand subtext, and show instead of tell, and get character consistency and pot consistency right, then it's going to be used a lot until it gets stale and the audience gets sick of AI stories that will tend to be pretty generic and formulaic. It's not about being paid for writing. It's about seeing the vision come to life. Back to writing my script.
You forget to tell us which script is ai and which not. Because A has more dialogue and less subtext in my opinion I hope you meant A to be ai written?
I'm going for Script A as the Human one. Trivial reason. VMA NipSlip comment sounds implausible for AI to come up with. There's a human brainswamp behind that one, in my estimation.
Script A was by human (I think), while I like though the bot version. Just seems more digestable and meaningful while reading. I can almost see the scene as I read (I guess maybe my understanding of the English slange in Script A was too hard to me to comprehend).
Hmm, script B seems more human to me, while it may be more ‘boring’, it flows way better and is more consistent, and it does seem like it could be an interesting story if continued. Script A on the other hand seems to jump into fights a lot quicker and randomly in a way that reminds me of AI hallucination…. 🤔
Interesting, I thought it was the other way around. Script A just seemed to be written by someone who doesn't know how things work in reality, which seems very familiar from modern screenwriters.
If they had challenged some rando on the street, then possibly. (I’ve definitely encountered some really badly written fan fiction before. lol). But my understanding is that the human challenger is a professional scriptwriter? 🤔
@@ThetaGraphics if A is human, it means the screenwriter guild deserved to vanished. I am still confused about the energy drink joke in it. This is like something, only a human would write. And after I give chatgpt the given logline, the generated first scene had more parallels to script B than script A.
@@MarionMicheleFilms But isn't that how a lot of Hollywood screenplays written by humans seem? And I found script B to be quite cliché as well. I'm very interested in the reveal.
Where do we find the answer to the screenwriting challenge? I read A first snd was convinced it was human. Having read B, I'm not so certain. Both could easily pass as human. I do think A was better, either way.
do you know of any good videos or articles or sites that get into a detailed technical description of exactly HOW generative AI works? Both for stills and video? I mean what do all the developers for each of these companies do exactly and how does the tech work? What software/hardware/tools are they using? And how are the different styles/aesthetics/models developed? What changes do they have to make from where it was say a year ago? I can find virtually NOTHING on the internet that gets into that technical aspect of it. Does ANYBODY know? Because nobody seems to know
I started with B first and as I read the opening descriptions, I thought the language was way too flowery, so I thought it was AI... BUT... In the end, I thought B had much more sophisticated, subtle humorous setups and A seemed like it was written 10 years ago with the much older, dated tropes and references. So I'm calling A is AI and B is human. I don't think AI is that good at humor yet. Maybe I'm wrong... if so, sorry human.
great analysis-- but, I think the most telling thing is that you have to say "maybe I'm wrong" we've come a LONG way in a short time! Scriptwriting in ChatGPT3 was so totally obvious! And that was-- what? less than a year ago?
A script was written with so much dark humor, I doubt and would be very surprised if AI wrote it. It was GREAT!! Script B was formulaic, cliche, and UNREADABLE after 3 pages which were more than I wanted to give it. Am I wong? Now that's human humor..
I was kinda thinking the opposite...script A seems over written and some of it illogical. Like why would the body guard plea with the crowd if he's so tough (and then it said he knew it was going to go that way...)?! 😃
If the scripts have not been edited, 'Script A' must be human, because, well, the word '$#@%nuts' is on page 1. If the LLMs _have_ started using that language, we're in for a good '&#@%ing' time. I don't know if you remember us commenting about 6 months ago, about how 'AI videos are not quite there yet'. Now we're zooming in, stopping frames, divorcing couples on the beach and the streets. Come a long way. And Tim, let me help you/us a little with Mystic 2.5beta. It allows mild NSFW. So, if you want to generate 'fine nude photography' and need to breathe outside the MJ ecosystem, go, people, go! I haven't tried that yet, of course. I would never. Magnific is finally offering more for their price, and that's fine by me. Thanks as always, and so good to watch a longer video. 👍
I sense a script coming with a mini series.. "The jaywalking man in a blue business suit and the girl in the red dress on the beach" (catchiest of titles). Why did she leave.. or more so, why did *he* leave?? I'm sure he knew he was jaywalking. .. unless.... hmmm. And why the beach? Who is that red dress for? .. or more so, who does that red dress belong to??! wait - is the man a real man or ... ?? - Lord I want to watch that so badly now.
It's getting so difficult to keep up with all of the changes. No more yearly subscriptions for me. Going month-to-month so I can change/keep up with the greatest and latest image/video generators!
I feel like a film student set that whole screenwriting thing up as a trick to get us to read his/her crappy screenplay. Cilantro is delicious and the soap people are twice as nefarious as the AI predictions of doom and gloom getting passed around. The man in the blue suit tries to get his nails done but he always walks into phone shop next door, to the left, due to his eyewonk issues... His wife stands one place over to the left, for this reason (after 3 frustrating years). The people working at the Verizon store next to the nail salon are sick of it, to be frank, and are starting to suspect AI. I think he breathes in the steam of children but that's out there.
