[Black hat] AI for video editing? Like ffmpeg just AI

Joined
Mar 11, 2013
Messages
5,542
Reaction score
9,755
This is for massive multiplication of videos.

I want to have AI cheats for my videos, so algorithms will think they're completely different.

Also which AI is actually for editing? All I find is video generation which is too much. It's more of a showcase what generative AI can do and not much of practical application.

I just want to add some smokes and fire to my current video like in after effects, just much more faster. Something like "noob friendly video editing AI that makes your videos like edited by an after effects pro".
 
What people are calling "AI" isn't actual intelligence, they're just pattern matching algorithms that can work on words and images. That's why they keep spewing out pictures of people with seven fingers or whatnot because no matter how many times we try and bash it into the system that humans only have five fingers the algorithm says "keep going and add another.
 
Yeah, that's just generation of something based on Reddit based dataset probabilities. It's truly artificial. As Redditors intelligence is.

I hope to see someone demystifying the whole "transformers" and LLM thing. For now they explain it like for experts calling it blackbox like it's some UFO...
 
It doesn't exist yet, but it can be made.
Here's my opinion on AI stuff (and it's not entirely subjective):
Some nerds took out the whole internet and put it in a box. In order to make sense of what's inside, there's an algorithm that looks for patterns (AI training). Now, by definition, patterns mean repetitive stuff. No wonder that when you want to extract something out of the box (they call it inference), you'll get generic stuff. Nothing special, just what was repeated most.
Of course, there are some little tricks that can improve creativity and so on, but that's what AI does, finding patterns. Try your best prompts to generate long and creative texts. Even with gpt4 you'll end up with some boring stuff if you go above certain limits. It simply can't handle big tasks because of a thing called Attention. Yup, even the big boys have this problem, no matter how good is hardware, it has limits when it comes to processing the whole internet.

What to do then? Split the task of creating long stuff in smaller chunks. Prompt the AI to create an outline/checklist for the task, then (with the help of Python or other programming language) make it respect it's initial plan (checklist, whatever). So, you would need to do things programmatically, do not expect AI to do anything good by itself.

Going back to the problem of repetitive, generic AI, that's one way to solve it: split the task (box extraction) and you'll obtain more (smaller) details (that's the advantage of having the internet in the box) . The content will now appear non repetitive and full of information. It's that creative? Not yet. But it's good enough to pass many human tests.

In regard to video editing, Ffmpeg can do so many things. It's so complex that it looks kinda like AI already. If you have the knowledge and time, you can use Python (or whatever) to combine AI with Ffmpeg. It's doable within current limits of technology and the results could be amazing. The only problem is that it takes time to make such thing. Time or .. sustained collaborative efforts. If enough people are interested, I'm sure it can be done.
 
We need to warm-up our GPUs for SORA clones.

They'll be probably biased and will ask 10 cents a video, but we will do it without them lowering price of a video to less than a broken nickel.

This is hyper-realistic. It cannot happen in real life. Normally glass would break - here it doesn't! Actually it's not working well enough. :anyway:
But it bounces on its own which is not possible.

https://cdn.openai.com/tmp/s/discussion_0.mp4

We won't need to do drugs in the future to experience DMT realms. :suspicious: Terence McKenna predicted that. He had a connection with inner mechanics of the consciousness.
 
Last edited:
I was just watching this video about that, it is on its way and this dude isn't happy about it

 
I was just watching this video about that, it is on its way and this dude isn't happy about it

Bluedrake42? This is a nice information.

I hope YouTube will collapse once we apply this and pollute their storage with whatever we want.

He speaks something about video trailers? This is lame. It's nothing I'm aiming for.

A deep learning model won't recommend those videos more than real high quality human production.

They're gonna improve their frameworks and so on with time.

- this is AI? I don't see any AI here. Maybe it's gone too far already! That'd be only good news!

Ahh

This is the content



So it sucks and he doesn't know how to use AI well. He put it in front of critically thinking audience - his viewers. And the way he's using it is weak. Mistral 7b? Text generation? For people who want something else? Pointless.

Gamers were always looking for entertainment in gameplays and so on. AI won't do it.
 
Last edited:
So it sucks and he doesn't know how to use AI well. He put it in front of critically thinking audience - his viewers. And the way he's using it is weak. Mistral 7b? Text generation? For people who want something else? Pointless.

