I used to rely on mobile filters or experiment with effects in After Effects to achieve cartoon effects on my videos few years back, but it was a the process was not smooth, and I was never quite satisfied with the outcomes. Now, thanks to you sharing your knowledge and your detailed video tutorial on the subject, I find it worth to try I made sure to hit the like and subscribe button in appreciation of your help!
You can find all the models that my workflow uses on huggingface.co . There are many more models you will need. That particular one is here: huggingface.co/wangfuyun/AnimateLCM/tree/main (sd15_lora_beta.safetensors) and put it in this folder 📁 ComfyUI_windows_portable\ComfyUI\models\loras I will work on a detailed workflow video with a list of where you can find each of the models.
I am a professional animator myself. I started in 1998 and we always used new tools that made the animation process faster. Not sure what offended you? Also aside from letting people know that there are new tools I can't stop the progress of technology. Neither can you. I can only assume that you are angry that AI is changing the animation industry. Then again technology always did that. I used to work with the old timers who worked on paper and when we first switched to scanning the hand drawn animation for clean up and final delivery, some of them also got angry. But it made work a lot faster so it was adopted and become part of the animation process. The same thing I seen when we first started incorporating 3D animation along side classical animation. Then 3D animation pretty much killed traditional animation. Which I personally love. I was sad when even Disney closed their animation studios in Cali and Florida. But now new tools are slowly bringing back classical animation. Things change. I didn't create this new tech. If you don't like it not sure why you would be angry at me?
There are many fundamental ethics issues that come with using AI tools, which both traditional artists and animators are trying to push back against. It is not simply a matter of efficiency. I would encourage you to look into how AI generative art tools are trained and how artists' work is stolen and fed to these algorithms.
@@TomAntosTom I support your move, everyone that are mad in the comments are frustrated with the quick changes especially when they just finished expensive classes to learn animations, but learning the newest techniques will always help you ahead of the curve, everyone else it would be too late for them to adapt because they were too late in the move, keep doing what you like, you have been in this business longer than any of your subscribers, don't listen to the hate, change is coming and there is nothing they can do about it
@@joeteti4958If you really think animators are upset simply because they don't know how to use the tools, then you completely misunderstand the issue.
@@troopercooper44 if your talking about artwork being stolen? It's too late and you don't understand the issue, if you don't know how the models work, they are trained on millions of artwork, meaning they study the style of art, AI is a algorithm of information of why a line is there and why a object in in a different location, it's exactly like you copying the style of Van Gogh, you study artists and replicate their art, but now with math and neural systems we can now skip a step and give concept art piece, art will always be art either handmade or AI, but just like everyone says including Tom in the video it doesn't replace creativity, your canvas is endless, now is about adapting to the change because it's here either you like it or not, pouting on unfairness will only leave you at the starting line, and another reason to pout, AI is a tool, use it to your advantage
This is amazing! Thanks for sharing. I worry though about the sources that the AI is using and the ethics behind it all. Is there a way to be sure prompts used don't source artists work uncredited?
@@roastable A "style" can not be copyrighted. If I went and traced the photo or video that was created by someone else then that would be illegal (in some cases) and artistically immoral. Even if I did that tracing in my own original style. The video is mine and the style is a MIX of various styles that convert my video into a cartoon look. Its a video filter. Same as photo filters that have been around for a long time that legally and artistically have never been a problem. Just because this filter can now be coherently applied to a video sequence thanks to AI does not mean anyone's work has been stolen. There is a reason why it was never a problem for artists to use a particular style and why it's not illegal. Even in US copyright laws under "fair use" case it has been proven again and again in court that a style can not be copyrighted unless it is a direct copy of a work or registered trademark. This is what allows parodies to be made or documentaries about any topic that criticize a particular person including artists and their work.
Honestly, of all the AI companies I would not recommend using at this point it would be Mid-Journey, and the reason is that the whole AI latent space copyright infringement issue has not been resolved in the courts yet, and historically how latent space stores the image permanently would not traditionally be considered fair use. Anyhow just an opinion. people can do what they want, but it might bite you if you're down the road. Just try to keep aware.
@@TomAntos That's a difficult to answer question, because a lot of it comes down to what a given AI was trained on. So while one might be good at one thing another might be better at something else. The thing to try and minimize is using AI that were trained on copyrighted material. some companies have put in warnings saying their results should not be used for commercial purposes, others ( like Mid-journy) are gambling on latent space being declared fair use (most lawyers with backgrounds on the topic are saying it's likely not, which is why I gave the warning), Others ( like adobe) are using their plethora of stock image/videos for training. I think one should try and find one that they're okay working with and see if they can get it to produce what they want. If they don't have luck try another.
Tom great video ! wow where tech is going is incredible - how long from concept to finished product did this take ? love your content ! hello from Toronto !
The whole thing took me almost a month as I tried to figure out how to setup the AI models and then how to do the whole process. Now that I know how to do it and have the workflow with AI models setup I could do this in 4 days. 1 day to shoot the live action, 2 days to do the AI filter conversion and 1 day to composite all the shots.