Script A is the human, I feel it’s quite obvious because how much swearing is in script A. If I’m wrong then the promoter intentionally threw those in to throw us off
I agree. The name of the pop star is also much less stereotypical in script A. Script B played everything a bit more "safe", but to be honest, I actually liked script B better. They are both not masterpieces, but A gave me some serious edge lord vibes and I felt like people acted a bit weird at times. But all in all I think they are of similar quality, which is pretty interesting considering the circumstances.
I think NoFilmSchool is going to publish the curtain reveal soon-- but, I'd tend to think you're right. That said, the fact that we aren't immediately calling one out as obviously AI? That says something.
@@TheoreticallyMedia I'm guessing someone with no AI experience would find it difficult to identify it. And that's almost all of the general public. That's it's value at this point.
I read both scripts completely and I say Script A(left) is ai generated. It had many random scenes thrown together, while the pathing of Script B is much more typical movie level. Also Script B missed a very important point of the task. In 99% of the cases AI's dont forget the main task.
@@sasbe1852 AIs don't forget the main task?!! How long have you been playing with AI? And you obviously could be 100% right, of course. It's all interesting fun.
@@thanksfernuthin After several prompts it could forget the goal and maybe the ai users used several minor prompts to complete the first 10 pages. But the goal wasn't really complex. I would say, AAA llm's would not forget this that fast. Also I think the human had several ideas in his week so he set the priority of the witnessed murder(there is none in script B) lower.
As it's fantasy you could explain the face change and ghosting as her about to unleash hell on her followers. She's digging into her darker side. Not ideal of course. I've seen and liked Doctor Sleep. In your opinion, is the 3 hour cut worth it, or is it padding?
Oh, it's 100% worth it! Much more novelistic? I wouldn't recommend going at it in one go-- in fact, the movie is structured with chapters (like a book) almost encouraging you to watch it in chunks. It almost feels like 3 episodes of a big budget Netflix show. (Which, to be honest, I'd probably binge watch anyhow...)
@@TheoreticallyMedia Thank you kindly for that. I like to watch new movies in one go but once I've seen them, I don't mind breaking them up. Cheers and have a great day.
I only had to look at the 2 script images and speed read them to see the first is definitely human written and hilarious while the second is generic and without life. Hey maybe I’m wrong. I’ll be surprised if I am.
@@TheoreticallyMedia To be fair, the other script reads like an amateur fanfiction so I really hope I'm right in thinking script B's the human-made one, even disregarding the error 😅
Just going by the first page the on the screen, I'd say the left one is AI. Can't imagine a human writing a girl scout as leader of a bunch of fans, plus the bodyguards opening comments are way too wordy. Could be wrong though!
oh, that's ODD! I didn't realize you could "grow out" of it! That's really interesting-- Are there other tastes you've found that you developed? Hmmm, might need to try Black Licorice again. ...nope. Still awful.
@ Kinda all occurred when I cut sugar out of my diet and went keto. I can also stand spice heat way more than before. Habaneros are like Bell Peppers to me.
I think Script A is the AI generated screenplay. As I started to read it, my first thoughts were it was written by a human, but only because someone made the effort, to use bold font style. Also an anounced match cut surprised me, if its would be ai. And I hoped, the two people behind the ai script don't wasted time for this unimportent stuff in the 2,5 hours. But it was a horrible script. Even for a comedy movie it's an unfunny over the top humor since the first second. Then too much dialogue and typical chat gpt fight choreography(i asked him also for this kind of stuff). Script B had some weird movie logic cliches in it(a woman beat a muscular guy), but at least the premise is more interesting, because both main characters has no relationship to each other at start and both get different action scenes, which doesn't end well for both of them but also give them more complex character traits. And you wouldn't think at a chinese woman as bodyguard and a male popstar in his 30's as first, when you read the logline. Also we have some typical but also solid pathing with some interesting scenes. But it missed the given task in its first 10 pages. But in all Script B beats Script A by miles.
After a night I have to admit, I am not sure anymore. I used the logline in chatgpt and the generated script had more parallels to Script B then to Script A. Also one punchline in Script A confuse me, if it was ai written. They made one good joke about energy drinks which sounds like something, only a human would come up with. Also the ai authors could used a great work flow for better complex scripting. B is still better imo. But it would anger me, if the professional screenwriter just came up with a trash movie idea.
The problem is: with Hollywood nowadays, ALL scripts seem to have been writing by AI. Or at least by braindead people. I don't know if I've seen a single movie (not even one) in the last decade that couldn't have been done by AI. So - the challenge is kind of pointless (it isn't helped by the very premise being something so rote that it would never make a good film. While it's unlikely Ai could come up with anything better, neither can real people, apparently).
Hi Tim, I was being facetious and was kind of poking fun at people who say things like... Dragon tastes like chicken... If you are what you eat well that might explain why my wife occasionally calls me a peckerhead. I've been doing a lot of work recently with AI and posting it on youtube, facebook, etc. Is there a way I can get it in front of you. I can't thank you enough for turning me on to Minimax.
Magnific is extremely overpriced and doesn’t offer much of a difference in results. You can achieve similar outcomes using free tools or even Photoshop. With pricing like this, it’s hard to see how they’ll stay in business for long.
