Oh!!! You have to buy/donate money to get access? Is this correct? You didn't mention that part in your video. Please do so in future. It won't feel so tricky. Thanks, Man. Love your videos by the way. Cheers!
@@Qmannn2UTube No, no...For the PDF? Or for Pika? Either way: Not at all. Both are 100% free. On the Gumroad side, just put in zero dollars for the cost.
Thanks, Tim! You are awesome! Nobody else even comes close to your quality of videos on AI generated videos; so glad I found you (also, thanks, algorithm!).
Excellent to-the-point video. I didn't know about the camera parameter getting included. I just read Kaiber is doing the same (or has done). Glad you brought that up. Hope YT fixes itself. 👍
I’ll double check on Kaiber (I’ve been wanting to circle back to them anyhow!), but I think that’s Gen2 you’re thinking of. But yeah, those camera controls are brand spanking new!
Incredible investigation into Pika Tim! I am just starting to investigate the tools and the PDF is extremely helpful for getting into some common commands. I also admire your channel and am excited to see where it goes!
So glad to hear! And I’m really happy the PDF is helpful! Those are a whole extra step in making these videos (and are hopefully typo free?) so I’m really happy when I hear they’re useful to you!
excellent to hear!! You caught this one at the right time too, since the new -camera feature was just released, and I think I'm the only tutorial that has it. I mean, they'll probably release some new amazing feature tomorrow, and I'll be out of date... ...but for TODAY!!
Personally, I think the neg prompts are the real gamechanging secret sauce. It’s like early SD, your neg has almost as much impact as your positives! Let me know how it works out for you!
That was totally what I was slyly implying in this video when I was saying: you can use the bot, but try out the gen rooms first, because there are a lot of great people there!
Very nice breakdown as always. Your understanding and knowledge of filmmaking in general definitely gives you an edge on deeper dives into the cinematic side of things - which I personally enjoy. And I feel those without that knowledge can and will learn a ton from your content. It's like a hidden secret society when you meet other filmmakers or those that worked in the industry, haha.
We need a secret handshake! And a clubhouse....hmmm, maybe that Crime Coffee shop? Meet ya there! (BTW: Are you in Chicago? Why do I think that? Anyhow-- just marathoned The Bear over the last week or so...my god, what an amazing show. High Recommend if you haven't watched it already!)
Crime Coffee would be an amazing hangout. Can have a double meaning, "Coffee is so good it's a crime". The Bear is amazing show, I haven't watched it all just yet. I just rewatched Murder By Death which I can never get tired of. I'm not in Chicago, currently Nashville but moving back to Los Angeles soon - Me and the family wanted to check out Nash for a bit and I needed a small vacation.@@TheoreticallyMedia
Brilliant tutorial. Really appreciate you putting in so much information rather than stringing the video out. Love the up-and-left camera moves. That's a game-changer. :-)
Awesome! Thank you so much!! And I'm glad the PDF is coming in handy! I honestly think it just beats screenshots or having to return to a video 30 times!
Good morning ! Thank you very much for your tip about hidden members. I've been looking for a private connection on Pika Labs for a long time without success. Thanks for thinking of reporting this.
Thank you so much! I’m really happy these PDFs are helpful to everyone! They do take a little extra time (and a fun game to play is “look for the typo” because there are 5 or more put in….unmmm, intentionally, yeah- that’s it; intentionally!) but it’s always worth it when I get a comment like yours!
That is fantastic to hear! Pika is quickly becoming my favorite Video Generator. The results are outstanding! Have a blast, and if you run into any questions, please feel free to shoot them over!
Thanks for all your advice as well Tyber! I didn’t get it all in (didn’t want to overwhelm folks getting started) but there is enough leftover for me to do an “advanced pika” video, so I’ll add your advice there!
Thank you so much! Haha, still got up to 14 minutes! But-- I tried not to waffle too much! (I mean, the Coffee Shop thing, maybe-- but that was too funny not to share!)
I have to say I just don’t understand at all the use of 4 sec hiccup repeat vids but at least in Splice I can reverse loop but of course hair this image/text to vid Amazing and every week better and better as Ai DO! 🎉
I think that 4 second limit will be broken fairly soon. Gen-2 already has an "extend" feature. My question is: What should the max length of a shot be? I personally don't think we need 1-min long shots. 10 sec? Is that enough? I'm curious to see where it ultimately lands.
