Haha, thank you! Yeah, it was a lot of digging around through various documents, the discord, and listening to an hour and a half long podcast with the devs. A lot of the other videos I saw on Kaiber only focused on one or two aspects as well. I wanted to make one that showcased the whole enchilada. I really do think it is a great platform and want to see them take this to the next level.
You just answered several questions I have had concerning not only Kaiber but diffusion itself. And I have become reliant on your reviews to learn. Thank you.🙏🙏
Hi Tim, thanks for putting this video together. I am using Kaiber to create music visualisers. Your advice on Stable Diffusion language and weighting prompts is going to help me. Thank you!
I love that effect that you have that starts at 08 seconds , could you please share the settings and the prompt for that effect, it would be greatly appreciated!!!!
So appreciated! This one is a bit older, but I’m really glad the info was still relevant for you! As soon as the 3.0 update fully launches, I’ll do an update for this one!
@@TheoreticallyMedia I just watched this and still found it highly relevant as well. Looking forward to your 3.0 update, i've used it a few times now and would love your insight, cheers, keep up the good work mate :)
i have been using Kaiber for my channel, and to tame the beast its a deep dive experience in experimentation, i do like your video i have had similar takes on how you are trying to tame the A.i generation, did learn a few tricks to get uniformity in renders i have generated, i like to browse and see how everyone is tackling the wild beast, pretty soon they will have a fine tuned control on A.I generation its just a matter of time.
I’d say once you can get that evolution scale down to zero, Kaiber will be a real beast. I’ve been having a lot of fun taking Gen2 or Pika videos and then “finishing” them in Kaiber as video to video. That is super powerful.
Excellent to hear! And glad you found this! I’m actually taking a look at Kaiber’s new mobile app (in beta) right now, so I’ll have some more Kaiber content coming up soon!
Phenomenal video my friend, thank you. The last section with the skateboard video to music is where I find the most fascination... do you have a tutorial for that process specifically?
I just subscribed to Kaiber for the annual plan. It's a cool app but does have some shortcomings. I'm a bit miffed that they charge you for every rendering, even if it's not what you wanted/prompted. I would also like to see an option to "teach" Kaiber when the rendering is close to what you want, but needs to be improved/tweaked.
I'll pass that note along to them!! It does have some shortcomings, agreed. Did you try out the new Motion 3.0 feature yet? I did a video going over it here: th-cam.com/video/F84LLCMRjMg/w-d-xo.html really cool!
@@TheoreticallyMedia yes, I've been playing with it all for a week now. The Motion feature is cool but I'm finding that the rendered result is often very different from the actual preview they give you. The other thing I find interesting is how Kaiber always wants to turn me, and all of my images, into Asian people. No matter what I render - I get Asian-styled houses, buildings, boats, etc. Not sure why.
WOW!!!!!! Mind blown!!! And that video was simply AWESOME!!!!!!!!!!!! Love when you post! Nothing but the best of good, informative content for the masses! We appreciate you!!!🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿
Heck yeah! Kaiber rocks. Thanks for this. Definitely going to have to do some additional experimentation with this newfound knowledge. Appreciate you, brother!
Thank you so much! Pop over some of your outputs when you've got them! And did you see Gen-2 is out in the wild? The last video I did had some super fun Gen-2 to Kaiber experiments-- you might want to check that out!
@@TheoreticallyMedia Yes! What a crazy day that it's out now. I have burned through nearly all my Gen-2 credits already. lol I will upload some of the Kaiber and other tool experiments to TH-cam. I've also been doing some exploration/beta testing stuff for ImagineAPP. Very similar to Kaiber.
Appreciate it!! Learned a lot from your channel-- and I really enjoyed that livestream! Picked up a lot of tips there! Hope I got everything more or less right! (also, I dig the lofi as your background music...people yell at me all the time here about mine being too loud! It's because I LIKE it! =))
This was awesome tips! Thanks a lot! BTW... what is the FX chain of stock Pluggins in Ableton Live that emulates Tom Morello's pedal board? I'm super curious.
Ha! Had to pull up the session to check...but here we go! Apparently I did cheat with the main guitar amp, that was a JCM800 in Guitar Rig (Tom's Amp)-- To be honest, I've never really gotten along with Ableton's Guitar Amp emulators...But the rest of the effects are as follows: Phaser Guitar > Tremolo > Slight Slap Delay > Dotted 8th Note Delay > Whammy Octave Down For the "kill switch" effect (on another channel) Trance Pad Gate. The Whammy part was Tremolo and Whammy While the "Scratching" was: Phaser/Trem/Slight Slap/and then I would automate the Dotted 8th on and off like I was stomping on it. Seemed like a very Tom thing to do!
Check out Kaiber here: kaiber.ai/?via=tim You can check out the insane Midjourney > Gen2 > Murf AI > Reface > Kaiber AI workflow here: th-cam.com/video/eUrtX432KUI/w-d-xo.html If you like the work I'm doing on the channel, please consider joining the Patreon! patreon.com/TheoreticallyMedia
Hello, is there a way to keep the same looking main character throughout the storyboard sequence without already having an image of that person for each storyboard in upload?
Thanks Tim. Some really useful info here especially the links for Replicate and the cheat sheet. Regarding negative prompts, is there any way to get rid of what seems to be "built in fillers" that the AI creates outside of your prompt. In my case, I'm trying to create a video that includes a soundtrack with storyboarding. The video is set in nature in a valley with a river. I want it to be uninhabited but kaiber keeps inserting buildings and houses into my scenes. The negative prompts don't seem to help! Thanks so much!
You´re writing about an issue i´m seraching for. Could you tell how negative prompts have to be written? When i use e.g. "disfigured, deformed, bad anatomy" in the style in kaiber, they will be amplified ...