Wow, really shocked by this. I use freepik who own magnific. I'm assuming it's similar technology. I think it adds waaaay too much. It's got a habit of turning all humans the same. It gives them similar features. It makes all the glamorous women a lot butcher, widens nose, puts muscle in face and neck, removes make up. Its greatest strength is fusing together messy composites. I can do rough cut outs in photoshop , put it into freepik and boy oh boy it gets rid of that cut n paste look. I think it's amazing and is a huge time saver.
In regards to the human v AI writing contest, I read both and script B runs away with the win! It's not even close! Script A is a contrived mess with no cohesive thought process and even the characters names seem made up! Script B all the way.
I don't know, I've seen a lot of movies that feel like script A that were (in all likelihood) written by humans. Might also have been chimpanzees, though.
See? You learn stuff here." I always feel like you are just chatting with us extemp about our favorite subject, even bringing up in-jokes like the man in the blue suit and his wife in the red dress divorcing him for too many jaywalking tickets.. I really appreciate the way you don't overhype stuff like Runway or Adobe or even the latest new editing feature of Midjourney. Your ai updates are the most fun and helpful I've found and I watch 'em all.
As someone who has written five scripts using Claude 3.5, I'm pretty certain script A is human, and script B is AI. A has a distinct voice and pacing that B doesn't. A uses non-standard slugline annotations (I./E., rather than INT./EXT.) that break the rules in a way the AI doesn't. B abuses "wrylys" and puts longer character descriptions in parens which is typical of AI output. The dialog in A is sharper and edgier than B, which is competent but not especially imaginative. The action lines in A use sentence fragments and caps for punch and to draw the spec readers attention like a pro writer would do. B uses more complete, grammatical sentences which slow the read and lack punch.
These are some of things I have to correct when using Claude by specifying detailed style prompts and exemplar's of my own writing style to add voice to the generated output. LLMs are like a reverse Turing test. If you ask it unintelligent questions, you get unintelligent answers. If you know what a good spec script requires and ask Claude nicely, it can surprise you with its creativity at least for a first draft.
A is terrible, though. If that's written by a human, their school is overpriced.
Agree. I read the first line of Script A and instantly thought that there is no way an AI wrote that line. On the other hand, Script B's first line sounded exactly like AI.
That said if they tried hard enough for enough hours I'm sure one of them could have gotten Claude (not chatGPT) to write about a nip slip.
I didn't have the patience to read the full scripts. I started with A and read the first page and a half, then went to B. Honestly, neither of them was good, but I got through the first two pages of script B a lot easier than with A. I can't say which is more likely to be human vs AI. They seemed pretty comparable to me, which doesn't bode well for the human writer I would say going forward. Lol. Part of the problem for me though is that reading a screenplay vs a regular story feels very unnatural. I would much prefer an AI that could write competent sci-fi or fantasy stories.
I love to discuss the question which could be ai and which could be human. After reading both scripts completely I came to the opposite conclusion that A must be ai and a little bit edited and B must be human.
A is going straight forward to the task, while B started to introduce both characters in their own way in non boring scenes first. A has too much dialogue, while B has more „show, don‘t tell“ aspects and subtext. For example the „chinese security lady“ had the opinion that the popstar‘s security precautions are too lax. It‘s also shown by her reaction, after she noticed the lame ‚6969‘-security pin. In script A both characters starts with a relationship to each other and the girl is good with her bodyguard. This means there are less potential for conflicts between both. I also find B much more exciting. I want to know what the reason is, why someone try to shoot the popstar. In A they just run in a mafia situation by accident and there is nothing different, if two other random people(no popstar) run into this situation. I am not interested to find out, why yakuza killed a random cashier. B doesn‘t include a murder yet. I think, the author prioritized his own ideas first than the task. And in his idea, the two characters need to meet each other first.
@@musicinthemachine B goes into the absurd almost from the start (the bodyguard beating up fans for no reason) and ends up with a random casher in a random store being murdered by Yakuza of all people. If a person wrote it, they should be ashamed to post this publicly. If a person was assisted by AI, they should also be ashamed to post this publicly.
B isn't spectacular, it's kinda safe and boring, but it works as a movie. Both characters are introduced properly, the main plot line is introduced properly, and the tension between the characters is set up properly (the bodyguard was a fan of the star). With A, the age difference is too big for a (non icky) romance, not right for a father-daughter type relationship. A is just nonsense.
At this point, the man in the blue suit is getting so much back story, he's developing his own lore.
At some point he'll have a Wiki Page!
He's Jay, Jay Walken
You should use AI to generate his lore to come full circle
yall are super nerds
I'm definitely ready for the blue suit man verse.
Did you EVER think you would hear: "We just ended up with an upscaled version of Ghosting" and it would actually make sense? Life in an AI world has stuff so weird you couldn't even dream it up LOL Great job as always Tim!
Haha, Upscaled Ghost needs to be a band name!
I still am amazed at how good midjourney is at cinematic. I've been working on an overlook hotel and the ones in flux look like the Ritz or the plaza.
MJ just captures the 1920s aesthetics, the dramatic light, shadows, lock stock the lot.
I did notice it absolutely struggles with mazes, where the flux model can produce detailed hedge mazes easily.