Gamers were always looking for entertainment in gameplays and so on. AI won't do it.
It sucks now, don't forget this is the worse version you will see of this bot, give it a year and you won't see the difference anymore...
 
It sucks now, don't forget this is the worse version you will see of this bot, give it a year and you won't see the difference anymore...
But for that you'd need to have a bot that literally does thing and record that.

AI can play games I believe. But can it do it well like a pro player?

I played with hardcore bots and as an average player I'd lose. But a pro player would not.

I'm more interested in emerging abilities of LLM than content like this.

People will switch to offline with time anyways. I think so. Because this is garbage state of the internet now.

Maybe next iteration like web3 will be more organized and more standardized. It might not be such a chaos where we share everything in such large groups. It's kinda pointless.

You could literally create a different version of the internet that doesn't even operate on websites and is something completely different without all these problems.

I think Meta or some other crazy company attempted that but the point is that it cannot be controlled by a large organization which is chaotic and controlled by their deep learning models which are biased and chaotic.

I think operating on smaller, isolated units is the future.

We don't need those large graphs and challenge ourselves indefinitely with handling as hardcore challenges as possible. We just need to get rid of supremacy of corporations and stop treating them like gods who dictate what's valid and what's not.

In my opinion, one of the best social media platforms these days is Discord. It does something better I think. I think there you'll find something positive unlike in other medias where algorithms dictate rules.
 
Last edited:
What's the benefit of using ffmpeg via AI?
What's the benefit of doing research of FFmpeg via AI?

What's the benefit of training AI on FFmpeg for suggestions and exclusive tricks / combinations / ideas?

What's the benefit of using AI to make unique videos using FFmpeg?

To make something out of it, it's best to find a niche, and find what tricks and stuff is needed for a niche, then try to find a solution combining AI and FFmpeg.

---

Start looking for business solutions, read papers and books to find out. Don't use google to search for products like this. It's counter-productive. Copy-cats of the same low quality product that is an example of how not yet saturated market looks. They can't provide sh*t for business.

It's like they use this AI for consumerism purposes and the whole point of those products is to sell something we know with "AI" added to it.

These companies don't even have money to utilize AI to make to its potential.

They're probably running on some external funding and it's useless...

I start with arxiv.org and research.google.com


---

Some companies like ElevenLabs and OpenAi and Gemini make sense. Other things, I can't find anything useful in this sea of nonsense.

Adobe has also some cool stuff but it's designed for solo producers, not people who're looking for mass production.

---

I'm looking forward to create some sort of compilations of videos with music + voice I think that are fueled by a large database.

"Yea, the Almighty shall be thy defence, and thou shalt have plenty of silver."

I am the Almighty - HAHAHA

---


Start with proper research sources that are Almighty like Google or Meta. There you shall find ideas that shall result in purses full of nickels.
 
Last edited:
Damnit.

This forum is polluted with FFmpeg propaganda.

Maybe they paid someone to promote it here?

Thou shalt look for alternative solutions.

FFmpeg syntax is what? I don't know. Maybe they smoke too much grass there. Albeit I managed to understand it a little bit in 20 hours.

They're based on ISO C11 language. Shouldn't make sense to anyone outside C family languages.

Open source is a plague and it's not the best option. Do you really care about something like open-source in black hat operations or automation? :rolleyes:

It's not necessarily coded with ease of use in mind like more commercial solutions.
  • Think PlayStation > MacOs > Windows > Linux.
  • Think normal keyboard > apple keyboard.
  • Think Adobe Premiere Pro > ffmpeg.
  • Think Adobe Illustrator / Photoshop > using CSS and HTML.
  • Think manual editing > machine learning editing.
  • Think iPhone > Android.
  • Think readymade optimized headphones > audiophile headphones to be tuned > studio headphones to be tuned.
  • Think Node.js + Typescript > C++.
  • Think Onlyfans > dating.
  • Think chat-GPT > GPT API.
  • Think limited area of possibilities > infinite possibilities.
  • Think specialized software > general use software.
  • Think creating video effects in Adobe After Effects > creating video effects in FFmpeg.
Maybe you subconsciously want to impress the girls from onlyfans by coding in ffmpeg instead of commercial software? :smirk:
  • Thou shalt carefully analyze everything they subscribe to.
  • Thou shalt carefully not reject wisdom.
  • Thou shalt carefully critically think about making the very next step.
Stinky ass developer that showers once a month stack may consist of low level, open source things. But that's not my goal to become one.