@@trabsrocks3102 I went into the dischord Kaiber group - they have admitted that negative prompts dont work - they're on it though and I'm sure we will hopefully. be lucky enough to work with them next year. 🤞
@@simplydigital.Thank you very much for info i didn't find in kaiber dischord and your time 👍. This info helps me a lot to spare time 😊. Maybe it soon works in an Update.
Begining with the prompt "The Image" or "The video" seems to circumnavigate having to put in a lot of that descriptive text at the beginning. Great video, doing a music video right now while playing around with all this stuff, and it's been a fun journey so far.
Currently working on a client's music video with Kaiber and finishing in Premiere pro and after effects, and using some stock footage as well. I've found the MJ describe and ChatGPT combined works well for coming up with styles for Kaiber. I'd love to share a clip if you like, It's turning out awesome this is good advice and direction thank you!
stable diffusion is way better for it. kaiber you can not use keyframe data for parameter controls. stable you can control any part with keyframes info from the song youre using. I have over 100 song specific audio reactive visuals i made for my own personal DJ sets
@@iamisobe that’s awesome! Do you have a link to any of them online? Would love to see them! (And possibly have a solo dance party during one of my 3am editing sessions…ha!)
@@TheoreticallyMedia hmm did it auto remove my responses? Well I have some examples on my youtube page if it did. I tried to link 2 but the comments look like they're gone. Theyre a few videos back and labeled with Stable Diffusion
This was very well put together from what you had to deal with Tim on equipment issues and of course your time. Appreciate you making this and thanks for sharing your experience and deep diving into your tutorial 👍
That’s awesome! A lot of people complain about that warping look it has (and they are working on minimizing that), but I always say: if you lean into it as a look? It is killer.
That transition in your awesome skate movie after the rotation of the building and then the zoom in... left me saying "Wow"... that was super smooth :). Want to give Kaiber a try... however looks like one would run out of credits super duper quickly when trying to iterate? 60credits per month -> with 8 credits per 8sec videogen if going through a few tries... credits already gone. Or am I misunderstanding? :) And love your content as always!
So I did the $30 for 2500 credits this month, since I had basically had 2 tutorials to do with it. I still have about 100 credits left. So, that’s between this one and the one I did on Gen2 to Kaiber. I do get that it’s pricey. Between this, MJ, and ChatGPT, you’re looking at 90/mo in AI tools. One plus side to Kaiber is that the credits roll over. So, that’s actually very good. But, I hear you. I think AI tools are going to be a bit like streaming services. Cancel Netflix, pick up Disney. Cancel Disney, pickup Apple. Etc etc. Oh, and thank you!!
Thank you. One quick question: Can I cancel the monthly subscription every month? Is the subscription automatically extended to the next month? Hie to quit? I couldn't find that information on their website.
I'll have to dig around. I do have a monthly sub as well, but I haven't tried cancelling yet. But yeah, you should be able to do it month to month. Like all sub services, it probably is buried somewhere...sigh.
Thanks for detailed video. In some kaiber videos, double heads and deformed heads occur. How to use negative prompts for animations like e.g. rotation? Any hint?
Kaiber videos come with a completely unique surrealistic style. (Two music videos made with Kaiber: th-cam.com/video/KSSI6N-1MTo/w-d-xo.html and th-cam.com/video/aX_pfJVbOEU/w-d-xo.html ). The good thing is that you can create an entire music video at once. The downside is that it's difficult to convey the plot, and what ultimately comes out is always a pleasant surprise. For comparison, here's a video made with Pixvers, composed of 4-second clips: th-cam.com/video/MarV_W1hZSo/w-d-xo.html
So, I’m going to talk a bit about that in Monday’s video. The short answer is, both? They both do different things. Runway handles image to video and video to video really well, with more or less a stable look. Plus, it also has a host of other things like auto green screen, and I think it even has a built in video editor, Kaiber is super wild with its output, and generates some super trippy and artistic stuff, and I always maintain it had the best Video to Video style around. It is really great for music videos, and is the only one that upscales to 4K out of the gate. That said, if you’re just digging in, I’d say kick off with Gen2 (also use Pika, since it is free right now!), and fire up the trial of Kaiber. One nice thing about Kaiber is that you can cancel your sub, and when you come back, your credits are still there. I really appreciate that!
Hey friend, great tutorials here, I have a question for you : I uploaded a picture of myself to Kaiber and I added some text and prompts to add a background for my picture but unfortunately Kaiber changed my whole looks, it made someone else out of me. The question is that, are there any ways for Kaiber not to change me so much that I no longer look the same or do you know of any other AI platforms, where I can achieve my goal ? Thanks.
So when you get to the end of a process of one prompt (choosing aspect ratio, initial screen, etc) but before you hit Generate, there should be a button that says Storyboard. Hit that, and you'll be taken back to the start of the process, but this will be Storyboard 2. You can repeat the process as needed. Quick note: If you're using an Image Prompt (inspiration), you CAN'T change it between Storyboards. I've tried a bunch of times, but no go.
Excellent info, thanks for your vid. I am currently playing with Kaiber and made some short clips, if you have the time have a quick look and let me know what you think. Again, thanks for the vid, some great info.
that's AWESOME! Ha, the intro of this video had a sly dig at Gen-2, with the line: "It's a video generator you can you today"-- Well, I guess that's moot now!
@@dmreturns6485 excellent to hear! I don't know if you caught it, but I did a kind of insane Gen-2 to Kaiber workflow last week: th-cam.com/video/eUrtX432KUI/w-d-xo.html Now that everyone has Gen-2, they can try that out! And thank god, I'll finally stop getting yelled at in the comments that Gen-2 wasn't available!