🎯 Key Takeaways for quick navigation: 00:00 🎬 Introduction to Pika Labs, an AI video generator you can currently use for free that was recently updated with new features 00:57 🖼️ You can prompt Pika with text or images. Text prompts follow a structure like: medium, style, scene description, action description. 03:45 🎥 List of common "called shots" that work well for text prompting in Pika 06:33 🖼️ Image prompting allows Pika to use the image for contextual clues. Can also animate images with /animate 08:24 ⚙️ Pika commands like FPS, motion, guidance scale, negative prompt, aspect ratio, seed, and camera controls 13:29 💡 Tips and advice from Pika community members on best practices for prompting Made with HARPA AI
Quelques mots en français histoire de donner une note exotique à ce fil. Merci pour les videos enthousiastes et souvent passionnantes, et toute cette généreuse contribution à notre créativité collective. Amitiés
What's very frustrating about all of the AI video generators is that when you get something you like, there is no clear cut way to save that and point to it so that you can keep developing that specific idea. You have make a photo of the last frame of video and start from that point again. If PIKA had a 'Continue From HERE' function it would rule the world. Because even using the "regenerate/cycle" button of a desirable image, creates an entirely different image. The question is: which company will create this simple function first? Conceptually, it's not hard to do.
@@TheoreticallyMedia Wow, Thank you! I’ve been searching and trying too many different AI video generators. This is awesome news for Pika. Can’t wait. Subscribed. :-)
Thank you so much! These deeper dives are a lot of work, but I really do love making them-- and it really means a lot to know that it is helping you all create!
"A Diner Named Crime"!! Your first gangster flick! Are you doing the music for it as well? One of the interludes could be, "Chewing My Coffee". Awesome guide, man! Thanks again.
Excellent! Honestly, try them all! They've been community tested. Some might work better than others, depending on what you prompt with the rest of your scene, but they are all worth a shot!
While i have done loads of Pika videos, i do appreciate that you are so organized with testing. Will check the pdf! Btw, did you include the "old" noise trick? By adding 4-10% noise to the input image in photoshop you get more movement to the video. I have not tried the movement parameter vs noise (it might be that pika adds the noise with movement parameter) but with noise one can experiment like adding noise only to the sky to get more movement to the clouds. Btw, are you aware of Full Journey? It's funny how ppl rave about camera controls in Gen2 and Pika when FJ has had those since the launch. It's also in discord and it does a quite lot more what pika and gen2 does. All three are based on Stable Diffusion so eventually the quality will be the same.
Hey, thanks so much for the comment! I didn't put the noise trick in, mostly because (if people want) I'll do a followup with advanced tricks/ideas (FinalFrame etc)-- I didn't want to overwhelm newer users with too much info! I do like Fulljourney too! At some point I'll have to do a video on them as well!
@@TheoreticallyMedia Finalframe trick has a problem in Pika: The follow up video that is done from the last frame is usually lower quality. In Gen2 it worked a lot better without difference on the quality.
When I try to upload the videos to TH-cam it isn’t full screen ..there are black spaces ..I get the images from midJourney ..does anyone kno how to fix this
Part of the reason I do those PDFs is just to keep everything straight in my own head! Bouncing back and forth between Pika, Midjourney, and all the other various AI apps can drive you crazy trying to remember all the various slash commands!
Awesome breakdown! From my early experiments, I really wanted to extend beyond the 3/4 seconds so I used an online tool to take a still of the last frame of the generation and used this as an image prompt for the next 3/4 seconds. However, the big issue I found with this is the still is lower resolution than the initial image prompt and so it generates a lower res video. I wonder what the way around this is. If there is a way around, and consistency can be achieved, the possibilities are greatly enhanced. Also need to work out how to get rid of the watermark!
Yup! Final frame is great! But yes, you do take a resolution hit every time you run it. Even in your first output with an image ref, you’ll notice that the first frame is the best, but the second frame (when the AI takes over) drops in quality. I think (still need to test) a way around this is to use final frame, then take that image and upscale/sharpen it, then run it through Pika. That might work/help. As far as watermark, I know you can reach out to them and pitch them a project or idea. They do have a non-watermark area where you can generate, but you have to get access to it,
@@TheoreticallyMediaNice, I didn't know that about the drop from the second frame on, makes sense though. The problem is that just there is a drop there, there will be another drop from this drop which is much more noticeable if a second generation is tagged on via final frame, as you know yourself! And if a third generation is tagged on, the quaIity will be even more downgraded. It reminds me of Chinese Whispers. Perhaps that's a great video idea actually, to incorporate the quirk. Chinese Whispers to the point of gross pixelation!I tried the upscale route in my early experiments but personally found that the final frame was downgraded to a point beyond satisfactory repair via this method. The drop was still noticeable and regrettable! I feel that once this barrier is overcome, floodgates will open in content creation.
@@austinmurphy8981 Ha! That Whispers though makes for an interesting narrative idea of someone who is losing their memories. I'd make it, but I think it sounds like it would depress me through production!
So, there’s kind of a work around by using a site called Final Frame. You export your last frame, and then use it to generate the next 4 second sequence. But so far, not yet officially. Hopefully coming soon!!