A person who tries to be as cheap as possible creates a pattern that kills a lot of their chances to become successful.
A genius admires simplicity - of use too.
A pizza restaurant that obsess about perfect pizza each time never delivers one.
 
Last edited:
Damnit.

This forum is polluted with FFmpeg propaganda.

Maybe they paid someone to promote it here?

Thou shalt look for alternative solutions.

FFmpeg syntax is what? I don't know. Maybe they smoke too much grass there. Albeit I managed to understand it a little bit in 20 hours.

They're based on ISO C11 language. Shouldn't make sense to anyone outside C family languages.

Open source is a plague and it's not the best option. Do you really care about something like open-source in black hat operations or automation? :rolleyes:

It's not necessarily coded with ease of use in mind like more commercial solutions.
  • Think PlayStation > MacOs > Windows > Linux.
  • Think normal keyboard > apple keyboard.
  • Think Adobe Premiere Pro > ffmpeg.
  • Think Adobe Illustrator / Photoshop > using CSS and HTML.
  • Think manual editing > machine learning editing.
  • Think iPhone > Android.
  • Think readymade optimized headphones > audiophile headphones to be tuned > studio headphones to be tuned.
  • Think Node.js + Typescript > C++.
  • Think Onlyfans > dating.
  • Think chat-GPT > GPT API.
  • Think limited area of possibilities > infinite possibilities.
  • Think specialized software > general use software.
  • Think creating video effects in Adobe After Effects > creating video effects in FFmpeg.
Maybe you subconsciously want to impress the girls from onlyfans by coding in ffmpeg instead of commercial software? :smirk:
  • Thou shalt carefully analyze everything they subscribe to.
  • Thou shalt carefully not reject wisdom.
  • Thou shalt carefully critically think about making the very next step.
Stinky ass developer that showers once a month stack may consist of low level, open source things. But that's not my goal to become one.

A person who tries to be as cheap as possible creates a pattern that kills a lot of their chances to become successful.
A genius admires simplicity - of use too.
A pizza restaurant that obsess about perfect pizza each time never delivers one.
I didn't want to quote all of that, but anyways. Ffmpeg has the advantage of being easily programmable (think bindings and stuff) and free. You repeat this black hat mantra, "black hat this and black hat that".
Check this out:
I don't want to waste time creating videos, but I want money. So, I'm putting AI at work to make me a script, via GPT API. Then I have this Python script that I cobbled together from various (GitHub) sources. The script finds some images and joins them together (via Ffmpeg) in a shitty video. I slap a TTS voiceover (from the AI generated text) on top of the shitty video.

Baboom!

I got one shit video in less than 5 minutes by just pressing a button.
Let's go black hat. I quickly Google/GitHub another few Python lines that keeps doing the above 24/7. On top of that, it uploads the videos to YouTube, TikTok and whatever platform I want. Cost is minimal, my work is zero.
Results are as follows. My videos quality is shit, we established that already. But because it's all automated (via f**ing Ffmpeg) I create thousands. Some shit sticks to the wall. I make money.
YouTube doesn't like me polluting their platform with shit. End of black hat.


PS. I had to
 
I didn't want to quote all of that, but anyways. Ffmpeg has the advantage of being easily programmable (think bindings and stuff) and free. You repeat this black hat mantra, "black hat this and black hat that".
Check this out:
I don't want to waste time creating videos, but I want money. So, I'm putting AI at work to make me a script, via GPT API. Then I have this Python script that I cobbled together from various (GitHub) sources. The script finds some images and joins them together (via Ffmpeg) in a shitty video. I slap a TTS voiceover (from the AI generated text) on top of the shitty video.

Baboom!

I got one shit video in less than 5 minutes by just pressing a button.
Let's go black hat. I quickly Google/GitHub another few Python lines that keeps doing the above 24/7. On top of that, it uploads the videos to YouTube, TikTok and whatever platform I want. Cost is minimal, my work is zero.
Results are as follows. My videos quality is shit, we established that already. But because it's all automated (via f**ing Ffmpeg) I create thousands. Some shit sticks to the wall. I make money.
YouTube doesn't like me polluting their platform with shit. End of black hat.