Oh, I know-- actually, the image UNDER the prompt that I typed out is actually MJ. Haha, it really did deliver on a straight up Overlook Hotel. Which...if I remember, I was being super lazy, and that was basically the prompt.
Yes, the director's cut of Doctor Sleep is great! Rebecca Ferguson is so creepy in that movie! Thanks for keeping us updated, Tim. So much going on!
So much going on! It was actually nice to slip away to the land of Danny Torrance for 3 hours to catch a (super tense) break! Man, that movie was good.
I would like to welcome, "woman in red dress, walking down beach" to our growing cast of AI test-characters. Thanks, Tim!
Jay walking is no longer illegal in NYC as of I think just this week... so maybe the man in the blue business suit and the woman in the red dress can get back together? Reunited and it feels so good.
I don't know. If she leaves him for something like that, it's just not meant to be. He can do better. I'm counting on a certain sporty pirate.
Good judging yesterday in Curious ref competition! There was some good action!
oh, man-- I'm so bummed I couldn't make the live premiere-- since oddly, I was working on this video! But yeah-- that was a TON of fun. SOOOO much good stuff there! That was probably the toughest one of those things I've done yet!
@@TheoreticallyMedia Really liked the winner and third place. Sucker for old aesthetics
Thank you for featuring my work! 🤍🤍🤍
1000%!! So Good!!!
Thanks Tim for keeping me abreast of AI developments over the last few weeks which I missed. Always love your succinct reviews.
14:16 its "rorotuck" but thanks for having my video in it :)
It IS! WOW. I think this is a Mandela effect for me. I think I misread your handle ages ago and my brain misreads it all the time now!
@@TheoreticallyMedia i get called rorotruck all the time, haha its fine
You left out Topaz Gigapixel 8 diffuser model. It crushes Magnific and Krea just because you can run it locally, and up to 16X, which none of the other models can do. Give it a try and you’ll be amazed by it. But wait, but there’s more! It’s coming to Topaz video as well!!! Without limits of the infamous 10 seconds.
I've got a thing coming up on Topaz pretty soon! Also...uhhh, I'm hearing their API might be popping up in a LOT of places soon!
@ Cool beans! Honestly, I tried it out this past weekend and is really better than Krea and Magnific. You also missed the new Video to 3D Scene from Wonder Dynamics. This is perhaps bigger than anything you showcased in the last 6 months.
i m using topaz giga.. i strongly recommend
@@MabelYolanda-c9i Video to 3D Scene if i understood it correctly is just a Camera Tracker + Motion Capture in one go. Which is convenient but not revolutionary. We've been doing this for years)
Great video as always! :) Thank you and Keep it up!
Thanks, Tim, for the great content, as always. I learned many new things from your channel. Keep it up.🤝
appreciate you!! Thank you!
Easily, script A is human, and script B is AI because at least the closed models don't allow cursing and generally won't produce any violence if it can get away with it.
Script B seems to pretty low stakes stories in general, which are generally the stories that AI make.
As someone who regularly has chatgpt scripts written, I can tell you that it has recently been allowing much more violent text. It does say somewhere that the generated text or request violates the guidelines, but they still leave it up for some reason.
@@sasbe1852 Well, I was assuming that the entire script itself was written by AI, and had little to no human input versus a script that was completely written by a human (so no restrictions).
Yes if a human prompts it, it will allow violence and such.
Thank you for the shoutout, my friend. Based on my experience, it appears that SellerPic AI may have downgraded their model. The output no longer aligns closely with the prompts, and face distortions seem more frequent compared to before when the results were more realistic.
Awesome video Tim!! thanks
I preferred script A Tim... Love you're work mate. Thanks
That cilantro fact is mind blowing
I welcome you. The video is very good, it helped a lot to navigate. I would like to ask how it can be set in the text description so that the video does not move in several directions? I mean that it should be in a side view and in a fixed position.
Hi Theoretically , any tips for creating a character in midjourney without blury background ?
Add --no depth of field at the end of your prompt and use wide-angle camera lenses in your prompt. It works for me sometimes.
@AI-EleganceStudio thanks i will try that
that's a solid go-to right there. I also (oddly) recommend trying v6 as opposed to v6.1. Six, for some reason or another, tends to behave a lot better on that front.
I have this problem with FluxPro. It would be great if Tim provided a tip for generating images without blurry backgrounds on FluxPro.
@@TheoreticallyMedia thanks 👍
What's the comic book page in the background from??
Books of Magic. Kind of a Sandman offshoot Vertigo did in the 90s. Totally random page. I've always just wanted to own a little piece from the analog era of comic book making. It's kinda cool, you can see tape marks and the blue pencils-- I got it off Ebay years ago, but only just got around to framing it! hah
@@TheoreticallyMedia my fav comic book series!!
@@TheoreticallyMedia Nice! And yeah, actual comic book pages will soon be hard to find. I've got art from the first bunch of comics I wrote on, but that's not possible anymore. That said, I do appreciate the speed of digitally drawn comics, and the ability to change anything on the page without much effort. Makes the art directing/publishing process so much easier.