There are video editing programs that will edit a whole scene for you based on music beats etc. I haven't tried AI for video editing myself other than little tests. Personally I prefer to edit myself and retain complete control over it. AI video editing feels too random to me and I was never really happy with the results.
Thanks for this tutorial! I have downloaded ComfyUI and everything else you mentioned, and loaded your workflow into ComfyUI and I chose a video to input. However, then I am getting stuck: the next nodes (height and width) are red and don't seem to let me input height or width (unlike in your video, there is nowhere to type, just red boxes). Let me know if you have any ideas!
It won't work until you install all the models that the workflow uses. You can find them on huggingface.co I will post detailed instructions on my website as to what models you will need and what folders to put them in. It's too long to explain here. I will let you know when that is up.
@@TomAntos Hey there, just wondering if there's an update on this? I seem to be running into some issues as well and couldn't find the node models. Appreciate your help!
Disappointed by this. Were artists whose work were used for the art style compensated? Unsubbed as this is a clear statement by you that artistic IP has no value.
In this video I created the work that AI used as reference material. I also created the idea and specifically storyboarded each shot. The style of others has always been used as an influence. When an artist is successful then its natural that his work then inspires other artists to do work that's similar. Sometimes artists base their style on a mix of various other artist's styles and that's how they by accident create a new style. Never has an artist just lived in complete vacuum and created something out of nothing. This happens all the time in paintings, music and writing styles. It's why there is Anime style of animation that is similar yet still created by many different artists. Same with different styles of music. AI does the exact same thing when used correctly. If however there is no personal input at all then AI will simply copy various other works and mash them together. I am not saying AI is good or bad. It's here regardless if we like it or not. It's affecting a lot more than art. I'd rather it would be used for good... as in speed up the workflow instead of straight up copy other people's work. But technology is just a tool. Like any tool it can be used for good or for bad. I am nearly experimenting with it and trying to see if it can be used to enhance and speed up my workflow not to steal other's work. So I reject your conclusion that I made a "statement... that artistic IP has no value". I've been working on films for over 25 years and I've had my own work stolen. I've had other artists take my design sketches and scripts and then financially benefit from it. I've also had films that won awards and got millions of online and each time I released a successful film, there was multiple knock off copies of my work. In my video here I specifically said I am trying to create a workflow where I can direct the AI instead of just letting AI on it's own copy other's work. Many programs and apps already had filters you can apply to your photos that will turn it into a sketch or drawing or Vincent van Gogh painting. The way I use the AI here is in exactly the same way. It's a filter. AI simply helps to apply that filter to a whole image sequence while keeping the motion coherent, instead of applying it to an individual image. I never once in the video encouraged people to steal other people work. I even said at the end that while AI will change art and eliminate some jobs I don't think it will eliminate creativity. I think some people will no doubt use AI to mimic or copy other people's creativity. However I think nothing can beat real human ideas, even if those ideas were brought to life with the help of AI.
@@TomAntosjust because other people or things are doing something doesn’t make it right. There is a difference between being inspired and learning an artist’s process to make it your own versus essentially tracing their art and claiming it as your own. It’s one thing to practice via tracing but don’t profit off of it or claim it as your own. Sure, you storyboarded and shot the video and that is yours, but the painted part is not and you shouldn’t pretend that it is. This is some really gross reasoning on your part.
@@roastable before you write a comment please slow down and think of your own logic. I didn't trace anyone's work. The video is mine and the style is a MIX of various styles that convert my video into a cartoon look. Its a video filter. Same as photo filters that have been around for a long time that legally and artistically have never been a problem. Just because this filter can now be coherently applied to a video sequence thanks to AI does not mean anyone's work has been stolen. There is a reason why it was never a problem for artists to use a particular style and why it was never illegal. Because a style can not be copyrighted. If I went and traced the photo or video that was created by someone else then that would be illegal and artistically immoral. Even if I did that tracing in my own original style.
@@TomAntos I didn’t mention anything about copyright in my previous comment and yet you’re feeling the need to get defensive. Tom, I know that you know that “filter” is not enough of a specific term to cover what you’ve done here. The point is not that your video has a look to it, it’s that you had no hand in creating the look. For example, you could have made your own artistic rendition of a frame as a painting and used AI to extrapolate across your footage but you did not - you took a software-generated bastardized amalgamation of other people’s paintings (also important to note that “people have always been inspired by other people’s art” doesn’t stand here because you didn’t get inspired, a piece of software “did”). There is nothing creative about this beyond the footage you shot. I’m frankly done with discussing as there’s not really another way I can put it for you to understand but I wish you well and I hope you take into account the impact your platform has and think beyond “other people are doing it so it must be okay”.
Don’t listen to the haters Tom you are definitely evolving. I would probably implement some subtle camera movements by adding foreground elements on top of 3D layers or in this case nodes in 3D to take away that obvious static green screen compositing result.