Thanks a lot for this great deep dive! Just found this channel and will start scrolling through the content - but can you indicate whether there's an AI where I can upload a short clip from me talking and ask it to morph me into an alien? Kind regards
Morph, I’m not sure about, but I think you can kinda do it. I’d use Gen-1 to create a stylized video of you as the alien, then take you original footage and (using a video editor) use a slow dissolve transition from the original to the alien version. It should work, since they’re both the same video. For extra points, you can mask areas so they don’t all transition at once, giving it less of a 50s sci fi film look.
@@TheoreticallyMedia Makes a lot of sense for sure. Should've thought it out myself, my bad haha. Thanks a lot for the great insights you share on your channel! Have a great evening
Great video, very good explained! If I already made a video using kaiber but i want to add more storyboards and make it a little longer is it possible? Thanks
I've been asking them the same-- not yet. So far you can't switch storyboards, each one generate off the same reference image. I know, I really want it to get there too. Right now, I'm just running individual shots and then putting them together in a video editor.
That’s been a bit of a long requested feature. Hopefully coming at some point. I fake it by creating a few shots in storyboard and then I’ll just create a new sequence with a different prompt, then merge them in a video editor. I’ll check in again with them to see if this is (hopefully) coming soon!
Thank you so much! Yeah, Replicate is really great for those SD engines. I also like to compare with /describe and sometimes cherry pick keywords from both. You can end up with some interesting ideas by combining both!
Somewhere in Hollywood: “ok, here’s the pitch…it’s Sister Act, but in Space! And wait for it…they’re vampires!!!” …actually, I totally would watch that movie.
Thank you! I was pretty happy with that one! I had a whole other version where I sampled Zack for a vocal version, but I’m pretty sure I would have ended up in copyright jail. Kind of ironic that the Machine would have raged on me.
I'm having an absolute nightmare with kaiber when it comes to rendering the final output of my videos,I don't know if anyone experiencing the same problem
@@TheoreticallyMedia It's not responding to image with text and also shorts videos already created converting them can be quite a challenge especially if they got text that matches the speech,redndering takes forever not quick at all even if you've upgraded
Interesting how you showed the order of operations then didn’t follow it in your very first prompt by adding artist second 😂 that tickled me but still and awesome and informative video!
Haha, yeah-- these tutorials tend to be very run and gun and sometimes I miss stuff. You'll note in the second Jon Snow output, it is in 9:16, and that's because OBS decided not to record that entire section, so I had to go and re-do it. I was so mad I forgot to do it in 16:9.
It basically has to do with the way the AI creates video-- each frame generates from the previous frame, and as we've learned from ChatGPT, it will often hallucinate, or just start making stuff up. I do find that the Video to Video option tends to be more stable though. I'm sure the shifting aspect will get worked out of AI Video fairly soon, so for now I'm just enjoying the weirdness. I mean, in 2 years, we'll probably have stuff that looks like it's on Netflix-- and I'll be over here trying to prompt for weird shifting backgrounds!
Nice and clear 😎 Although you make it sound easy, it would probably still take me a lot of time 😆 I'm considering using AI to produce a music video for a self-released band I manage. If I get it right, key is to have a clear storyboard for the whole thing, find images or video footage approaching the storyboard, edit the results via prompts to reach the desired video and that would do it?
Oh, I think that's going to be a thing at some point. I've seen a few early models that are in development that remove the instability of Kaiber, which I'm sure they'll adopt. But you're right, for text to video, there has to be a major library of video sources. It's actually why I think Google's Text to Video is going to smash everything when they finally release it. They've got this entire platform to train from. Huh...so if you prompt me, I'll probably be in there. /imagine some dude at a desk yammering into a camera...
These tools seem to be really random. I just want to add some smokey effects and smog and it repaints my characters to be smoke...too early for practicality.
The best way I’ve found to do that is to create some Smokey effects, and then overlay it with a video editor, and then go back to Kaiber or an AI video generator. Or, use a photo with characters with smoky effects and then use gen2 or Pika. I can totally see a text based AI tool getting confused with the idea and turning the characters into wisps of smoke. That’s both frustrating and kind of funny!
@@TheoreticallyMedia Mathilda May is a top beauty. I just watched it last week again. The ending is so classic. The credits roll at the peak climax, literally, when the ancient sword is driven into the protagonist and antagonist at the same time. Timeless way to end a film, they just don’t make ‘em like they used to. Thanks for the video will be making a short zombie movie this weekend likely with runway as I don’t have time to learn kaiber for it and turn it around in time.
I'm pretty sure I taped it off Cinemax. It was a very...inspiring movie to young me. haha...I also vaguely remember accidentally taping an episode of Knight Rider over it. I think naked vampires and talking cars were probably equally weighted to me at that time. Just below Ninjas and Pizza, which of course, were top tier.
working on it-- I think I figured out if I lower it to -30db it moves into a "it's kind of there" are. If you have a second, check out this video and let me know if the vol level is good: th-cam.com/video/DtB_dAsPsTI/w-d-xo.html
It looks so easy when you go over to replicate. I went over and couldn't even find a way to sign up for it. Ya just upload an image into replicate and it will spit you out a prompt. Easy stuff! Nope!
I’ve got something coming up this week that is a big step forward. That said, I think given the output of Kaiber at 4K, it can technically be used in a pro context. Or even generating and then passing off to a VFX artist to add more elements in for a dreamy or trippy narrative. I mean, I get what you’re saying, but I also think that if you look at the limitations of the current tech as a stylistic choice, you can make some pretty interesting stuff. Not mainstream for sure, but interesting! Ultimately, as with all things AI, if you aren’t satisfied with the results, wait a week. Haha!
