Ran Prieur🔒 SSL Verified

Table of Contents

• ↓ Authority Metrics: Score 69/100
• ↓ Global Reach of Ranprieur
• ↓ Competitor Analysis for .COM Domains
• ↓ Top Alternatives to ranprieur.com

6-Month Traffic Trend Estimate

Official Contact & Profiles

Visual representation extracted from ranprieur.com

Content Snippets from ranprieur.com

"Blessed is the lion which becomes man when consumed by man; and cursed is the man whom the lion consumes, and the lion becomes man."

"July 24. Continuing on AI, I've said this before: The value of anything that AI does, seems to be inversely proportional to the hype. So the most hyped things, like AI becoming actually intelligent, or replacing all jobs, are going to fail spectacularly. Meanwhile, AI is doing cool stuff in areas that are either too specialized, or too silly, to get hyped. An example of the former: AI is designing radio chips that humans couldn't even imagine. As an example of the latter, I want to go into more detail about my process of making videos, because I haven't heard of anyone else doing it the same way, using free images and being very picky. I just love to exercise aesthetic judgment, and this practice is independent of the source: you can do it with houses or seashells or random paint splatters. Doing it with AI is like walking through an infinite museum. Before AI, I did the same thing with human-made images, and the human ceiling is definitely higher, but it's also harder to get there. For International Dateline I scoured the Imaginary Colorscapes subreddit and used everything good that I found. For Holiday I scoured Danbooru, and if you think being a human illustrator is a badge of honor, you should have seen Danbooru before AI, it was even worse. On a scale of 1-10, most human-made images are 1-5 -- but 10s are possible, and finding a 10 among 3s is worth a thousand thumbnails. Most AI images are 4-7, and picking 7s from 6s is subtle. The less I can explain it, the more I like it. It's funny, both my dad and my grandfather loved fishing, and I never got into it, but this is not that different. I keep throwing lines into the deep and seeing what's on the other end. Getting an 8 is only a matter of patience, and 9s are possible. For prompts, I always start with lyrics from the song. If they're dense enough, I can go line by line. That's not the case with the video I just finished, Hawkwind's Motorway City, which has just one verse, one chorus, and a long outro. I'm matching sounds as much as words, but the full chorus prompted its own images surprisingly well, and the first half of the verse was perfect for a whole other part of the song. To cover the rest of the song, I had to develop new prompts, and this is a satisfying iterative process of testing different combinations of words until I find one that clicks. The first prompt I tried was "sleek cars zip along winding motorways in a future utopian city." I plugged it into a bunch of styles, and the results were lame. It turns out that AI is bad at speculative cars, because it likes to copy cars that already exist. The image above was the best of those, and it's not in the video, because the image that opens the video convinced me to redo the whole thing as a journey from the POV of the driver. That way I could focus on cityscapes and landscapes, which AI is better at. Fruitful prompts included "a rain-slicked street in a colorful futuristic city" and "zipping away at blurring speed down a psychedelic highway." Even at the sweet spot I'm only saving around one image in five. They go in a folder for further filtering, with numbers at the fronts of their filenames, so I can test the video by clicking through the folder while listening to the song. There must be an app that makes this easier, but I like low-level work, and I do a lot of fiddly changing of filenames. For this video, I had so many decent rejected images that I stacked them at the end under the prefix "99-". Then whenever I still wasn't satisfied, I played through the 99s and brought some back. One of them kept catching my eye until I decided to "recast" a whole part of the song to make it fit. You won't guess which one (3:41) because this is so subjective. It's not like Galileo looking through a telescope. I can't prove that what I'm seeing is real, but it's an empirical practice that anyone can do. I use Perchance, but there are other free generators that I haven't tried. I don't use paid generators because they tend to make images more perfect when I want them weirder. Also, cheaply made images have a much smaller footprint. I've done the math and my carbon emissions over nine months are roughly the same as driving a car ten miles. By the way, the final image contains a pretty accurate front view of a blue 1948 Dodge pickup, and I never prompted any specific vehicle, but I recognized it immediately. An old blue Dodge pickup is what the protagonist drives in the book that most influenced this video, Roger Zelazny's Roadmarks."

Key Topics & Focus Areas for ranprieur.com:

Authority Metrics: Score 69/100

Global Reach of Ranprieur

Competitor Analysis for .COM Domains