PS. I had to
Black Hat originally was about greatness.

I found some products better than both FFmpeg and Adobe Premiere Pro which API is for something else and it's not simple enough.

The mantra is a classic. It's what this community is about. People tend to forget that real black hat is like a religion. It's multi step system of beliefs.

And being able to keep up and progressing, growing is one of the rules to it.

I can't see FFmpeg as business software. It's good for developers who want to create other software based on that.

But it's not gonna work for someone who doesn't study muxing, demuxin and all that complicated video terminology.

If I was just starting I would just do FFmpeg but I know it doesn't work as a business solution where you need to connect 20+ modules.

If you do 20 weak modules vs 20 ok modules, you're winning in multiple ways, where money is not the priority but greatness.

Put things into diagrams and then compare solutions. Simple process is like drawing a single diagram or just a set of instructions - it doesn't take much effort. It's not enough. There's a need t asses pros and cons, then reactions of deep learning algorithms.

Creating shitty videos, products and solutions is a recipe for shitty results - long-term wise.

If for 20 years we focus on just shitting, it's gonna take us nowhere. Year 21 we will be still shitting. It's boring as hell and depressing as you're not moving anywhere with it.

In the past I used to shit something, not put much thought into it, get some revenue, then put all my money back into stuff I didn't understand why I'm buying. Well such experiences have to influence us somehow.
 
Last edited:
Straight from gemini.google.com:
Black hat marketing is essentially the opposite of ethical marketing. It refers to using deceptive and manipulative tactics to promote a business or product online. These tactics prioritize short-term gains over long-term customer trust and can even break the law.

Here's a breakdown of the key aspects of black hat marketing:

* **Unethical practices:** Black hat techniques prioritize getting results by any means necessary, often misleading or harming users in the process.
* **Deception and manipulation:** Black hat marketers often use tactics that trick or pressure users into taking unwanted actions, like buying a product or clicking on an ad.
* **Short-term gains:** Black hat tactics may lead to a quick boost in traffic or sales, but they are unsustainable and can damage a business's reputation in the long run.

Some common black hat marketing tactics include:

* **Spam:** Sending unsolicited emails or messages to promote a product or service.
* **Keyword stuffing:** Filling web content with irrelevant keywords to try and trick search engines into ranking a website higher.
* **Cloaking:** Showing different content to search engines and users in order to manipulate search results.
* **Fake reviews:** Planting positive (or negative) reviews about a business or product to mislead consumers.

Black hat marketing can have serious consequences for businesses, including:

* **Search engine penalties:** Search engines like Google can penalize or even ban websites that use black hat SEO tactics.
* **Damaged reputation:** Consumers can quickly identify and avoid deceptive marketing practices, which can damage a business's reputation.
* **Legal trouble:** Some black hat tactics, like cookie stuffing, can be illegal and lead to fines or even lawsuits.
Not a religion, not a system of beliefs. Bending the rules would be the best definition.
 
Straight from gemini.google.com:

Not a religion, not a system of beliefs. Bending the rules would be the best definition.
Yeah, exactly. Corporations bend them how they want. Gemini just described itself and its creators. :smirk:
 
Canva no longer work's lol ?
Does that have an API?

Edit:
I found something https://chat.openai.com/g/g-alKfVrz9K-canva

It won't do anything other than giving ideas. They go to trashcan.

Is not a good business solution.

Think of things that can be integrated with Zapier and Make.com. Now that's something.

I'll read again over again https://www.canva.com/developers/

But first I'll throw up.
 
Last edited:
So I found a level of video I want to produce

I hope anyone can help me make it!

Looks better than just slides put together!

There is this company called Runway and it was hidden like a snake in the grass from me *fcking CIA*. Their whole thing is available as SaaS with UI.

Capitalism is too huge today. They'll have to regulate it soon or everyone and my dog (that I don't have) will have a company.

Edit: frankly reading this thread makes me laugh. :oops:

Some time ago we used to do our own operational systems. Today there's 10000 things on top of those systems and other systems. It's overwhelming technological market size.
 
Last edited:
Back
Top