I think this is one of the best tutorial videos for Kaiber. It helped a great deal. I'm really pleased with the result of the first video using Kaiber AI. Looking forward into diving deeper in for the next videos. th-cam.com/video/f_CRznDGgSM/w-d-xo.htmlsi=VixHW4CsUgCU6_RX
Great tutorial! Thanks! I could not yet work out the storyboard well but thanks to your tips on the weighting the ghosts in my story got much better than I ever dreamed of! Creepier, but cooler: th-cam.com/video/nHDEIDAG7IM/w-d-xo.html thank you also for explaining the evolve factor! That was the game changer for me because it gave focus to the video evolve 🎉🎉🎉
Yeah, it has that effect. I kinda think it's cool, but I don't know if I could watch a feature length film with this style. Currently, I think it'd be super ideal for shorts, or for say-- a weird dream sequence?
kaiber is just a wrapper for stable diffusion utilizing deforum. you're paying for less settings and optimizations. stable diffusion with deforum is completely free.
True. But I think there’s a valid service being provided to those who don’t have the GPU/hardware to run SD, or the technical knowledge to get it running/maintaining it. I feel the same way about Leonardo. It’s just SD/Controlnet etc, but I think it remains popular because it removes the barrier of entry, I’m glad we have options for both. At the end of the day, it’s all about everyone being able to create things
Can't disagree, but not everyone has the horsepower/GPU to run it-- or, the technical knowhow to install. I don't mind platforms that bring creativity to everyone.
Even Kaiber couldn’t explain their own program as well as you did…detailed and highly informative ❤
Haha, thank you! Yeah, it was a lot of digging around through various documents, the discord, and listening to an hour and a half long podcast with the devs. A lot of the other videos I saw on Kaiber only focused on one or two aspects as well. I wanted to make one that showcased the whole enchilada. I really do think it is a great platform and want to see them take this to the next level.
You just answered several questions I have had concerning not only Kaiber but diffusion itself. And I have become reliant on your reviews to learn. Thank you.🙏🙏
Kaiber 2.0 is coming soon! Keep an eye out!
Hi Tim, thanks for putting this video together. I am using Kaiber to create music visualisers. Your advice on Stable Diffusion language and weighting prompts is going to help me. Thank you!
Excellent to hear! I’m shooting a video on Kaiber today, so keep an eye out!
I love that effect that you have that starts at 08 seconds , could you please share the settings and the prompt for that effect, it would be greatly appreciated!!!!
After viewing many videos on how to use Kaiber. It was good to come across one that actually shows how to do things. Many many thanks.
So appreciated! This one is a bit older, but I’m really glad the info was still relevant for you!
As soon as the 3.0 update fully launches, I’ll do an update for this one!
@@TheoreticallyMedia I just watched this and still found it highly relevant as well. Looking forward to your 3.0 update, i've used it a few times now and would love your insight, cheers, keep up the good work mate :)
i have been using Kaiber for my channel, and to tame the beast its a deep dive experience in experimentation, i do like your video i have had similar takes on how you are trying to tame the A.i generation, did learn a few tricks to get uniformity in renders i have generated, i like to browse and see how everyone is tackling the wild beast, pretty soon they will have a fine tuned control on A.I generation its just a matter of time.
I’d say once you can get that evolution scale down to zero, Kaiber will be a real beast. I’ve been having a lot of fun taking Gen2 or Pika videos and then “finishing” them in Kaiber as video to video. That is super powerful.
Wow thanks for the great explanation! I had no idea about being able to adjust the weighting of individual parts of the prompts
Excellent to hear! And glad you found this! I’m actually taking a look at Kaiber’s new mobile app (in beta) right now, so I’ll have some more Kaiber content coming up soon!
Phenomenal video my friend, thank you. The last section with the skateboard video to music is where I find the most fascination... do you have a tutorial for that process specifically?
Not really beyond what was in the video, but I am shooting a video on Kaiber today, so I’ll try to sneak that in for you!
I just subscribed to Kaiber for the annual plan. It's a cool app but does have some shortcomings. I'm a bit miffed that they charge you for every rendering, even if it's not what you wanted/prompted. I would also like to see an option to "teach" Kaiber when the rendering is close to what you want, but needs to be improved/tweaked.
I'll pass that note along to them!! It does have some shortcomings, agreed. Did you try out the new Motion 3.0 feature yet? I did a video going over it here: th-cam.com/video/F84LLCMRjMg/w-d-xo.html really cool!
@@TheoreticallyMedia yes, I've been playing with it all for a week now. The Motion feature is cool but I'm finding that the rendered result is often very different from the actual preview they give you. The other thing I find interesting is how Kaiber always wants to turn me, and all of my images, into Asian people. No matter what I render - I get Asian-styled houses, buildings, boats, etc. Not sure why.
WOW!!!!!! Mind blown!!! And that video was simply AWESOME!!!!!!!!!!!! Love when you post! Nothing but the best of good, informative content for the masses! We appreciate you!!!🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿🙌🏿
Thank you soooo much!! Honestly, comments like these keep me going at that 8th hour of editing, when the coffee isn’t working anymore!
I have been using Kaiber AI, but I still learned some new things from your video. Thank you for sharing.
excellent to hear! Hey, if you don't mind, I just swapped out the thumbnail-- does that look too washed out to you?
Heck yeah! Kaiber rocks. Thanks for this.
Definitely going to have to do some additional experimentation with this newfound knowledge. Appreciate you, brother!
Thank you so much! Pop over some of your outputs when you've got them! And did you see Gen-2 is out in the wild? The last video I did had some super fun Gen-2 to Kaiber experiments-- you might want to check that out!
@@TheoreticallyMedia Yes! What a crazy day that it's out now. I have burned through nearly all my Gen-2 credits already. lol
I will upload some of the Kaiber and other tool experiments to TH-cam. I've also been doing some exploration/beta testing stuff for ImagineAPP. Very similar to Kaiber.
Great stuff man! Awesome to see more creators helping people find and use Kaiber!
Appreciate it!! Learned a lot from your channel-- and I really enjoyed that livestream! Picked up a lot of tips there! Hope I got everything more or less right! (also, I dig the lofi as your background music...people yell at me all the time here about mine being too loud! It's because I LIKE it! =))
Life Force was one of my favourite movies as a teenage boy, if you know, you know.
