with open AI solver getting all the hype in the last couple months I can imagine some people are wondering what existing software do is and how it performs we've all seen the quality that open AI aims to achieve but as of writing this it seems to be getting pruned of anything that might be Doom defensive don't want some another chy BT situations about hands stey but I'd be lying if I didn't admit to making it say some really dumb things when it was first released I wonder if open AI would have d
eleted conversations had I not deleted them myself forever ago but while we waiting for that why don't we take a look at what else we can use instead so p collabs is an online tool that was released in late September 2023 we can sign up using Discord and Google signing up is completely free and the first place you'll be taken to is the explore page where you can see other videos people have made if you want to watch them at their current resolution you can just hover over them or to increase the
m we can click the enlarge icon at the bottom right hand corner PE UPS have recently added the ability to enable sound effects so try not to set your volume two high before you play them if you look along the top you'll see different features that will be used like for this sizzling bacon one we've got video inot and sound effects and for those fireworks we have added four seconds running along the bottom here we've got redo retrim edit at 4 seconds and you can click the three dots to the right
of that and upscale beneath that unless video input was used you will have the PRS for the generation alongside a share button before I describe what these buttons do I'll point you towards where we can actually generate our own videos we'll be able to describe what we want in our video here below that we've got the options attach reference media enable or disable sound effects adjust the aspect ratio or reduce frame rate maximum is 24 FPS some useful effects you might like to use when generatin
g a video can be found here we can do things like making the camera move around or zoom in with a motion setting here the majority of generative AI models have the ability to include the negative prompts and you can find that to the right of motion control similarly we've got text consistency which controls how the video Will Follow The Prompt so going back to these buttons directly underneath each video retry and reprompt just two sides of the same coin Just One lets you alter the pumps first u
sing the existing video as an input so we don't stray too far edit will not only let you alter the pumps but also a portion of the video think of it like in painting and expand canvas is like out painting as it will give you a zoomed out look lip sync is also fairly new if you've got a character you want to animate saying subscribe to koal in a lab you can either type it in here and click generative Voice or you can record yourself seeing it and attach it here but do be mindful before recording
a speech the basic plan only allows you to make videos up to 3 seconds long and that's if the character in the video is already talking before adding lip sync he'll look like he's muttering under his breath like this unfortunately the basic plan doesn't let you scale Fage either if you are considering paying for it then we put a comparison here of all the different features if you're committing to a year then it's going to be cheaper in the long run but that depends on what you really plan to us
e it for if you are interested in some use cases outside of what you can already see on the website here's the page on Reddit showing the recreation of the advertisement of course it's not 100% And I'm not sure why anyone would want to R sh salt on the face but it's still worth a Lu you can also find many examples on YouTube too if we go back to the ones I've made you can see I've already filled about with some RS and promps and modifying things I've seen in the explore tub but I was curious to
pay a little less than £10 after conversion and see what I can generate with a month of the standard plan for my first project I wanted to go all out with the different options I wanted to see if I could generate a short clip of John Wick as an anime running through a building I gave it a bit of motion and had the camera pan to the right for the negative prompt I put photo real stick and stood still I didn't realize it before but if you click retry it will appear under the preview next to your o
riginal video As for the new one yeah it wasn't perfect either anime John Wick would try and move and then he would morph to an entirely different angle so then I tried to repump but if I specify the direction he's running put more thing in a negative prompt again he's not running there's no cam work he's got his gun out which I didn't ask for but hey I still like it nonetheless and at least he's giving it some posture problem with working with AI like this is that you kind of settling for compr
omise I didn't initially ask for this specific scene but I was interested in seeing what would happen if I haded sound linken the clip and upscaled it at this point though I don't know if I was doing it wrong but I wasn't really impressed extending the footage just B him and stop morphing again and continuing the weird little dance he was doing so my next experiment was to try an ogre lurking through his swamp but while let was somewhat accurate to what I typed in my head I was envisioning Shrek
stomping through his swamp like it was a 2d platformer so instead of clicking the we pump because I know it would try and make it similar it's always already there I just changed it to clarify wer said pixel art style Shrek stomping through swamp 2D platformer video game and while this D1 definitely resembled Shrek it made an attempt to get the art style correct but Shrek wasn't moving the background did resemble something I'd expect to see in the video game however so for the next experiment I
thought if the AI seemed to get a somewhat reasonable idea of who certain characters are like John Wick andrek what would happen if I made up some ridiculous crossover like The Simpsons made Star Trek or to be more specific Ned Flanders shaking hands with Mr spark and I don't know what I was expecting AI has never been able to differentiate between two people in a single image so moving on from those I haven't actually tried using a reference yet so I generated an image from Microsoft designer/
D 3 of something like man in business attire giving speech on stage I was quite impressed with this one as it made made the image come to life the only problem is that it won't let me apply motion controls so I could zoom out but it would let me change the aspect ratio so I tried that and then I added 4 seconds and upscaled the footage I'm still not sure why though because I went back a day later and the options W grade out there is the option to upload a video as well but one thing I don't thi
nk P has yet is a slider to indicate how much he want the AI to reference whatever you've attached like for example if I was using mid journey and attached the image of a man giving a speech I could reduce the image weight so that all maturity understand is that there's a humanoid figure standing in the center of the screen stable diffusion has the option as well before I wrap up the video it did occur to me that maybe I should do some experiments to show hands so I'll show you the ideas I came
up with I now previously had prompted net Flanders shaking hands with Mr spark so I thought I'd just change the art style of that this is what it gave me and I'm not entirely sure why it instead decided to give me a close-up of an Atlantis so I increased the text consistency and it gave me this which I guess is a little closer to what I had intended but not really I thought if I increase the duration then maybe i' seen more of the hands so I added showing hands to The Prompt then this happened w
hich at least showed me some hands another test I did to demonstrate this was a miniature person waving at the camera from the center of a coffee cup I thought adding stuff like videography and too realistic would give me something more like what I want but instead it gave me this and I think that's where I'll leave it for now so a bit of a short video but hopefully I've given you a good enough of look into pups on whether you want to play around with it and see what you can come up with these w
ere just some quick experiments but with some more tweaking of prompts and settings more references I can imagine you could come up with some really creative or trippy looking videos I say trippy based on what I've seen but there's nothing to stop you editing the region to fix anything that looks wrong the AI generated sound effects leave a lot to be desired but you can always just edit in your sound sorts from elsewhere but if you did like this video do make sure to like And subscribe it's alwa
ys fun to look into these different Services I hope you all have a very pleasant week bye now
Comments