Currently, you can reach out to them to pitch them a project, or an idea, and they can give you access to a special channel that generates without a watermark. I'm sure at some point they'll go paid, and then it'll be automatic.
Yeah, but that’s across the board. I still say that we’re about 3 months from constancy in image generators (just a hunch)- which, obviously will lead to video as image references
Honestly I think that's kind of the charm of it though. It's outside of human normalcy. It doesn't adhere to what humans think of as realistic because it isn't really human. It's like an unconscious entity and it's really interesting and contradictory.
Thanks for the video. I just wanted to ask you how do you change camera? Is it possible to change them in the first time span of movie or you make for example 2 same videos but one with move left and other move up? Thanks
You can! the command would be -camera pan up left Granted, it is kind of one diagonal movement, not a hard Left and Up. You might try adding in a text prompt from one of the shot descriptions in the PDF, but I'm not 100% sure that would work. I tend to think that -camera is going to overpower any of those prompts...but hey, that's the fun in experimenting!
Great job on this video. Can vou please help me understand the distortion that comes in the videos from runway? I am getting massive distortion. Missing limbs, character going from anime to human when I want it to stay anime. Distortion on face. Thank you kindly
Thanks a lot for sharing! I wanna ask about the mentioned amount you get from TH-cam for this video. I do not understand what you have meant by the invalid traffic bug? I wanna build own channel and it might be something to be aware of?
Not yet. Still Discord based, so you have to basically run a search on your username. Or generate in the Pika Bot DMs. I feel like a web launch is coming soon though.
So, not yet. Since pika is still in beta, they don’t have a paid/watermark free version yet. BUT… You can message them if you have a project you’d like to do and if approved, you can get access to a watermark free channel on the discord.
@@HauntedMinutes haha, I spend a stupid amount of time looking through half baked documentation and do a LOT of experimenting! One of these days, a company will just hire me from the start to do their training videos!
The difficulty ran into is How to put multiple character in the same scene and talking to each other? For example I want 3 rabbits jump onto the same rock at the same time, how can you do that?
Hey Tim. Another great tutorial. Q: Have you done any testing with the quality of your input images when using image to video? Will you get a better result from using an upscaled or enhanced initial image? Or will generative AI programs just take the first frame from your reference image and turn it into its own AI soup, regardless? I haven't seen any one actually test/demonstrate their results. Would be interesting to see.
So, I think you take a slight hit on the second frame. I have noticed that as well. There is some advice out there on "Garbage in/Garbage Out" (In fact, that was mentioned by Tyber White)-- If I get to an "Advanced Pika" lesson, I'll include a section on upscaling a ref image to start. I think it does make a difference. (Again, just didn't want to overwhelm newcomers!!)
@@TheoreticallyMedia Thanks, Tim. It would be interesting to view a parallel test using an original MJ image straight from the AI oven, then run the same prompt using an enlarged, photo-enhanced version of the same image. As for the noobies... Pro surfers and grommits alike can all share the same ocean. As long as they don't cut in on my wave. 👋🏄♂️
@@morpheus2573 Ha! I always say: the most pro of us, once was a guppy! Great idea there, though! I'll make sure to pop it in. I also want to add some stuff like Video Upscaling and using a 3rd party interpolate. A number have mentioned that running pika at 12fps and then interpolating outside of it, has yielded some amazing results.
There seems to be diffence on using AI images vs photos. Every photo i have tried to use usually ends up looking quite bad. It messes up faces etc. With AI generated "photos" there is less problems.
That will be coming at some point. Currently, you can contact them with a project idea and if accepted, you'll get access to an area that generates without a watermark
If I could GIF comment back, it would be the Keanu Reeves one where he says "You're Breathtaking!" Why can't we have .gif comments, TH-cam?! They'd be so much fun!
@@TheoreticallyMedia All the comment is more than enough. You probably can't see it but I had to use inside face for this profile pic which I love. Guess who I learned that from? The best that's who
I mean, I think MJ just tends to be used/referenced a lot since it is kind of a major player in the field. If you're happy w/ NightCafe, stick with it! AI Art needs different looks and different perspectives!
Yeah, they just went paid after a very long having a free unlimited tier. I did just post up a new option that is free and unlimited (for now at least): th-cam.com/video/v-V8hVv2veE/w-d-xo.html
Nope! No shady business here at the Crime Diner! Here is your To-Go bag. Oh, you only had coffee? Well, it’s just a little slice of…uhh, the pie. Yeah, that’s the ticket!
wait-- what? That shouldn't be the case at all. If you put in 0 dollars, it is free. Please let me know if you're running into anything weird there, because that should 100% NOT be the case. In a worst case scenario, please email me at theoreticallymedia@gmail.com and I will email one back to you!
yeah, that should not be the case at all! Again, if you are running into any issues, please just email me at theoreticallymedia@gmail.com and I will email you a PDF back!