8:21 important to note, everything after bridge should be in the style prompt. That’s the actual style.
Ahhhh-- good tip!! I'll pop that into a pinned comment later!
This was awesome tips! Thanks a lot! BTW... what is the FX chain of stock Pluggins in Ableton Live that emulates Tom Morello's pedal board? I'm super curious.
Ha! Had to pull up the session to check...but here we go!
Apparently I did cheat with the main guitar amp, that was a JCM800 in Guitar Rig (Tom's Amp)-- To be honest, I've never really gotten along with Ableton's Guitar Amp emulators...But the rest of the effects are as follows:
Phaser Guitar > Tremolo > Slight Slap Delay > Dotted 8th Note Delay > Whammy Octave Down
For the "kill switch" effect (on another channel) Trance Pad Gate.
The Whammy part was Tremolo and Whammy
While the "Scratching" was: Phaser/Trem/Slight Slap/and then I would automate the Dotted 8th on and off like I was stomping on it. Seemed like a very Tom thing to do!
Super! Thanks again! Can't wait to give it a try. @@TheoreticallyMedia
the skateboarding video is amazing! I watched it 4 times
Check out Kaiber here: kaiber.ai/?via=tim
You can check out the insane Midjourney > Gen2 > Murf AI > Reface > Kaiber AI workflow here: th-cam.com/video/eUrtX432KUI/w-d-xo.html
If you like the work I'm doing on the channel, please consider joining the Patreon! patreon.com/TheoreticallyMedia
Hello, is there a way to keep the same looking main character throughout the storyboard sequence without already having an image of that person for each storyboard in upload?
Great video and explanation! Thanks for the "KaiberSpace" tour!
Haha, Kaiberspace still cracks me up! Thank you!!
Thanks Tim. Some really useful info here especially the links for Replicate and the cheat sheet. Regarding negative prompts, is there any way to get rid of what seems to be "built in fillers" that the AI creates outside of your prompt. In my case, I'm trying to create a video that includes a soundtrack with storyboarding. The video is set in nature in a valley with a river. I want it to be uninhabited but kaiber keeps inserting buildings and houses into my scenes. The negative prompts don't seem to help! Thanks so much!
You´re writing about an issue i´m seraching for. Could you tell how negative prompts have to be written? When i use e.g. "disfigured, deformed, bad anatomy" in the style in kaiber, they will be amplified ...
@@trabsrocks3102 I went into the dischord Kaiber group - they have admitted that negative prompts dont work - they're on it though and I'm sure we will hopefully. be lucky enough to work with them next year. 🤞
@@simplydigital.Thank you very much for info i didn't find in kaiber dischord and your time 👍. This info helps me a lot to spare time 😊. Maybe it soon works in an Update.
@@trabsrocks3102 You're welcome. Good luck with your. projects!
@@simplydigital. Thanks 😊. All the best for your channel and your projects, too.
How many minutes kaiber can converter?
up to 8 minutes!
@@TheoreticallyMedia thanks!
Thanks for your superb tutorials. Great stuff. Your music and video at the end were excellent!
Begining with the prompt "The Image" or "The video" seems to circumnavigate having to put in a lot of that descriptive text at the beginning. Great video, doing a music video right now while playing around with all this stuff, and it's been a fun journey so far.
Currently working on a client's music video with Kaiber and finishing in Premiere pro and after effects, and using some stock footage as well. I've found the MJ describe and ChatGPT combined works well for coming up with styles for Kaiber. I'd love to share a clip if you like, It's turning out awesome this is good advice and direction thank you!
oh, 100%!!! Send it on over: theoreticallymedia@gmail.com
stable diffusion is way better for it. kaiber you can not use keyframe data for parameter controls. stable you can control any part with keyframes info from the song youre using. I have over 100 song specific audio reactive visuals i made for my own personal DJ sets
@@iamisobe that’s awesome! Do you have a link to any of them online? Would love to see them! (And possibly have a solo dance party during one of my 3am editing sessions…ha!)
@@TheoreticallyMedia hmm did it auto remove my responses? Well I have some examples on my youtube page if it did. I tried to link 2 but the comments look like they're gone. Theyre a few videos back and labeled with Stable Diffusion
@@iamisobe It might have? I'll check in the spam folder-- sometimes YT gets a little aggressive with that. Looking forward to checking them out!!
This was very well put together from what you had to deal with Tim on equipment issues and of course your time. Appreciate you making this and thanks for sharing your experience and deep diving into your tutorial 👍
Much appreciated!! I really do mean it when I say I can't wait to see how Kaiber develops! There's so much style here!
Amazing platform and luv that last video! 🎉
Thank you so much! And happy you enjoyed the little mini-Rage session! That was so much fun to make!
I learned about Kaiber like two days ago. 5 minutes and $5 later I had 6 awesome b-roll shots for a friend's rap video. I'll be using it a lot.....
That’s awesome! A lot of people complain about that warping look it has (and they are working on minimizing that), but I always say: if you lean into it as a look? It is killer.
@@TheoreticallyMedia super impressed with the tool.....I'll post a link when we finish our project up.
super cool stuff. I feel like I've only scratched the surface with this one.
there's a LOT to explore in here. Check this video out, where I bash Kaiber with Gen2! th-cam.com/video/eUrtX432KUI/w-d-xo.html
Hey Tim. still enjoy this video. appreciate the guide.
Excellent to hear! Thank you so much!
That transition in your awesome skate movie after the rotation of the building and then the zoom in... left me saying "Wow"... that was super smooth :). Want to give Kaiber a try... however looks like one would run out of credits super duper quickly when trying to iterate? 60credits per month -> with 8 credits per 8sec videogen if going through a few tries... credits already gone. Or am I misunderstanding? :) And love your content as always!
