AI Video Generation - Google Labs Veo 3 (requests)

[phoenix2020]

I have a Gemini Ultra subscription and have a few credits to burn this evening before it rolls over tomorrow with the next billing cycle. Happy to take requests and generate an 8 second clip and post it somewhere for you to get at. Be detailed (scene, atmosphere, characters, dialogue, style aka realistic or animation) and obviously no violence/sexual stuff/etc. all of which are against the terms of Gemini usage (as well as this site).

I can't generate an infinite number of these, so if there are a lot of requests I may end up picking and choosing.

You can find dozens of examples of what this is capable of on YouTube and elsewhere, just search for "Veo 3 demo"

 

[phoenix2020]

Random stream of clips that people have generated.

Flow TV

Explore Flow TV.

labs.google

 

[ASHES]

Lovely. Been talking about it here for a few days.

Please make me a Prompt Theory clip. I love those.

If you want me to be specific about the prompt, try this (originally written by me, then refined by Gemini):

Through the lens of a handheld phone, we witness a city council meeting's abrupt descent from eerie stillness to violent chaos. A lone woman, poised at the podium, articulates the principles of the Prompt Theory with conviction. The council's sneering rejection is met not with resignation, but with a groundswell of public outrage. As police attempt to seize her, the crowd transforms into a protective, vocal wall, their defiant shouts drowning out official commands. The final moments are a blur of uncontrolled pandemonium.

 

[phoenix2020]

Lovely. Been talking about it here for a few days.

Please make me a Prompt Theory clip. I love those.

If you want me to be specific about the prompt, try this (originally written by me, then refined by Gemini):

Through the lens of a handheld phone, we witness a city council meeting's abrupt descent from eerie stillness to violent chaos. A lone woman, poised at the podium, articulates the principles of the Prompt Theory with conviction. The council's sneering rejection is met not with resignation, but with a groundswell of public outrage. As police attempt to seize her, the crowd transforms into a protective, vocal wall, their defiant shouts drowning out official commands. The final moments are a blur of uncontrolled pandemonium.

Came out a bit... weird! But, this was a lot to pack into 8 seconds.

 

[ASHES]

Came out a bit... weird! But, this was a lot to pack into 8 seconds.

LOL

Clearly my prompt requires some work. Looks like a choreographed dance parody of the prompt theory. But I've also never written a video prompt before, so . Live and learn, must be more specific.

 

[phoenix2020]

LOL

Clearly my prompt requires some work. Looks like a choreographed dance parody of the prompt theory. But I've also never written a video prompt before, so . Live and learn, must be more specific.

Yeah, I've been playing with it and have learned that one must be thoughtful and precise with the ask. I've seen good results by specifying dialogue and voice characteristics, any kind of ambient sound or music, the motion in the scene, key characteristics of the scene (e.g. "crowd" --> "4 adults of mixed gender and ethnicity dressed in normal American winter attire") and so forth.

Interestingly, Veo 2 did a bit better with the same prompt - no audio though.

 

[phoenix2020]

I have a few extra credits and am again taking requests.

 

[bomberfox]

I prefer not to help art theft.

 

[ASHES]

I prefer not to help art theft.

You've made that abundantly clear in multiple AI threads. Make it your signature or something instead of making all the AI threads about your opinion.

 

[bomberfox]

You've made that abundantly clear in multiple AI threads. Make it your signature or something instead of making all the AI threads about your opinion.

I didnt make any ai thread about my opinion. Simply stated the fact that these AIs sucked up all the hard work of actual artists.

 

[ASHES]

I didnt make any ai thread about my opinion. Simply stated the fact that these AIs sucked up all the hard work of actual artists.

Yes you did. Or you've tried. We get it.
See you're doing it right now, in this very moment.

 

[bomberfox]

Yes you did. Or you've tried. We get it.

You are free to ignore me o.o.

 

[ASHES]

You are free to ignore me o.o.

No.

 

[ASHES]

You are free to ignore me o.o.

You're also free to ignore AI image generation threads without making them about your opinion you believe it is art theft. I think it would be a good topic for a philosophical or moral debate. Not one about what this is about. Or the others where it has been mentioned.

 

[bomberfox]

You're also free to ignore AI image generation threads without making them about your opinion you believe it is art theft. I think it would be a good topic for a philosophical or moral debate. Not one about what this is about. Or the others where it has been mentioned.

Unless you are going to issue moderator action lets stop cluttering the thread then. I described exactly what it is. Move along.

 

[ASHES]

Since this thread is about an opportunity for fellow forum members to interact with veo3, not about whether AI image/video generation is art theft, I will disengage now. But I will continue to challenge you whenever I see you doing it. It isn't even that I necessarily disagree with you fundamentally, it's just about sticking to the topic and having appropriate conversations at appropriate times.

 

[phoenix2020]

I prefer not to help art theft.