Yeah, I get you-- I but I also get yelled at if I gloss over things like How to use Discord, or how to DM bots. Better to err on the side of comprehensive than leave newcomers out.
I don't know, it's ok, but not that great, 3 seconds of video. I guess we should be grateful, a lot of editing matching is required, it has not saved any time at all, more time is spent. But I don't have to pay an actor. Whatever. Another time consumption. .
Download the FREE PDF Workbook here: theoreticallymedia.gumroad.com/l/pika
Oh!!! You have to buy/donate money to get access? Is this correct? You didn't mention that part in your video. Please do so in future. It won't feel so tricky. Thanks, Man. Love your videos by the way. Cheers!
@@Qmannn2UTube No, no...For the PDF? Or for Pika? Either way: Not at all. Both are 100% free. On the Gumroad side, just put in zero dollars for the cost.
@@Qmannn2UTube And thank you!!
Thanks, Tim! You are awesome! Nobody else even comes close to your quality of videos on AI generated videos; so glad I found you (also, thanks, algorithm!).
Not free my friend
Thanks for the feature! Solid Pika Guide. Sharing with friends.
Thank you Sway!!
Having the ability to camera control is truly awesome!!
And this is just the start! In the next few months I think we’re going to see a pretty massive explosion in 3d control of 2D images.
@@TheoreticallyMedia that’s crazy!
One of the best Pika tutorials of all time ❤ Thank you Tim!
Oh, thank you sooo much!! Really looking forward to Pika 2.0!!
this is the best ai channel on here right now.
Excellent to-the-point video. I didn't know about the camera parameter getting included. I just read Kaiber is doing the same (or has done). Glad you brought that up. Hope YT fixes itself. 👍
I’ll double check on Kaiber (I’ve been wanting to circle back to them anyhow!), but I think that’s Gen2 you’re thinking of.
But yeah, those camera controls are brand spanking new!
@@TheoreticallyMedia Yes! I believe it's Gen-2. 💯 There are just way too many 'toys' right now!
Incredible investigation into Pika Tim! I am just starting to investigate the tools and the PDF is extremely helpful for getting into some common commands. I also admire your channel and am excited to see where it goes!
So glad to hear! And I’m really happy the PDF is helpful! Those are a whole extra step in making these videos (and are hopefully typo free?) so I’m really happy when I hear they’re useful to you!
Thank you so much for this! Perfect timing as I just found the discord yesterday and needed a tutorial like this!
excellent to hear!! You caught this one at the right time too, since the new -camera feature was just released, and I think I'm the only tutorial that has it. I mean, they'll probably release some new amazing feature tomorrow, and I'll be out of date...
...but for TODAY!!
Thank you! I've been waiting for a tutorial so I can start using Pika. Appreciate all the hard work you do for the community!
Not a problem at all! I LOVE Pika...It has some really special sauce going on!
Loved this. So helpful. Wasnt aware of the negative prompt options!
Personally, I think the neg prompts are the real gamechanging secret sauce.
It’s like early SD, your neg has almost as much impact as your positives!
Let me know how it works out for you!
Love the ai community and meeting many wonderful souls in the creator rooms
That was totally what I was slyly implying in this video when I was saying: you can use the bot, but try out the gen rooms first, because there are a lot of great people there!
GREAT CHANNEL! Great informative videos! I bought your pdf, gonna spend the day learning PIKA because of you! Thanks!!! -Tommy
That's awesome to hear, Tommy! Hope it was a super fun day!
Very nice breakdown as always. Your understanding and knowledge of filmmaking in general definitely gives you an edge on deeper dives into the cinematic side of things - which I personally enjoy. And I feel those without that knowledge can and will learn a ton from your content. It's like a hidden secret society when you meet other filmmakers or those that worked in the industry, haha.
We need a secret handshake! And a clubhouse....hmmm, maybe that Crime Coffee shop? Meet ya there!
(BTW: Are you in Chicago? Why do I think that? Anyhow-- just marathoned The Bear over the last week or so...my god, what an amazing show. High Recommend if you haven't watched it already!)
Crime Coffee would be an amazing hangout. Can have a double meaning, "Coffee is so good it's a crime". The Bear is amazing show, I haven't watched it all just yet. I just rewatched Murder By Death which I can never get tired of. I'm not in Chicago, currently Nashville but moving back to Los Angeles soon - Me and the family wanted to check out Nash for a bit and I needed a small vacation.@@TheoreticallyMedia
Brilliant tutorial. Really appreciate you putting in so much information rather than stringing the video out. Love the up-and-left camera moves. That's a game-changer. :-)
Thank you so much for this. From Midjourney to Pika you have been so helpful.
Oh man, thank you so much! It's been a real pleasure learning these tools and passing the info to you all!
