Anyone tried AI to make thumbnails?

I did get this working. It only makes those Reddit video thumbnails, but takes a video title, an image related to the title, and combines it with a border and background and stuff. Using python, pil, and some other stuff.
 
I did get this working. It only makes those Reddit video thumbnails, but takes a video title, an image related to the title, and combines it with a border and background and stuff. Using python, pil, and some other stuff.
And this is not AI
 
Wrong, OP looking for AI, just try to read a title
A program that takes a string (title) as an input and thumbnail as an output :p.

OP said he wants to put a title in and generate a thumbnail based off that. Using Rake (which is an AI Python library) for keywords you can make a python script with PIL to make a thumbnail. I was saying that's what I did. It's not technically AI but it does what op wants.

If you have a better idea please feel free to post it :)
 
OP said he wants to put a title in and generate a thumbnail based off that. Using Rake (which is an AI Python library) for keywords you can make a python script with PIL to make a thumbnail. I was saying that's what I did. It's not technically AI but it does what op wants.

If you have a better idea please feel free to post it :)
Thanks for the tips. Is keyword extraction not an AI task?
 
good thumbnails
Problem with AIs is to define "good" in this case. When do you call it a day? What is a "good" thumbnail?
Of course you could give an AI thousands of thumbnails in relation to the video title (clean your data and be careful of the quality) and train it on said dataset. Afterwards you could give the AI the title and possible an image input, after which it might edit it / include texts; but I think it's highly unlikely that this will look ANY good unless you've got a professional team covering this project.

If it works, I'll happily see it at "Two Minute Papers" the other day ;)
 
Problem with AIs is to define "good" in this case. When do you call it a day? What is a "good" thumbnail?
Of course you could give an AI thousands of thumbnails in relation to the video title (clean your data and be careful of the quality) and train it on said dataset. Afterwards you could give the AI the title and possible an image input, after which it might edit it / include texts; but I think it's highly unlikely that this will look ANY good unless you've got a professional team covering this project.

If it works, I'll happily see it at "Two Minute Papers" the other day ;)
I don't think training an AI with thumbnails and titles would do anything. I think that the AI part should be in keyword extraction from the title and image positioning. The programm would need a hard-coded style, and then it would generate thousands of thumbnails in that style. If you look at Youtube thumbnails, they are often quite repetitive, at least the Reddit TTS ones. Just a Simpsons character / famous person / person with strong emotional expression, and then some text nicely positioned and you're set.

Good thumbnail would be one that you couldn't guess was generated by a programm (but if you see 100 of them it's obvious)
 
I don't think training an AI with thumbnails and titles would do anything. I think that the AI part should be in keyword extraction from the title and image positioning. The programm would need a hard-coded style, and then it would generate thousands of thumbnails in that style. If you look at Youtube thumbnails, they are often quite repetitive, at least the Reddit TTS ones. Just a Simpsons character / famous person / person with strong emotional expression, and then some text nicely positioned and you're set.

Good thumbnail would be one that you couldn't guess was generated by a programm (but if you see 100 of them it's obvious)
Besides watching some videos in my spare time, I barely got any AI expertise, hence I withdraw my previous statement. Your approach seems well-founded and logical. :)
 
Back
Top