So I did the $30 for 2500 credits this month, since I had basically had 2 tutorials to do with it. I still have about 100 credits left. So, that’s between this one and the one I did on Gen2 to Kaiber.
I do get that it’s pricey. Between this, MJ, and ChatGPT, you’re looking at 90/mo in AI tools.
One plus side to Kaiber is that the credits roll over. So, that’s actually very good. But, I hear you. I think AI tools are going to be a bit like streaming services. Cancel Netflix, pick up Disney. Cancel Disney, pickup Apple. Etc etc.
Oh, and thank you!!
@@TheoreticallyMedia thx mate. Makes sense. Will give kaiber another shot :)
Hello Tim, what prompt can i use to keep my character in view at all times? Please help :)
Thank you. One quick question: Can I cancel the monthly subscription every month? Is the subscription automatically extended to the next month? Hie to quit? I couldn't find that information on their website.
I'll have to dig around. I do have a monthly sub as well, but I haven't tried cancelling yet. But yeah, you should be able to do it month to month.
Like all sub services, it probably is buried somewhere...sigh.
@@TheoreticallyMedia This is what I feared. I will not try it out if it is so difficult to cancel the subscription. Thank you.
@@Gitaro64 I hear you. Let me scope it out for you real quick. I haven’t actually tried cancelling, so I’d feel bad if it was actually super easy.
Thank you for this, I have only recently gotten into creating ai art with kaiber and you have gave me a lot of really good tips.👍🏻🍻
Fantastic to hear! Hoping to have some more Kaiber videos next week!
Thanks for detailed video. In some kaiber videos, double heads and deformed heads occur. How to use negative prompts for animations like e.g. rotation? Any hint?
The best tutorial I’ve seen so far. Thank you 🙏
excellent to hear! Thank you!
Thank you Tim, that was great and informative. Now I'm off to create some lively things
I have been using video bolt . I love . Can I use this with ease. What’s the difference between the two
I haven't tried out Video Bolt, I'll look into it!
Great stuff! Did you add that music to that skateboard video or was it created randomly by Kaiber??
Oh thank you! Nope, that music was 100% me!
Kaiber videos come with a completely unique surrealistic style. (Two music videos made with Kaiber: th-cam.com/video/KSSI6N-1MTo/w-d-xo.html and th-cam.com/video/aX_pfJVbOEU/w-d-xo.html ). The good thing is that you can create an entire music video at once. The downside is that it's difficult to convey the plot, and what ultimately comes out is always a pleasant surprise. For comparison, here's a video made with Pixvers, composed of 4-second clips: th-cam.com/video/MarV_W1hZSo/w-d-xo.html
hi any advice on kaiber vs runwayml?
pros /cons...etc...trying to choose the one to start with. tx
So, I’m going to talk a bit about that in Monday’s video. The short answer is, both? They both do different things. Runway handles image to video and video to video really well, with more or less a stable look. Plus, it also has a host of other things like auto green screen, and I think it even has a built in video editor,
Kaiber is super wild with its output, and generates some super trippy and artistic stuff, and I always maintain it had the best Video to Video style around. It is really great for music videos, and is the only one that upscales to 4K out of the gate.
That said, if you’re just digging in, I’d say kick off with Gen2 (also use Pika, since it is free right now!), and fire up the trial of Kaiber.
One nice thing about Kaiber is that you can cancel your sub, and when you come back, your credits are still there. I really appreciate that!
Can you designate a specific model/character to be used across multiple images and/or video?
Hey friend, great tutorials here,
I have a question for you : I uploaded a picture of myself to Kaiber and I added some text and prompts to add a background for my picture but unfortunately Kaiber changed my whole looks, it made someone else out of me. The question is that, are there any ways for Kaiber not to change me so much that I no longer look the same or do you know of any other AI platforms, where I can achieve my goal ?
Thanks.
how did you give the last image 3 separate prompts? did you show that? Great tutorial btw
So when you get to the end of a process of one prompt (choosing aspect ratio, initial screen, etc) but before you hit Generate, there should be a button that says Storyboard. Hit that, and you'll be taken back to the start of the process, but this will be Storyboard 2. You can repeat the process as needed.
Quick note: If you're using an Image Prompt (inspiration), you CAN'T change it between Storyboards. I've tried a bunch of times, but no go.
Is it possible to transform individual objects in video to video mode by masking them through image prompts?
Does the water mark go away if you pay for the subscription?
It does!
i love your videos please keep making them
thank you so much! And will do!!
Excellent info, thanks for your vid. I am currently playing with Kaiber and made some short clips, if you have the time have a quick look and let me know what you think.
Again, thanks for the vid, some great info.
100% Shoot it over to Theoreticallymedia@gmail.com
Runway Gen-2 just dropped!
that's AWESOME! Ha, the intro of this video had a sly dig at Gen-2, with the line: "It's a video generator you can you today"--
Well, I guess that's moot now!
@@TheoreticallyMedia Thanks for the Kaiber vid btw, very helpful to see it in action.
@@dmreturns6485 excellent to hear! I don't know if you caught it, but I did a kind of insane Gen-2 to Kaiber workflow last week: th-cam.com/video/eUrtX432KUI/w-d-xo.html
Now that everyone has Gen-2, they can try that out!
And thank god, I'll finally stop getting yelled at in the comments that Gen-2 wasn't available!
@@TheoreticallyMedia I need to go back and watch your Gen-2 to Kaiber workflow. The video you generated at the end of the video was really good.
Made my first video-to-video short from watching this! Turned some clips into anime
I honestly think that’s Kaiber’s real superpower. I still want to get some time to swing back and experiment with that more!
How to add multiple prompts? I am on Pro plan
tim, you're a legend! thank you.
Amazing’ thank you!!!