Awesome overview and comprehensive details. Thanks for another great one AND the handy Workbook!
Awesome! Thank you so much!! And I'm glad the PDF is coming in handy! I honestly think it just beats screenshots or having to return to a video 30 times!
Good morning ! Thank you very much for your tip about hidden members. I've been looking for a private connection on Pika Labs for a long time without success. Thanks for thinking of reporting this.
thanks for the pdf, you're a hero ❤
Thank you so much! I’m really happy these PDFs are helpful to everyone! They do take a little extra time (and a fun game to play is “look for the typo” because there are 5 or more put in….unmmm, intentionally, yeah- that’s it; intentionally!) but it’s always worth it when I get a comment like yours!
Great tips and demo. Thanks! Very fun!
Excellent to hear! Pika is a REAL blast!!
Thank you, just what I needed to get to grips with this. Subbed
That is fantastic to hear! Pika is quickly becoming my favorite Video Generator. The results are outstanding! Have a blast, and if you run into any questions, please feel free to shoot them over!
Excellent video thanks!
Fantastic to hear!! Thank you so much for the comment! Really means a lot!
Great guide Tim! Thanks so much for putting this together. 🔥
Thanks so much, Cory!! This one was super fun to dive into!!
Pretty gobsmacking stuff. Thank you for your exceptional content.
Thank you so much for this. I just started using Pika today
Very in-depth thanks. Came here for the image uploader bc the MidJourney process did not work.
Which Midjourney process didn’t work?
@@TheoreticallyMedia I got it working. Thanks.
That pdf is amazing. Thank you
Thank you so much! I know Pika has updated to the web UI since, but I'm happy that the info is all still valid!
Thanks for the share. Pika is amazing
It legit impresses me more and more every day!
Thanks for the video and full explanation
Super happy it was helpful to you!
Great work, Tim! Thanks for sharing.
Thanks for all your advice as well Tyber! I didn’t get it all in (didn’t want to overwhelm folks getting started) but there is enough leftover for me to do an “advanced pika” video, so I’ll add your advice there!
Thank you for PDF. WoW
Excellent! Always happy to hear the PDFs are useful!!
Excellent right-to-the-point explanation.
Thank you so much! Haha, still got up to 14 minutes! But-- I tried not to waffle too much! (I mean, the Coffee Shop thing, maybe-- but that was too funny not to share!)
I have to say I just don’t understand at all the use of 4 sec hiccup repeat vids but at least in Splice I can reverse loop but of course hair this image/text to vid Amazing and every week better and better as Ai DO! 🎉
I think that 4 second limit will be broken fairly soon. Gen-2 already has an "extend" feature. My question is: What should the max length of a shot be? I personally don't think we need 1-min long shots. 10 sec? Is that enough? I'm curious to see where it ultimately lands.
Great video ! Thanks !
100%! Glad it was useful to you!
🎯 Key Takeaways for quick navigation:
00:00 🎬 Introduction to Pika Labs, an AI video generator you can currently use for free that was recently updated with new features
00:57 🖼️ You can prompt Pika with text or images. Text prompts follow a structure like: medium, style, scene description, action description.
03:45 🎥 List of common "called shots" that work well for text prompting in Pika
06:33 🖼️ Image prompting allows Pika to use the image for contextual clues. Can also animate images with /animate
08:24 ⚙️ Pika commands like FPS, motion, guidance scale, negative prompt, aspect ratio, seed, and camera controls
13:29 💡 Tips and advice from Pika community members on best practices for prompting
Made with HARPA AI
I mean...there are chapters too.
@@TheoreticallyMedia its my ai assistant's comment if u want I'll delete it, ur video is great BTW thanks
Quelques mots en français histoire de donner une note exotique à ce fil. Merci pour les videos enthousiastes et souvent passionnantes, et toute cette généreuse contribution à notre créativité collective. Amitiés
Amazing application and this is a brilliant presentation.
Thank you! You're amazing!
No, no, YOU’RE amazing!!
What's very frustrating about all of the AI video generators is that when you get something you like, there is no clear cut way to save that and point to it so that you can keep developing that specific idea. You have make a photo of the last frame of video and start from that point again. If PIKA had a 'Continue From HERE' function it would rule the world. Because even using the "regenerate/cycle" button of a desirable image, creates an entirely different image. The question is: which company will create this simple function first? Conceptually, it's not hard to do.
So, check out the latest video I did on Pika 1.0 (new model releasing soon)- extend is in there!
@@TheoreticallyMedia Wow, Thank you! I’ve been searching and trying too many different AI video generators. This is awesome news for Pika. Can’t wait. Subscribed. :-)
How do you increase the video length to more than 3 secs???
Hi, thanks for this video. Very helpful.
Excellent to hear! You are quite welcome!
