Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

blog.google

Post reply on HN
Print view View original

321–330 of 570 posts

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#321

Post by jryle70 »

onemoresoop wrote: post #223
hollowturtle wrote: post #97

Earlier quoted context omitted.

Recordings of who? Not only sad but a disaster, I'm sorry but anyone that ever tried to play an instrument seriously knows how much human touch/imperfections come into play, otherwise you're just an anonymous guy playing in a cover band(like the ai will do)

A tsunami of effortless content is upon us and that will change many things including tastes, probably for the worse. People not having to learn instruments because same can be done with a prompt is a tragic loss for humanity, not because human work is better but because of the lost experience and joy of learning, connection with the self and others and so many other things.

> because of the lost experience and joy of learning, connection with the self and others and so many other things.

Nothing prevents human from continue doing just that, precisely because it brings joy and satisfaction. Painting, photography classes are still popular, if not more, in the age of digital photography.

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#322

Post by bigmadshoe »

vunderba wrote: post #319
bigmadshoe wrote: post #243

Earlier quoted context omitted.

I'm curious why you decide to declare victory after one successful attempt, but try many times for unsuccessful models. Are you trying to measure whether a model _can_ get it right, or whether it frequently _does_ get it right? I feel like success rate is a better metric here, or at least a fixed number of trials with some success rate threshold to determine model success.

It's hard to nail down a good objective metric on something that is always going to be marginally qualitative in nature but it's a good call out - I should probably add a FAQ to the site. To clarify this test is purely a PASS/FAIL - unsuccessful means that the model NEVER managed to generate an image adhering to the prompt. So as an example, Midjourney 7 did not manage to generate the correct vertical stack of transl…

Makes sense. It just set off some statistical alarm bells in my head to see a model marked as passing with 1 trial, and some models marked as failing with 5. What if the probability of success is 5% for both models? How confident are we that our grading of the models is correct? It's an interesting problem.

Cool site btw! Thanks for sharing.

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#324

Post by goosejuice »

rcarr wrote: post #238
julianpye wrote: post #126

An indie film with poor production values, even bad acting can grip you, make you laugh and make you cry. The consistency of quality is key - even if it is poor. The directing is the red thread throughout the scenes. Anything with different quality levels interrupts your flow and breaks your experience. The problem with AI video content at this stage is that the clips are very good 'in themselves', just as LLM result…

You might want to look up NeuralViz on YouTube. 180k subscribers. They've been building out an entire cinematic universe using AI video tools. And it's by far the funniest show I've watched in years. So the claim that "let you engage beyond an individual clip will not be possible for a long time" isn't true. People are already doing it. https://www.youtube.com/@NeuralViz

This is news is hilarious.

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#325

Post by stevenhuang »

rcarr wrote: post #238
julianpye wrote: post #126

An indie film with poor production values, even bad acting can grip you, make you laugh and make you cry. The consistency of quality is key - even if it is poor. The directing is the red thread throughout the scenes. Anything with different quality levels interrupts your flow and breaks your experience. The problem with AI video content at this stage is that the clips are very good 'in themselves', just as LLM result…

You might want to look up NeuralViz on YouTube. 180k subscribers. They've been building out an entire cinematic universe using AI video tools. And it's by far the funniest show I've watched in years. So the claim that "let you engage beyond an individual clip will not be possible for a long time" isn't true. People are already doing it. https://www.youtube.com/@NeuralViz

https://www.youtube.com/watch?v=3XkrhhsV6zg

> You're not the monolith of me!

These other universe memes are too good.

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#326

Post by Jordan-117 »

MattGrommes wrote: post #160
jampa wrote: post #145

Earlier quoted context omitted.

The first two paragraphs of your argument could be used to discuss whether Photography (Camera is doing most of the work) or Digital Drawing (Photoshop is doing most of the work) are art. Both things which were dismissed as not art at first but are widely accepted as an art medium nowadays.

I see this comparison to a camera a lot but I don't think it works (not that you're saying this, I'm just contributing). I'm not an expert but to me the camera is doing very little of the work involved in taking an artistic picture. The photographer chooses which camera to use to get a certain effect, which lenses, the framing, etc. All the camera is doing recording the output of what the person is specifying.

I think there's a sliding scale in both cases. Vanilla prompting something like DALL-E 3 and uncritically accepting what it spits out is the AI equivalent of dime-a-dozen smartphone snapshots of the Eiffel Tower or an ocean sunset. But like your description of professional photography, there are more intricate AI approaches where an expert user can carefully select a model, a fine-tune/LORA, adjust the temperature or seed, inpaint or layer different elements, and of course have the artistic vision to describe something interesting in the first place.

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#327

Post by brunoborges »

_ncuy wrote: post #311

Google hit the jackpot with their acquisition of YouTube and it's now paying dividend. YouTube is the largest single source of data and traffic on the Internet, and it's still growing fast. I think this data will prove incredibly important to robotics as well. It's a shame they sold Boston Dynamics in one of their dumbest ever moves because of bad PR.

"Growing fast" is questionable these days.

There is an ever growing percentage of new AI-generated videos among every set of daily uploads.

How long until more than half of uploads in a day are AI-generated?

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#328

Post by bamboozled »

brunoborges wrote: post #327
_ncuy wrote: post #311

Google hit the jackpot with their acquisition of YouTube and it's now paying dividend. YouTube is the largest single source of data and traffic on the Internet, and it's still growing fast. I think this data will prove incredibly important to robotics as well. It's a shame they sold Boston Dynamics in one of their dumbest ever moves because of bad PR.

"Growing fast" is questionable these days. There is an ever growing percentage of new AI-generated videos among every set of daily uploads. How long until more than half of uploads in a day are AI-generated?

In the future, a new intelligent species will roam the earth, they will ask, "why did their civilization fall?" The answer? These homo-sapiens strip mined the Earth and exacerbated climate change to generate enough power to make amusing cat videos...

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#329

Post by autobodie »

sandspar wrote: post #251
pedalpete wrote: post #224

Earlier quoted context omitted.

I think you're contradicting your own argument. Painting didn't die from photography. Photography increased the abstract and more creative aspects of painting and created a new style because photography removed much of the need to capture realism. Though, I am still entranced by realist painting style myself, it is serving different purpose than capturing a moment.

I'm not an expert so I may be wrong about all this. But my impression is that Pictorialist photography aped painting for 50 years. Photography only came into its own as a "photography native" art form with Stieglitz and people like that around ~1905. By that time, non-representational painting styles like Cubism had already sucked the remaining juice out of painting, with Duchamp's 1917 urinal perhaps deserving credi…

Sounds right, except calling Dali or Warhol "middlebrow." That's just weird.

Re: Veo 3 and Imagen 4, and a new tool for filmmaking called Flow

#330

Post by sethaurus »

ipnon wrote: post #310
bigmadshoe wrote: post #243

Earlier quoted context omitted.

I'm curious why you decide to declare victory after one successful attempt, but try many times for unsuccessful models. Are you trying to measure whether a model _can_ get it right, or whether it frequently _does_ get it right? I feel like success rate is a better metric here, or at least a fixed number of trials with some success rate threshold to determine model success.

It is indicative of marginal improvements instead of new breakthroughs. iPhone 1 was a paradigm shift. iPhone 10 was essentially iPhone 9 with tweaks. As an AI optimist I would be disappointed to find we are already seeing diminishing returns on R&D.

At the risk of unbearable pedantry, there's never been an iPhone 9. (There was never a 2 either; there was kind of a 3, although it was really called the 3G.)
Post reply on HN