It's simple: I see a Theoretically Media video, I click
What are you looking at up on the left and up on the right all the time? Other than that, good video!
to be honest, I think I visually process while I'm talking-- but, usually only on camera? I think it's because I feel like a psychopath staring directly into the lens.
I must say your reviews are quite inclusive and I look forward to new episodes. Side note" free King is now not performing in the free version. They claim over 700 hours to create a video so they have it on hold unless of course you are a paying customer.
Hey, have you tried Freepik and their integration of Magnific? I used you prompt for that lady in armor, I added a huge bear to and chose their model Mystic and Bam! Damn... :O
Tx again matey 😁😁
Excellent Content.😊
appreciate that!
Pixverse is great for namecards like your brand name in different environments such as horizon, smokey room, sunrise etc.
Just the fact that the A.I script at this point competes, spells DOOM, now studios have decades of bought scripts, feed them into a specialized A.I dedicated program, GAME OVER, output.
How are the pricing model at maginific these days? I was so impressed by the upscale when i tried it for trial version but stopped using it because of pricing. But if i can now cancel my other subscriptions and generate directly in magnific then this might just be worth it.
I think the monthly price is still the same, but the overall cost of runs is cheaper. So, you get more for your money now.
I’ve had a few talks w/ the big guys at the various AI platforms, and I guess that’s the way it works with most of those cloud GPUs. The more you run it, the cheaper it gets. So, there’s this double edged sword for them: it costs a lot for earlier adopters, but can get cheaper as more users start running it.
I presume that’s why a lot of them end up folding 3 or 4 months in. If they don’t get that audience wave, the cost of GPUs eventually crushes them.
Yes, the founder said it's FLUX under the hood
I do write quite a bit, and I do use ai aswell. but I can never give ai the credit that it can create a good story by itself.
it needs a massive amount of involvement and direction, which I dont mind.
oh 100%. I liken it to having a very eager intern. Who...might also be drunk?
Yup I second that.
I'm great at coming up with ideas , but the absolute noob at sentence structure.
This is a god send, I just throw all my bullet point ideas in and it can smooth out the edges.
It's great for cheesy puns which are hard to come up with. 99% are garbage but every now and then you'll get a gem.
Asking it for variations or give me 10 examples is my number 1 goto.
@@TheoreticallyMedia and has sugar related mind fog where they forget plot points a few hours long.
honestly thats fairly accurate to having an actual assistant, except a little more enthusiastic.
@@armondtanz oh yeah!, developing ideas is definitely it's strong side. though it really got no chance of handling an actual full story script.
personally I dont mind it as a "very eager intern" who also has special-needs. carrying more weight is good for me. it should only iron out the kinks. otherwise I'll lose my muscles.
@steve-g3j6b it is getting better. Maybe one day we can move onto asking for advanced ideas.
Be great if they could introduce movie catalogue , like a folder with all your fave movies. It could then analyse everything from cinematic, costume, dialogue, camera, coloring.
Then you could have sliders saying how much of each you want.
So if u wanted a sci fi meets old skool disney , you could tweak and lean into many of the aspects that makes that genre work.
Great video I haven't tried Mystic Yet.
@11:26 is like creepy Lazytown
ooooh, good call!! Lazytown...man, that is a name I haven't heard in an age!
Is hyper detailed more detailed than ultra detailed or is it okay to stick with mega-super-duper detailed?
as long as you add "8k, ultra realistic, IMAX, Ray Trace, Octane, RTX ON" to your prompt, I think either will work.
haha...that of course, is not true!
Magnific not being honest is normal. Upgraded UI just got grid view which is a simple integration for a dev. They need to take some of that money and build out a better UI. Mystic is yet anothe rmoney grab as the wonkey eyes shows the quality of the model.
Well, they've integrated w/ Freepik at this point, so I'm tending to think that Magnific might be the test bed for what we'll eventually see in Freepik?
I do give them a break on the gird view thing-- mainly because Magnific is actually just 2 people. So, one REALLY overworked Dev!
You did it!!! What was the one thing that could improve a theoretically media video? Matt vidpro😂
Haha, I gotta send that over to him-- I'm not sure he's seen it yet!
yes, what is next , what we go next week? lol, this is growing faster then anybody could thought
As the (now) old saying goes: This is the worst it'll ever be!
@@TheoreticallyMedia yes, I dont know where to spend my money :)
10:56 thats Dutch Van Der Linde 100% 😂
Ha! It is!! Arrrruuuuuutttttther!
Cilantro had me thinking I was watching a cooking channel for a split second 😅
Compared to the spaghetti eating Will Smith, we've come a long way...
You know what's crazy? That Will Smith Video was in March of 2023. That's NUTS! NOT that long ago!
@TheoreticallyMedia yeah, just give it another year!
@AarreLisakki strangely enough, watching the video gives you the same question, lol
@@TheoreticallyMedia recently had a little opinion exchange about the perspectives of AI visual generation and realistically it doesn't seem to be that bright. It feels like we're somewhere near of hitting a plateau. And then i stumbled across a "Has Generative AI Already Peaked?" video which has some serious research paper to support the point.