Excellent Tim :)
Thank you! This was so much fun to put together!
This could be great for making GIFs!!
I have actually done that, and it works great!
Thanks for these vids they really help a lot🙏🏿😁
Thank you so much! These deeper dives are a lot of work, but I really do love making them-- and it really means a lot to know that it is helping you all create!
Really helpful!🤝
Thank you so much for this
100%! Glad you found it helpful!!
leave becky alone! hey im new here and Im liking the content... keep them coming old man!🥰
Haha. It’s been a bit since I saw this video! I totally forgot about Becky! She’s sooooo weird!! Hate Becky! Ruins everything!
very well explained.
Thank you!!!
"A Diner Named Crime"!! Your first gangster flick! Are you doing the music for it as well? One of the interludes could be, "Chewing My Coffee". Awesome guide, man! Thanks again.
Yes!! Yes to all of this!!! Seriously, my biggest laugh of the day here!
@@TheoreticallyMedia Cool. Anytime! I got a million of 'em. Hey, so when you do a rom-com: "And Then There Was Becky"!
Does anyone have cheat sheet for runway ml prompts such as for the neg prompt’s camera prompts ?
Thank you
5:52 i like the shots list which can help in making shirt film
Excellent! Honestly, try them all! They've been community tested. Some might work better than others, depending on what you prompt with the rest of your scene, but they are all worth a shot!
Geniooo! Excelent tutorial !!!
Awesome to hear! I have an updated sneak peek at 1.0 here: th-cam.com/video/cte-UPyrhIs/w-d-xo.html
While i have done loads of Pika videos, i do appreciate that you are so organized with testing. Will check the pdf! Btw, did you include the "old" noise trick? By adding 4-10% noise to the input image in photoshop you get more movement to the video. I have not tried the movement parameter vs noise (it might be that pika adds the noise with movement parameter) but with noise one can experiment like adding noise only to the sky to get more movement to the clouds.
Btw, are you aware of Full Journey? It's funny how ppl rave about camera controls in Gen2 and Pika when FJ has had those since the launch. It's also in discord and it does a quite lot more what pika and gen2 does. All three are based on Stable Diffusion so eventually the quality will be the same.
Hey, thanks so much for the comment! I didn't put the noise trick in, mostly because (if people want) I'll do a followup with advanced tricks/ideas (FinalFrame etc)-- I didn't want to overwhelm newer users with too much info!
I do like Fulljourney too! At some point I'll have to do a video on them as well!
@@TheoreticallyMedia Finalframe trick has a problem in Pika: The follow up video that is done from the last frame is usually lower quality. In Gen2 it worked a lot better without difference on the quality.
When I try to upload the videos to TH-cam it isn’t full screen ..there are black spaces ..I get the images from midJourney ..does anyone kno how to fix this
What software are you using to edit?
@@TheoreticallyMedia capcut
great instruction. thanks
Excellent to hear!! Pika should be dropping an update soon, so I’m looking forward to circling back!
It’s like learning a new language
Part of the reason I do those PDFs is just to keep everything straight in my own head! Bouncing back and forth between Pika, Midjourney, and all the other various AI apps can drive you crazy trying to remember all the various slash commands!
Awesome breakdown! From my early experiments, I really wanted to extend beyond the 3/4 seconds so I used an online tool to take a still of the last frame of the generation and used this as an image prompt for the next 3/4 seconds. However, the big issue I found with this is the still is lower resolution than the initial image prompt and so it generates a lower res video. I wonder what the way around this is. If there is a way around, and consistency can be achieved, the possibilities are greatly enhanced. Also need to work out how to get rid of the watermark!
Yup! Final frame is great! But yes, you do take a resolution hit every time you run it. Even in your first output with an image ref, you’ll notice that the first frame is the best, but the second frame (when the AI takes over) drops in quality.
I think (still need to test) a way around this is to use final frame, then take that image and upscale/sharpen it, then run it through Pika. That might work/help.
As far as watermark, I know you can reach out to them and pitch them a project or idea. They do have a non-watermark area where you can generate, but you have to get access to it,
@@TheoreticallyMediaNice, I didn't know that about the drop from the second frame on, makes sense though. The problem is that just there is a drop there, there will be another drop from this drop which is much more noticeable if a second generation is tagged on via final frame, as you know yourself! And if a third generation is tagged on, the quaIity will be even more downgraded. It reminds me of Chinese Whispers. Perhaps that's a great video idea actually, to incorporate the quirk. Chinese Whispers to the point of gross pixelation!I tried the upscale route in my early experiments but personally found that the final frame was downgraded to a point beyond satisfactory repair via this method. The drop was still noticeable and regrettable! I feel that once this barrier is overcome, floodgates will open in content creation.
