Quicksand Fans

I’ve been at work generating Sora clips. I wish to share my best ones with you all. Go nuts.

https://drive.google.com/drive/folders/ ... SCiPs8m5ui

I will add new ones every few days if they’re good.

This is a gold mine of sinky goodness. Absolutely phenomenal. It’s like these poor girls in their hot little outfits seemingly respawn over and over into the next clip, only to find themselves yet again at the mercy of that inescapable muck.

Thank you for your top notch contribution. I can’t wait to see more!

A few quick needle-drops tells me that I've just planned tonight's entertainment!

Lovely work!!

How easy was it to prompt in Sora 2? Most AIs I find are an absolute nightmare to prompt properly for quicksand stuff, they don't seem to properly understand the concept.

doltonutani wrote:How easy was it to prompt in Sora 2? Most AIs I find are an absolute nightmare to prompt properly for quicksand stuff, they don't seem to properly understand the concept.

You definitely have to describe it as thick, tar like, peat bog, a lot of ai content posters (you know who you are) generate images or videos that just look like water and call it a day. It’s also able to understand depths relatively well, so you need to tell it phrases like, “her legs are shown sinking until she is sunk down to her hips”, or “another angle shows her sunk to her breasts and she sinks further down to her neck”.

AI is not some magic tool. You need to write prompts as if you’re writing a movie scene. Someone needs to be able to recreate the image you’re describing. Same with AI.

These are amazing! I would love it if you could make them reach while they're going under a few times. Want to see them struggle Great work!

SecretDude77 wrote:AI is not some magic tool. You need to write prompts as if you’re writing a movie scene. Someone needs to be able to recreate the image you’re describing. Same with AI.

Intresting. It seems like the more advanced or ‘intelligent’ these generators get the more dynamic the prompts need to be. It’s not as simple as dumping key words and half sentences into a box anymore. It’s more like using yellow and blue to make green. That will definitely separate the good quality productions from ‘amateur hour’ filler.

I've been having a lot of success with Sora 2 providing backgrounds of real life sinking locations I took and then describing something

SecretDude77 wrote:

doltonutani wrote:How easy was it to prompt in Sora 2? Most AIs I find are an absolute nightmare to prompt properly for quicksand stuff, they don't seem to properly understand the concept.

You definitely have to describe it as thick, tar like, peat bog, a lot of ai content posters (you know who you are) generate images or videos that just look like water and call it a day. It’s also able to understand depths relatively well, so you need to tell it phrases like, “her legs are shown sinking until she is sunk down to her hips”, or “another angle shows her sunk to her breasts and she sinks further down to her neck”.

AI is not some magic tool. You need to write prompts as if you’re writing a movie scene. Someone needs to be able to recreate the image you’re describing. Same with AI.

Grok imagine is definitely the worst for the water quicksand stuff. I've tried all sorts of descriptions and it'll just generate water every time. Kling is a bit better but obviously expensive and can be a bit hit and miss with whether it'll full grasp the prompt or do something a bit mad. Sora looks like it might be the best so far.

It's amaing how far the AI has come with video creation. You even got some of them to tilt their heads up at the end. The AI seems to be learning? Ah well I look forward to where this ability is with the release of 18+ chat/AI in a few months!