Thanks a lot for this great deep dive! Just found this channel and will start scrolling through the content - but can you indicate whether there's an AI where I can upload a short clip from me talking and ask it to morph me into an alien? Kind regards
Morph, I’m not sure about, but I think you can kinda do it. I’d use Gen-1 to create a stylized video of you as the alien, then take you original footage and (using a video editor) use a slow dissolve transition from the original to the alien version. It should work, since they’re both the same video.
For extra points, you can mask areas so they don’t all transition at once, giving it less of a 50s sci fi film look.
@@TheoreticallyMedia Makes a lot of sense for sure. Should've thought it out myself, my bad haha. Thanks a lot for the great insights you share on your channel! Have a great evening
Great video, very good explained! If I already made a video using kaiber but i want to add more storyboards and make it a little longer is it possible? Thanks
I've been asking them the same-- not yet. So far you can't switch storyboards, each one generate off the same reference image. I know, I really want it to get there too. Right now, I'm just running individual shots and then putting them together in a video editor.
Your video is in point! ❤
Thank you so much!!!
Really good information. I'll be subscribing to Kaiber. Thanks.
excellent! It's a lot of fun. Interesting note: If you don't use your monthly credits, they actually roll over. That's super nice!
What if i want to change my prompt mid video?
That’s been a bit of a long requested feature. Hopefully coming at some point. I fake it by creating a few shots in storyboard and then I’ll just create a new sequence with a different prompt, then merge them in a video editor.
I’ll check in again with them to see if this is (hopefully) coming soon!
Very instructive. So helpful. Many thanks.
So happy to hear!!
Another informative video, will definitely be using replicate.
Thank you so much! Yeah, Replicate is really great for those SD engines. I also like to compare with /describe and sometimes cherry pick keywords from both. You can end up with some interesting ideas by combining both!
😄(10:13) "This was the kind of thing" wanted, "not nuns boarding a spaceship"
Great video, good info, thanks!👍
Somewhere in Hollywood: “ok, here’s the pitch…it’s Sister Act, but in Space! And wait for it…they’re vampires!!!”
…actually, I totally would watch that movie.
incredibly well made and detailed video, thanks!
Excellent to hear! I think the new Motion Feature dropped today, so you might want to take that out for a spin!
@@TheoreticallyMedia paid app hai free mai use kese kare 😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😢😮😢😢😢😢😢😮
i liked the skateboard with rage against the machine :)
Thank you! I was pretty happy with that one! I had a whole other version where I sampled Zack for a vocal version, but I’m pretty sure I would have ended up in copyright jail. Kind of ironic that the Machine would have raged on me.
@@TheoreticallyMedia yup . Damn true! 😄
for some reason i dont have the settings tab?
i only have Evolve tab
inside replicate where would one go to generate a prompt as mentioned in the video? TIA :)
I found it! Image to text i believe!!
Excellent! Sorry, fell behind on comments over the last few days!!
Is there anything that can do a stylization over a video with some temporal coherence? This wasn't good for that
They’ve got an update that looks pretty promising. Hoping to dive into it soon
I'm new to this but trying to figure it out. I'm trying to transform myself into a vampire it's been fun to use . Do you have any tips for that?
I'm having an absolute nightmare with kaiber when it comes to rendering the final output of my videos,I don't know if anyone experiencing the same problem
How’s so? Like, what is it doing?
@@TheoreticallyMedia It's not responding to image with text and also shorts videos already created converting them can be quite a challenge especially if they got text that matches the speech,redndering takes forever not quick at all even if you've upgraded
Superb!Thanks
Excellent! You’re welcome!!
Dope video. Thanks!
100%! Might have some new Kaiber videos very soon!
Awesome tutorial man, thank you!
SICK!!
Hey dude ❤❤great content!!!
I wanna know how to generate texts from images using replicate
Did I leave a link in the description? Let me know if not, I’ll drop one in there!
Insane‼️ Thanks.
It's so fun! High recommend to give the free trial a shot, if you haven't already!
man! how the hell did you do that skate boarding video, is excellent, you should make a video on how to do something like that.🤩
Interesting how you showed the order of operations then didn’t follow it in your very first prompt by adding artist second 😂 that tickled me but still and awesome and informative video!
Haha, yeah-- these tutorials tend to be very run and gun and sometimes I miss stuff. You'll note in the second Jon Snow output, it is in 9:16, and that's because OBS decided not to record that entire section, so I had to go and re-do it. I was so mad I forgot to do it in 16:9.
Why does background change so radically...
It basically has to do with the way the AI creates video-- each frame generates from the previous frame, and as we've learned from ChatGPT, it will often hallucinate, or just start making stuff up. I do find that the Video to Video option tends to be more stable though.
I'm sure the shifting aspect will get worked out of AI Video fairly soon, so for now I'm just enjoying the weirdness. I mean, in 2 years, we'll probably have stuff that looks like it's on Netflix-- and I'll be over here trying to prompt for weird shifting backgrounds!
Just put the evolve string on 1 to keep your same image and transform image once
Nice and clear 😎 Although you make it sound easy, it would probably still take me a lot of time 😆 I'm considering using AI to produce a music video for a self-released band I manage. If I get it right, key is to have a clear storyboard for the whole thing, find images or video footage approaching the storyboard, edit the results via prompts to reach the desired video and that would do it?
Hmmmm this makes me think a possible new job is agencies that make videos to feed into Ai. As these are (so far) the best results.
Oh, I think that's going to be a thing at some point. I've seen a few early models that are in development that remove the instability of Kaiber, which I'm sure they'll adopt. But you're right, for text to video, there has to be a major library of video sources.
It's actually why I think Google's Text to Video is going to smash everything when they finally release it. They've got this entire platform to train from.
Huh...so if you prompt me, I'll probably be in there.