@@austinmurphy8981 Ha! That Whispers though makes for an interesting narrative idea of someone who is losing their memories. I'd make it, but I think it sounds like it would depress me through production!
@@TheoreticallyMedia maybe it should be reversed so, a journey towards clarity!
@@austinmurphy8981 oh, that's brilliant. And...maybe made me cry just a bit. Haha, GO MAKE THIS!!
Is there any way to continue the video from the 3s to form a storyline
So, there’s kind of a work around by using a site called Final Frame. You export your last frame, and then use it to generate the next 4 second sequence.
But so far, not yet officially. Hopefully coming soon!!
how do we get the water mark off?
Currently, you can reach out to them to pitch them a project, or an idea, and they can give you access to a special channel that generates without a watermark. I'm sure at some point they'll go paid, and then it'll be automatic.
basic problem still prevails. No consistency in characters, clothes, location etc
Yeah, but that’s across the board. I still say that we’re about 3 months from constancy in image generators (just a hunch)- which, obviously will lead to video as image references
Yes I notices that also.
Honestly I think that's kind of the charm of it though. It's outside of human normalcy. It doesn't adhere to what humans think of as realistic because it isn't really human. It's like an unconscious entity and it's really interesting and contradictory.
Because you just write simply, he need to re imaginate it always, using same seed helps a little just a little
is there a way to stop pikalabs from making my image I upload so blurry?
Damm just spent the whole day prepping some video and should have seen this. I ended up using Kaiber instead.
Oh sorry! Well, next video in that case! I do like Kaiber quite a bit as well though!
Superb run through!!! Fantastic, valuable content!!
Thank you so much!! Props to everyone in the Pika community for helping out and offering advice as well! That crew is great!
Thank Tim!
Thanks for the video. I just wanted to ask you how do you change camera? Is it possible to change them in the first time span of movie or you make for example 2 same videos but one with move left and other move up? Thanks
You can! the command would be -camera pan up left
Granted, it is kind of one diagonal movement, not a hard Left and Up. You might try adding in a text prompt from one of the shot descriptions in the PDF, but I'm not 100% sure that would work. I tend to think that -camera is going to overpower any of those prompts...but hey, that's the fun in experimenting!
@TheoreticallyMedia thank you so much for your quick response. Yeah, it would be fun exploring it.
Great job on this video.
Can vou please help me understand the distortion that comes in the videos from runway? I am getting
massive distortion. Missing limbs, character going from anime to human when I want it to stay anime.
Distortion on face.
Thank you kindly
Thanks a lot for sharing! I wanna ask about the mentioned amount you get from TH-cam for this video. I do not understand what you have meant by the invalid traffic bug? I wanna build own channel and it might be something to be aware of?
Does it have a gallery where i can check my genrates ?
Not yet. Still Discord based, so you have to basically run a search on your username. Or generate in the Pika Bot DMs.
I feel like a web launch is coming soon though.
@@TheoreticallyMedia I figured out by clicking mentions that's sorta a way but thanks man
thank you!
1000%! If you’re looking for another free video generator (Pika has gone paid) you might want to see Haiper. I did a recent video on it as well!
great channel
Hey, Tim, is there any way to upgrade to get rid of the Pika Labs logo? I can't fins anywhere to pay!
So, not yet. Since pika is still in beta, they don’t have a paid/watermark free version yet. BUT…
You can message them if you have a project you’d like to do and if approved, you can get access to a watermark free channel on the discord.
@@TheoreticallyMedia Cool! Thanks, mate. How do you know all this??
@@HauntedMinutes haha, I spend a stupid amount of time looking through half baked documentation and do a LOT of experimenting!
One of these days, a company will just hire me from the start to do their training videos!
What do you prefer- Pika or Mid Journey?
Once you've created a short AI video, are we allowed to use it in our personal videos, and publish them on YT?
Sure. In fact, I think Pika encourages it.
cant join pika lab says unable accept invite
this isn't working? discord.com/invite/pika
The difficulty ran into is How to put multiple character in the same scene and talking to each other? For example I want 3 rabbits jump onto the same rock at the same time, how can you do that?
Hey Tim. Another great tutorial.
Q: Have you done any testing with the quality of your input images when using image to video? Will you get a better result from using an upscaled or enhanced initial image? Or will generative AI programs just take the first frame from your reference image and turn it into its own AI soup, regardless? I haven't seen any one actually test/demonstrate their results. Would be interesting to see.
So, I think you take a slight hit on the second frame. I have noticed that as well. There is some advice out there on "Garbage in/Garbage Out" (In fact, that was mentioned by Tyber White)-- If I get to an "Advanced Pika" lesson, I'll include a section on upscaling a ref image to start. I think it does make a difference.
(Again, just didn't want to overwhelm newcomers!!)