@AarreLisakki First of all, i was talking about visuals not ChatGPT-like text models. Secondly, we don't see major improvements in - hands tracking, object tracking, prompt understanding, consistency, massive undertrained areas, etc. Lastly, if someone was shouting wolf before it was there it doesn't make current concerns less plausible. Especially when it has a scientific base not just someone's premonition.
Majority doesn't understand neither technical side deep enough nor the current state of the models to take their personal opinions into account anyway which you cannot say about researchers and people who work with multiple models intensively.
Nailed the script detection in under half a page. Basically, the AI copies how most scripts are written. Most published scripts are shooting scripts, ie, scene numbers. Additionally, AI tends to tell, not show. And dialogue tends to be extremely on the nose. AI's don't really understand subtext. The AI script might be better but it would need serious work to fix the issues and add in all the subtext. But it is getting there. Once AI's understand subtext, and show instead of tell, and get character consistency and pot consistency right, then it's going to be used a lot until it gets stale and the audience gets sick of AI stories that will tend to be pretty generic and formulaic. It's not about being paid for writing. It's about seeing the vision come to life. Back to writing my script.
You forget to tell us which script is ai and which not. Because A has more dialogue and less subtext in my opinion I hope you meant A to be ai written?
@@sasbe1852 If I had to guess, I'm thinking script A is the AI and B is the human.
@3:36 That's only realistic if that crazy long sword is made out of graphite and wrapped in aluminum foil. :)
18:10
I havent red the script. but only the first 2 lines on top. I am guessing the one on the right is the AI version.
What AI screen writing program was put through the challenge?
detailed toward the end of the video. CGPT and Claude, basically.
I'm going for Script A as the Human one. Trivial reason. VMA NipSlip comment sounds implausible for AI to come up with. There's a human brainswamp behind that one, in my estimation.
My suggestions:
1.Upscaling should happen automatically.
2. All AI should free for 5 years, because all are in Beta stages.
None of them have licked they eyeline problem yet!
Haha you crack me up. Suddenly we are talking about cilantro and it's actually relevant to the AI video generators 🤣
Script A was by human (I think), while I like though the bot version. Just seems more digestable and meaningful while reading. I can almost see the scene as I read (I guess maybe my understanding of the English slange in Script A was too hard to me to comprehend).
Hmm, script B seems more human to me, while it may be more ‘boring’, it flows way better and is more consistent, and it does seem like it could be an interesting story if continued. Script A on the other hand seems to jump into fights a lot quicker and randomly in a way that reminds me of AI hallucination…. 🤔
Interesting, I thought it was the other way around. Script A just seemed to be written by someone who doesn't know how things work in reality, which seems very familiar from modern screenwriters.
If they had challenged some rando on the street, then possibly. (I’ve definitely encountered some really badly written fan fiction before. lol). But my understanding is that the human challenger is a professional scriptwriter? 🤔
@@ThetaGraphics if A is human, it means the screenwriter guild deserved to vanished. I am still confused about the energy drink joke in it. This is like something, only a human would write.
And after I give chatgpt the given logline, the generated first scene had more parallels to script B than script A.
I was thinking the same thing...script A seems over written and some parts cliche and illogical.
@@MarionMicheleFilms But isn't that how a lot of Hollywood screenplays written by humans seem? And I found script B to be quite cliché as well. I'm very interested in the reveal.
Where do we find the answer to the screenwriting challenge? I read A first snd was convinced it was human. Having read B, I'm not so certain. Both could easily pass as human. I do think A was better, either way.
On twitter they said they will reveal the answer next week. The poll there is still open for the next 2 days.
do you know of any good videos or articles or sites that get into a detailed technical description of exactly HOW generative AI works? Both for stills and video? I mean what do all the developers for each of these companies do exactly and how does the tech work? What software/hardware/tools are they using? And how are the different styles/aesthetics/models developed? What changes do they have to make from where it was say a year ago? I can find virtually NOTHING on the internet that gets into that technical aspect of it. Does ANYBODY know? Because nobody seems to know
These pictures are so good. Insanity.
I started with B first and as I read the opening descriptions, I thought the language was way too flowery, so I thought it was AI... BUT...
In the end, I thought B had much more sophisticated, subtle humorous setups and A seemed like it was written 10 years ago with the much older, dated tropes and references.
So I'm calling A is AI and B is human. I don't think AI is that good at humor yet. Maybe I'm wrong... if so, sorry human.
great analysis-- but, I think the most telling thing is that you have to say "maybe I'm wrong"
we've come a LONG way in a short time! Scriptwriting in ChatGPT3 was so totally obvious! And that was-- what? less than a year ago?
Very interesting
A script was written with so much dark humor, I doubt and would be very surprised if AI wrote it. It was GREAT!! Script B was formulaic, cliche, and UNREADABLE after 3 pages which were more than I wanted to give it. Am I wong? Now that's human humor..
My answer: Script B - AI , Script A - human, answered via GPT's help. 😁
I was kinda thinking the opposite...script A seems over written and some of it illogical. Like why would the body guard plea with the crowd if he's so tough (and then it said he knew it was going to go that way...)?! 😃
@@MarionMicheleFilmsohk! Well if you are right then guess even GPT cant recognise work from its own species
@@nadk666 I might be wrong! :) I'm super curious for the reveal.