/imagine some dude at a desk yammering into a camera...
@@TheoreticallyMedia interesting! I didn't know Google was planning a text to video.
Will be interesting to see how it differs from other models
I'm still learning from Kaiber. All my shorts are made by kaiber.
It’s such a good tool! I’m hoping to have another Kaiber video up next week! They’ve got some pretty cool stuff brewing!
Awesome video
These tools seem to be really random. I just want to add some smokey effects and smog and it repaints my characters to be smoke...too early for practicality.
The best way I’ve found to do that is to create some Smokey effects, and then overlay it with a video editor, and then go back to Kaiber or an AI video generator.
Or, use a photo with characters with smoky effects and then use gen2 or Pika.
I can totally see a text based AI tool getting confused with the idea and turning the characters into wisps of smoke. That’s both frustrating and kind of funny!
The generated video does not really look like a video at all , but rather as an image which has changing random colors on it.
Yeah, that’s the current look. I think a newer model is inbound that takes care of that
Lifeforce is one of my favourite movies.
Same. I wouldn't call it a good movie, but it is a favorite movie. I've got a TON of those!
@@TheoreticallyMedia Mathilda May is a top beauty. I just watched it last week again. The ending is so classic. The credits roll at the peak climax, literally, when the ancient sword is driven into the protagonist and antagonist at the same time. Timeless way to end a film, they just don’t make ‘em like they used to.
Thanks for the video will be making a short zombie movie this weekend likely with runway as I don’t have time to learn kaiber for it and turn it around in time.
Young me really liked "Lifeforce"..... no reason..... no reason parts of the VHS became worn out either...
I'm pretty sure I taped it off Cinemax. It was a very...inspiring movie to young me. haha...I also vaguely remember accidentally taping an episode of Knight Rider over it. I think naked vampires and talking cars were probably equally weighted to me at that time. Just below Ninjas and Pizza, which of course, were top tier.
@@TheoreticallyMedia 0:24
@@FerdinandFerrer-k3m when AR is everywhere we will ALL be in Kaiberspace!
I like your background music, but it's a little loud.
working on it-- I think I figured out if I lower it to -30db it moves into a "it's kind of there" are. If you have a second, check out this video and let me know if the vol level is good: th-cam.com/video/DtB_dAsPsTI/w-d-xo.html
@@TheoreticallyMedia Much better.
The thing i didnt like was hugest Kaiber watermarks on everything.
That goes away in the paid version. But yeah, I get it.
Space Nuns. Could easily be a movie.
Ha! Let’s call Mel Brooks! “It’s The Sound of Music, but…wait for it…in Space!”
@@TheoreticallyMedia "Cosmic Confessions Are A Nasty Habit"
Helpful
Glad to hear that!! I've been circling back to Kaiber more and more-- I just did a few other videos recently featuring the new 3.0 update!
A space vampire Star Wars like movie or show sounds awesome.
I’m actually surprised we haven’t seen that particular genre mashup in something mainstream in all these years!
"Lifeforce" rules! :)
It looks so easy when you go over to replicate. I went over and couldn't even find a way to sign up for it. Ya just upload an image into replicate and it will spit you out a prompt. Easy stuff! Nope!
Well explained. good man I have a music channel and I make my music videos with Kaiber. Greetings from Germany Tony 👍
Thank you Tony!! Yeah, Kaiber is such a killer app for Music videos! Do you have any of your videos on YT?
I know it will get better but honestly, these examples are no where near ready for professional use.
I’ve got something coming up this week that is a big step forward. That said, I think given the output of Kaiber at 4K, it can technically be used in a pro context. Or even generating and then passing off to a VFX artist to add more elements in for a dreamy or trippy narrative.
I mean, I get what you’re saying, but I also think that if you look at the limitations of the current tech as a stylistic choice, you can make some pretty interesting stuff. Not mainstream for sure, but interesting!
Ultimately, as with all things AI, if you aren’t satisfied with the results, wait a week. Haha!
@@TheoreticallyMedia Hi is it out? I would really like to see that.
Kaiber has helped me create TERRIFYING TALES! Enjoy and have a Terrifyingly goodnight!
th-cam.com/channels/L-V2Jgi0m3NWaIdLw64YHg.html
I think this is one of the best tutorial videos for Kaiber. It helped a great deal.
I'm really pleased with the result of the first video using Kaiber AI. Looking forward into diving deeper in for the next videos.
th-cam.com/video/f_CRznDGgSM/w-d-xo.htmlsi=VixHW4CsUgCU6_RX
Lifeforce movie female character is beautiful!
She will eat you if you get too close!!
Great tutorial! Thanks! I could not yet work out the storyboard well but thanks to your tips on the weighting the ghosts in my story got much better than I ever dreamed of! Creepier, but cooler: th-cam.com/video/nHDEIDAG7IM/w-d-xo.html thank you also for explaining the evolve factor! That was the game changer for me because it gave focus to the video evolve 🎉🎉🎉
I feel dizzy
Yeah, it has that effect. I kinda think it's cool, but I don't know if I could watch a feature length film with this style. Currently, I think it'd be super ideal for shorts, or for say-- a weird dream sequence?
kaiber is just a wrapper for stable diffusion utilizing deforum. you're paying for less settings and optimizations. stable diffusion with deforum is completely free.
True. But I think there’s a valid service being provided to those who don’t have the GPU/hardware to run SD, or the technical knowledge to get it running/maintaining it.
I feel the same way about Leonardo. It’s just SD/Controlnet etc, but I think it remains popular because it removes the barrier of entry,
I’m glad we have options for both. At the end of the day, it’s all about everyone being able to create things
All these apps. Sadly automatic 11 and civit ai are still the best.
Can't disagree, but not everyone has the horsepower/GPU to run it-- or, the technical knowhow to install. I don't mind platforms that bring creativity to everyone.