@@TheoreticallyMedia Thanks, Tim. It would be interesting to view a parallel test using an original MJ image straight from the AI oven, then run the same prompt using an enlarged, photo-enhanced version of the same image. As for the noobies... Pro surfers and grommits alike can all share the same ocean. As long as they don't cut in on my wave. 👋🏄♂️
@@morpheus2573 Ha! I always say: the most pro of us, once was a guppy!
Great idea there, though! I'll make sure to pop it in. I also want to add some stuff like Video Upscaling and using a 3rd party interpolate. A number have mentioned that running pika at 12fps and then interpolating outside of it, has yielded some amazing results.
There seems to be diffence on using AI images vs photos. Every photo i have tried to use usually ends up looking quite bad. It messes up faces etc. With AI generated "photos" there is less problems.
@@TheoreticallyMedia Thanks. Looking forward to that. 🤩✨
how to remove pika labs watermark?
That will be coming at some point. Currently, you can contact them with a project idea and if accepted, you'll get access to an area that generates without a watermark
you're the best
If I could GIF comment back, it would be the Keanu Reeves one where he says "You're Breathtaking!"
Why can't we have .gif comments, TH-cam?! They'd be so much fun!
@@TheoreticallyMedia All the comment is more than enough. You probably can't see it but I had to use inside face for this profile pic which I love. Guess who I learned that from? The best that's who
All I ever see are midjourney refrences. Did I make a mistake in going all in on NightCafe? Like what.
I mean, I think MJ just tends to be used/referenced a lot since it is kind of a major player in the field. If you're happy w/ NightCafe, stick with it! AI Art needs different looks and different perspectives!
So according to my understanding it’s $58 monthly for there plan unlimited use and users only get 3 seconds of video ?
Yeah, they just went paid after a very long having a free unlimited tier. I did just post up a new option that is free and unlimited (for now at least): th-cam.com/video/v-V8hVv2veE/w-d-xo.html
I love you bro, u funny as hell but I don't know why🤣
Run on TT Bro! 😎
The link aint workin pops.
Which link? To the PDF?
It is
How to add dialogues to videos...?
"... think her name is Becky" LMAO
BECKY IS THE WORRRRRST!!!
super
Thank you!! Hoping to do a part 2 with advanced Pika tricks in the near future!
In this tutorial only 2 prompts are shown. Please write more prompts for character scenes.
I mean, the format is there, which I think is the more important part. I’d rather give you a kitchen than give you one recipe, y’know?
@@TheoreticallyMedia Thanks🙏 you made a nice tutorial. But it will be nice if you show us few more example for longer scene creation and all that...!
come on down to the crime diner for nothing more than coffee and definitely NOT money laundering
Nope! No shady business here at the Crime Diner! Here is your To-Go bag. Oh, you only had coffee? Well, it’s just a little slice of…uhh, the pie. Yeah, that’s the ticket!
Ur so sweet thanks for all u do
Happy to do it!!
Diner named “Crime” 😂
“Try the special tonight…it’s to (ahem) die for…”
These PDF workbooks are not free. You have to pay something before you receive the PDF. Not that I mind donating, but free they are not.
They are free, Its asked for a donation you don't have to give one.
The page would not let me download the file without a donation. So that's not free.
,@@MONTY-YTNOM
wait-- what? That shouldn't be the case at all. If you put in 0 dollars, it is free. Please let me know if you're running into anything weird there, because that should 100% NOT be the case.
In a worst case scenario, please email me at theoreticallymedia@gmail.com and I will email one back to you!
yeah, that should not be the case at all! Again, if you are running into any issues, please just email me at theoreticallymedia@gmail.com and I will email you a PDF back!
Did you put a 0 in the donation box. Mid you it's worth at least a dollar. Won't break your bank @@artrellamack2380
6:43 actually starts creating a video
Yeah, I get you-- I but I also get yelled at if I gloss over things like How to use Discord, or how to DM bots. Better to err on the side of comprehensive than leave newcomers out.
@@TheoreticallyMedia It's your channel, do you, I have no issue with it... was just putting that there for those of us who wanted to advance forward.
@@SpiritTracker7 Right on-- I think I do have chapters. no?
Its GoCHHH no Go :)
Wait...wha?
LOL The Dutch say Goch, no van Go , you pronounce the ch :)
@@TheoreticallyMedia
I don't know, it's ok, but not that great, 3 seconds of video. I guess we should be grateful, a lot of editing matching is required, it has not saved any time at all, more time is spent. But I don't have to pay an actor. Whatever. Another time consumption. .
It’s getting there! I’d say a few more months until we see something pretty major pop in the AI video space!
discord - yuk, bye bye
Don't miss out! There are a lot of great tools (free!) that you can use on Discord! Admittedly, not the most handsome of UIs, but still!
Thanks 👍🏽, 💯 ,
100%!!