Man in blue business suit doesn't have to worry about J-Walking tickets but now has a new worry... covid-19
haha, poor guy can never catch a break! But at least he looks good in that suit!
That video of Matt Vid Pro got me laughing like crazy 😂😂😂
If the scripts have not been edited, 'Script A' must be human, because, well, the word '$#@%nuts' is on page 1. If the LLMs _have_ started using that language, we're in for a good '&#@%ing' time. I don't know if you remember us commenting about 6 months ago, about how 'AI videos are not quite there yet'. Now we're zooming in, stopping frames, divorcing couples on the beach and the streets. Come a long way. And Tim, let me help you/us a little with Mystic 2.5beta. It allows mild NSFW. So, if you want to generate 'fine nude photography' and need to breathe outside the MJ ecosystem, go, people, go! I haven't tried that yet, of course. I would never. Magnific is finally offering more for their price, and that's fine by me. Thanks as always, and so good to watch a longer video. 👍
thank you for researching the cilantro issue, thankfully sans that gene lmao
pace of ai starting to feel exponential for realz
I sense a script coming with a mini series.. "The jaywalking man in a blue business suit and the girl in the red dress on the beach" (catchiest of titles). Why did she leave.. or more so, why did *he* leave?? I'm sure he knew he was jaywalking. .. unless.... hmmm. And why the beach? Who is that red dress for? .. or more so, who does that red dress belong to??! wait - is the man a real man or ... ?? - Lord I want to watch that so badly now.
It's getting so difficult to keep up with all of the changes. No more yearly subscriptions for me. Going month-to-month so I can change/keep up with the greatest and latest image/video generators!
Your freaking jokes lol so catchy and I totally enjoy watching your videos because I'm entertained in more ways than one
I feel like a film student set that whole screenwriting thing up as a trick to get us to read his/her crappy screenplay. Cilantro is delicious and the soap people are twice as nefarious as the AI predictions of doom and gloom getting passed around. The man in the blue suit tries to get his nails done but he always walks into phone shop next door, to the left, due to his eyewonk issues... His wife stands one place over to the left, for this reason (after 3 frustrating years). The people working at the Verizon store next to the nail salon are sick of it, to be frank, and are starting to suspect AI. I think he breathes in the steam of children but that's out there.
You missed a Matrix movie reference with the woman in the red dress by putting her on the beach my bro.
❤❤❤❤
Script A is the human, I feel it’s quite obvious because how much swearing is in script A. If I’m wrong then the promoter intentionally threw those in to throw us off
I agree. The name of the pop star is also much less stereotypical in script A. Script B played everything a bit more "safe", but to be honest, I actually liked script B better. They are both not masterpieces, but A gave me some serious edge lord vibes and I felt like people acted a bit weird at times. But all in all I think they are of similar quality, which is pretty interesting considering the circumstances.
14:27 Those boys have skirts on. I didn't see the prompt, but that looks wrong.
haha, I did notice that after I shot this-- hey, not judging! Maybe they're kilts and the kids are going to Braveheart camp?
Just scanning the first couple lines of each script I'll say the one on the right is AI generated.
I think NoFilmSchool is going to publish the curtain reveal soon-- but, I'd tend to think you're right. That said, the fact that we aren't immediately calling one out as obviously AI? That says something.
@@TheoreticallyMedia I'm guessing someone with no AI experience would find it difficult to identify it. And that's almost all of the general public. That's it's value at this point.
I read both scripts completely and I say Script A(left) is ai generated. It had many random scenes thrown together, while the pathing of Script B is much more typical movie level. Also Script B missed a very important point of the task. In 99% of the cases AI's dont forget the main task.
@@sasbe1852 AIs don't forget the main task?!! How long have you been playing with AI? And you obviously could be 100% right, of course. It's all interesting fun.
@@thanksfernuthin After several prompts it could forget the goal and maybe the ai users used several minor prompts to complete the first 10 pages. But the goal wasn't really complex. I would say, AAA llm's would not forget this that fast. Also I think the human had several ideas in his week so he set the priority of the witnessed murder(there is none in script B) lower.
As it's fantasy you could explain the face change and ghosting as her about to unleash hell on her followers. She's digging into her darker side. Not ideal of course.
I've seen and liked Doctor Sleep. In your opinion, is the 3 hour cut worth it, or is it padding?
Oh, it's 100% worth it! Much more novelistic? I wouldn't recommend going at it in one go-- in fact, the movie is structured with chapters (like a book) almost encouraging you to watch it in chunks.
It almost feels like 3 episodes of a big budget Netflix show.
(Which, to be honest, I'd probably binge watch anyhow...)
@@TheoreticallyMedia Thank you kindly for that. I like to watch new movies in one go but once I've seen them, I don't mind breaking them up. Cheers and have a great day.
I need money from ai
Right now I think you more SPEND money on AI. haha...
If all of us bright brains can come together to create something using ai, we can make money
A is written by the Bot, and B is written by the Human
I only had to look at the 2 script images and speed read them to see the first is definitely human written and hilarious while the second is generic and without life. Hey maybe I’m wrong. I’ll be surprised if I am.
