Mom! I’m a YouTuber! Or how to Generate a YouTube channel.

How to create a YouTube channel using only artificial intelligence and persistence.

Ever since my school days I have been dreaming about my youtube channel, to put up videos and watch the view count increase, yes it’s not a bad dream. But what if I told you that if you also want to make youtube videos you can already start right now.

First of all let’s understand what is the thing about youtubers, you are a person who in some niche tells his point of view, makes videos on certain topics, or just video documentation of his life sharing with everyone who wants to, also they are opinion leaders and active participants of cammunity. But how surprised I was that many popular channels can be without important aspects such as author, and creative editing. Many youtube niches involve the usual storytelling with a visual background. This is the niche I chose — hour-long videos in which a monotone voice tells bikes and stories that you can fall asleep to.

How to write a video script?

And so, where to start. First, as in all absolutely works you need a text that is the script of the video. With this task is not bad to cope with ChatCPT, worthy conditionally free analogs I have not found. Yes through ChatCPT in the free version gives a small result in 1–2k words, which for an hour video is very little. But what if you first make a plan of different points and chapters, then write a promt in which you clearly formulate the tasks that you want to achieve. Yes, then you will be able to achieve a result of 20,000 words, which is enough for an hour-long video.

ChatGPT

Voiceovers

I’ve also always wanted to make a grid of channels in different languages, this idea has haunted me for several years after my experience with a similar company. So what would you need if you wanted to do a channel grid in normal conditions, with the footage you shot with the actors in the studio and edited? Quite right, it’s translation and dubbing in different languages. But what if you don’t have money for translators and people who speak Spanish, or German, then we turn to good old neural networks, although it’s not as good as with a live person, but still for a budget project it’s quite acceptable results I think.

Here ChatCPT has written everything for us beautifully, if possible, we have already made edits and open Deepl is a text translator, based on AI.

Deepl

Here everything is simple, copy-paste, huge texts are translated in a couple of minutes, which can not but please. Now we have two or more versions of the text in different languages and we need a voice that will dub it all. I have tried many text-to-speech services, but how many nerves and time used. From what I can conclude, if you have some budget or your timekeeping is designed for 10–15 minutes use ElevenlabsIO, a comfortable convenient resource with a fairly high quality sound, which is not in the competition (this is not an advertisement). If the budget for video creation is limited, then to your attention Clipchamp. I use MasOS so I do not know what possibilities Clipchamp opens on the Windows platform. In my disposal only the web version, which I was quite pleasantly surprised, a good basic video editor and in your browser, yes technology is moving kilometer steps. You have a lot of different languages, and a huge number of speakers, but the sound quality is far inferior to the aforementioned ElevenlabsIO. Also what else is cool Clipchamp, in the immediate after the generation of voice can create subtitles with the ability to upload with timecodes.

Clipchamp

After creating a voiceover on the first videos I ran them through Adobe Podcast but everyone who used Adobe products know that at the time of work you are lovers and enemies, Adobe Podcast is not an exception, after that I decided to just manually tighten the sound by pulling the equalizer and threw a couple of effects, the difference is not particularly noticeable and saves time and effort incredibly how much.

Visual

And so, video is not video without a visual component, to help come AI on the type of text-to-image, here everything is simple, the budget allows us to open the ambasador of this niche Midjourney, if at the first stages you are greedy, then look for analogues or use the analog of ambasador LeonardoAi. At the moment I am experimenting and using the results of both neuroites. For good results you need a good promt, it can also be written with the help of good old ChatCPT, well or generation by short query, yes the result is not always as you want but still more or less acceptable.

Promt

LeonardoAI

Further it is purely my whim, if you do not pursue a high from the visual picture, this there can safely skip. I’m about increasing the quality of the picture, do with hd to 4k, and at the same time kill your storage, because with 1mb you become more than 50 or 100mb image. If you have already decided to such a stupidity TopazAI at your service, I use their software since the first versions, and very satisfied with the result, free or other worthy analogs do not know.

TopazAI

I glue everything together in DaVinci Resolve. If you will use LeonardoAI and DaVinci Resolve you will most likely have a Florentine work (just kidding). Montage is simple and without any tricks, just put everything into a beautiful picture, if you want you can use futazhi for visual dilution, stretch our images on the timing of the video, add background music and season with light color correction and your video is ready.

DaVinci Resolve

Afterwords

We live in incredible times of opportunities, even 10–20 years ago only those who had startup capital or other resources that help to get into certain circles of people could achieve their desires, but now you only need an internet connection, and you don’t always need a computer. The competition in the neural network market is amazing, incredible products that allow you to create incredible things and communicate your thoughts and ideas to the world more freely, without spending a lot of time and effort on analog methods. I’m currently creating visuals for my debut short film with the help of AI, and I’m afraid to imagine how much money and effort it would take to actually shoot it, and for such enthusiasts neural networks are not just a tool but an opportunity.

After the release of Midjourney, I remember everyone shouted “designers are no longer needed” “Midjourney has taken away the work of artists and designers” and so on, but I have always been of the opinion that a good performer will remain good even with incredibly high quality AIs and will perceive it not as a competitor but just an assistant for its convenience.

[embed]The video created by the method described in this article