Thank you for offering to help out. I really appreciate it! Without people like yourself to drop terrific suggestions on threads like these, we can't push this technology forward. Here is the prompt that you inspired; I ran two inferences but only one generated audio for some reason so I'm posting just the one.

An anthropomorphized fox wearing a classic leather bomber jacket sits in front of an open laptop in a dimly lit room. His fists are clenched and resting on the table to the side of the laptop. His eyes are wide open and full of anger. Suddenly, in one very fast yet smooth motion he simultaneously stands up, slams his fists on the table and begins to yell "Art Theft! Art THEFT! ART THEFT!", his somewhat high pitched and whiny, coarse voice getting louder with each word. As this happens, objects in the room around him begin to explode spectacularly.

 

[bomberfox]

Thank you for offering to help out. I really appreciate it! Without people like yourself to drop terrific suggestions on threads like these, we can't push this technology forward. Here is the prompt that you inspired; I ran two inferences but only one generated audio for some reason so I'm posting just the one.

An anthropomorphized fox wearing a classic leather bomber jacket sits in front of an open laptop in a dimly lit room. His fists are clenched and resting on the table to the side of the laptop. His eyes are wide open and full of anger. Suddenly, in one very fast yet smooth motion he simultaneously stands up, slams his fists on the table and begins to yell "Art Theft! Art THEFT! ART THEFT!", his somewhat high pitched and whiny, coarse voice getting louder with each word. As this happens, objects in the room around him begin to explode spectacularly.

At least the thing got hands right this time. The voice is good and it has a model that actually works.

 

[phoenix2020]

At least the thing got hands right this time. The voice is good and it has a model that actually works.

I wanted a different mood, and wasn't happy with the missing audio so I decided to prompt in more aural elements to see if I could provoke the model to generate audio. Four inference passes this time. And the result? Score! Prompting an intro audio cue (classical music) does the trick. So, we have four more takes. Posting just a couple due to time.

We begin with a photorealistic scene featuring an anthropomorphized fox wearing a classic leather bomber jacket who is sitting in front of an open laptop in a dimly lit room. Soft classical music is playing in the background. His fists are clenched and resting on the table to the side of the laptop. His eyes are wide open and full of anger. The atmosphere is dark and foreboding. Then, in a sudden yet smooth motion he simultaneously stands up, slams his fists on the table and begins to yell "Art Theft! Art Theft!", his somewhat high pitched and whiny, coarse voice getting louder and angrier with every word, eventually drowning out the music, his facial expression one of anger and disgust. As this happens, objects in the room around him begin to explode spectacularly, culminating with the laptop itself exploding into fragments.

 

[bomberfox]

I wanted a different mood, and wasn't happy with the missing audio so I decided to prompt in more aural elements to see if I could provoke the model to generate audio. Four inference passes this time. And the result? Score! Prompting an intro audio cue (classical music) does the trick. So, we have four more takes. Posting just a couple due to time.

We begin with a photorealistic scene featuring an anthropomorphized fox wearing a classic leather bomber jacket who is sitting in front of an open laptop in a dimly lit room. Soft classical music is playing in the background. His fists are clenched and resting on the table to the side of the laptop. His eyes are wide open and full of anger. The atmosphere is dark and foreboding. Then, in a sudden yet smooth motion he simultaneously stands up, slams his fists on the table and begins to yell "Art Theft! Art Theft!", his somewhat high pitched and whiny, coarse voice getting louder and angrier with every word, eventually drowning out the music, his facial expression one of anger and disgust. As this happens, objects in the room around him begin to explode spectacularly, culminating with the laptop itself exploding into fragments.

the first has a better ending that expresses genuine anger while the second is better overall though the smile looks like an evil smile. The first one looks the best with the fox’s appearance.

 

[ASHES]

I wanted a different mood, and wasn't happy with the missing audio so I decided to prompt in more aural elements to see if I could provoke the model to generate audio. Four inference passes this time. And the result? Score! Prompting an intro audio cue (classical music) does the trick. So, we have four more takes. Posting just a couple due to time.

We begin with a photorealistic scene featuring an anthropomorphized fox wearing a classic leather bomber jacket who is sitting in front of an open laptop in a dimly lit room. Soft classical music is playing in the background. His fists are clenched and resting on the table to the side of the laptop. His eyes are wide open and full of anger. The atmosphere is dark and foreboding. Then, in a sudden yet smooth motion he simultaneously stands up, slams his fists on the table and begins to yell "Art Theft! Art Theft!", his somewhat high pitched and whiny, coarse voice getting louder and angrier with every word, eventually drowning out the music, his facial expression one of anger and disgust. As this happens, objects in the room around him begin to explode spectacularly, culminating with the laptop itself exploding into fragments.

Second one lands better, but the simultaneously exploding into pieces yet perfectly intact laptop opening again kicks me back out of my suspension of disbelief.