Script B is definitely the one written by a human, right? Way less cliche but also misspells "peashooter" as "pee-shooter"
Haha, love it that we can ID human by the mistakes!
@@TheoreticallyMedia To be fair, the other script reads like an amateur fanfiction so I really hope I'm right in thinking script B's the human-made one, even disregarding the error 😅
damn Tim, you stole my ScarJo Aloy prompt huh, that's it... you me outside in the schoolyard at 3 pm sharp... it's on sucka! lol kidding obviously.
Just going by the first page the on the screen, I'd say the left one is AI. Can't imagine a human writing a girl scout as leader of a bunch of fans, plus the bodyguards opening comments are way too wordy. Could be wrong though!
I thought it was the other way around. But the fact that we are not really sure says a lot.
Hi 👋
Heya Louis!!
Have the gene. Tastes like soap. Hated for like ever then one day it didn’t anymore. Now love it. Weird
oh, that's ODD! I didn't realize you could "grow out" of it! That's really interesting-- Are there other tastes you've found that you developed? Hmmm, might need to try Black Licorice again.
...nope. Still awful.
@
Kinda all occurred when I cut sugar out of my diet and went keto. I can also stand spice heat way more than before. Habaneros are like Bell Peppers to me.
❤❤❤
all Ai is doing is mimicking human styles and patterns of writing so technically its kinda still human....just generated really fast
I think Script A is the AI generated screenplay. As I started to read it, my first thoughts were it was written by a human, but only because someone made the effort, to use bold font style. Also an anounced match cut surprised me, if its would be ai. And I hoped, the two people behind the ai script don't wasted time for this unimportent stuff in the 2,5 hours. But it was a horrible script. Even for a comedy movie it's an unfunny over the top humor since the first second. Then too much dialogue and typical chat gpt fight choreography(i asked him also for this kind of stuff).
Script B had some weird movie logic cliches in it(a woman beat a muscular guy), but at least the premise is more interesting, because both main characters has no relationship to each other at start and both get different action scenes, which doesn't end well for both of them but also give them more complex character traits. And you wouldn't think at a chinese woman as bodyguard and a male popstar in his 30's as first, when you read the logline. Also we have some typical but also solid pathing with some interesting scenes. But it missed the given task in its first 10 pages.
But in all Script B beats Script A by miles.
After a night I have to admit, I am not sure anymore. I used the logline in chatgpt and the generated script had more parallels to Script B then to Script A. Also one punchline in Script A confuse me, if it was ai written. They made one good joke about energy drinks which sounds like something, only a human would come up with. Also the ai authors could used a great work flow for better complex scripting.
B is still better imo. But it would anger me, if the professional screenwriter just came up with a trash movie idea.
The problem is: with Hollywood nowadays, ALL scripts seem to have been writing by AI. Or at least by braindead people. I don't know if I've seen a single movie (not even one) in the last decade that couldn't have been done by AI. So - the challenge is kind of pointless (it isn't helped by the very premise being something so rote that it would never make a good film. While it's unlikely Ai could come up with anything better, neither can real people, apparently).
Umm. Tom Selleck is rugged Tom Selleck. 😄
VALID POINT!!!
Oh cool!!!…I’m just kiddin’
I am so screwed... Cilantro tastes like... CHICKEN TO ME!!!! What the hell gene is that!!!!
I...think that's better than soap? That is super weird though!
Hi Tim,
I was being facetious and was kind of poking fun at people who say things like... Dragon tastes like chicken... If you are what you eat well that might explain why my wife occasionally calls me a peckerhead.
I've been doing a lot of work recently with AI and posting it on youtube, facebook, etc. Is there a way I can get it in front of you. I can't thank you enough for turning me on to Minimax.
The one on the left is human. Waaaaay better written
Script A seems like AI
Fun fact: If you click on a comment twice, you’ll like it, but if you click three times, you’ll reply to it.
The model turning was alright until it wasn't
Ah man you were doing so good my guy!...cilantro?🤢🤮Kids love it, wife loves it. Not me.
Magnific is extremely overpriced and doesn’t offer much of a difference in results. You can achieve similar outcomes using free tools or even Photoshop. With pricing like this, it’s hard to see how they’ll stay in business for long.
Wow, really shocked by this. I use freepik who own magnific. I'm assuming it's similar technology. I think it adds waaaay too much.
It's got a habit of turning all humans the same. It gives them similar features. It makes all the glamorous women a lot butcher, widens nose, puts muscle in face and neck, removes make up.
Its greatest strength is fusing together messy composites. I can do rough cut outs in photoshop , put it into freepik and boy oh boy it gets rid of that cut n paste look.
I think it's amazing and is a huge time saver.
12:04
😆 🤣 😂
In regards to the human v AI writing contest, I read both and script B runs away with the win! It's not even close! Script A is a contrived mess with no cohesive thought process and even the characters names seem made up! Script B all the way.
Amen. But I still wonder if B is also the human one. I am 95% pretty sure it is but by these screenwriters these days...
I don't know, I've seen a lot of movies that feel like script A that were (in all likelihood) written by humans. Might also have been chimpanzees, though.
Can you please make a video about deep fake video in real time.
👀why do you want this?
Rugged Tom Sellek 🤣