OpenArt Under the Microscope: Art or Algorithmic Noise?  ## Metadata - Author: [[Hard Fork AI]] - Full Title: OpenArt Under the Microscope: Art or Algorithmic Noise? - Category: #podcasts - URL: https://share.snipd.com/episode/0d3a7c74-edd0-42de-b504-9ecf3c03e750 ## Document Notes ## Highlights - **OpenArt Is Behind Viral AI Content** - OpenArt powers the viral 'brain rot' videos that flood social platforms but also serves many other creators. - The platform has millions of monthly users and nearly $20M ARR, showing broad commercial traction. Transcript: Jaeden Schafer Now, I think one thing that's interesting here, OpenArt, basically, we see these quote unquote, brain rot videos everywhere. There's like, you know, you'll see like a video or a picture of a shark that's wearing sneakers. And just I don't know, like all sorts of ridiculous things, right? So these are the classic brain rot videos, which if you're watching on YouTube or Spotify, I'm sharing my screen, you can see some examples of these videos. But otherwise, I think you get the picture. So people are making these and they're getting millions of views. I'm looking at TikTok right now, 2.2 million views, 22,000 views, 280,000 views, 33,000 views, 250,000 views, a lot of views for these quote unquote brain rot videos. Now, is this whole podcast about making brain rot videos and content? No. So this is the company behind basically most of these, most of these videos. But I think there's something much more interesting here and a tool that's much more interesting than it gets credit for. So OpenAR, oh, OpenArt was founded by two Google employees or former Google employees in 2022. So they actually have around 6 million monthly active users. So it's got a very big user base. I ([Time 0:00:59](https://share.snipd.com/snip/06aabbc3-c433-448d-81fb-57656f005647)) - **One-Click Story Demo** - Jaeden explains OpenArt's new 'one-click story' that converts text, images, or audio into a full video. - He recounts that the feature can make music videos, explainers, or short social clips from a single input. Transcript: Jaeden Schafer What they just launched that I think a lot of people are talking about right now is called a one-click story. So basically, this is the ability for you to upload a piece of text, an image, an element, some audio, and it will create a full video from it. So they're even saying like, this is great for making music videos. If you if you made a song, you could upload the song and it would create an entire music video with a whole story arc through it, making a video on here. Now, the I think one of the main models I've seen mentioned on here is VO3. This is Google's newest, most famous video model, which is really good. I've tested out at Google head green on Google campus when it first came out. It's a great model. There's a couple other players that I think they have on the platform. I think they have about 50 different AI models that they use to generate their images and video. But this is this kind of one click thing is really interesting. A lot of people are saying, oh, no, everyone's going to just use it for making more AI brain rot but I actually think there's a ton of very interesting use cases where you can get videos Made with just a image or just a prompt or just a concept and basically it will it will come out with that so right now that feature is in open beta so basically you can put a single sentence, A script, a song, it turns it into a video, anything from a, you know, lighthearted story for TikTok or some more, you know, serious content, like an explainer video, a music video for YouTube, I would imagine you could do something like, you know, an explanation of why your product is great when you upload a screenshot. ([Time 0:02:50](https://share.snipd.com/snip/a2ec62b0-8c1d-44cd-aff1-1c19735d2d0d)) - **Multiple Models Power Fast Video Generation** - OpenArt uses strong video models (e.g., Google's VO3) and about 50 different models to produce content. - Combining multiple foundational models lets the platform generate high-quality videos quickly. Transcript: Jaeden Schafer Now, the I think one of the main models I've seen mentioned on here is VO3. This is Google's newest, most famous video model, which is really good. I've tested out at Google head green on Google campus when it first came out. It's a great model. There's a couple other players that I think they have on the platform. I think they have about 50 different AI models that they use to generate their images and video. But this is this kind of one click thing is really interesting. ([Time 0:03:16](https://share.snipd.com/snip/ac54b8c0-19b2-41fc-a848-837923d6fdbe)) - **Use One-Click Stories For Cheap Ad Creative** - Use the one-click story to generate ad creatives by uploading product images and prompting a narrative. - Leverage it for solo founders or small teams to produce marketing videos without large production budgets. Transcript: Jaeden Schafer Open it, open art is basically hoping this feature is going to be used for advertising. So for making ads, and I can see this as a great use case where basically you take a picture of your product and you put a prompt in there and then you say, like, make a whole story about this Product, right? You put in like a special kind of purse from your company. And all of a sudden, it's got a whole story about why this lady loves this purse and this whole, you know, emotional story. I mean, basically, it's what ads are doing today. But they're just happening so fast. And it's interesting that they're not just integrating the execution, but the creativity of a project in advertising like that, right? Like, I think so often, we think of, okay, like, I like my backgrounds in marketing. And so like, we think of, okay, like we have this product, we want to market it. Someone's got to write a good script. They got to write a catchy thing. And then we move it over to like a protection company that's going to do all the production for it. This is like, basically, I think for solo people, for small companies, you just upload a picture and it creates the whole thing. ([Time 0:04:26](https://share.snipd.com/snip/30c2ba10-01c3-4aef-819b-1191d2518bba)) - **Templates And Storyboard Editing** - Jaeden outlines the three templates: character vlog, music video, and explainer, and shows how users can edit storyboards. - He shares that uploading a song lets the tool sync visuals to lyrical themes automatically. Transcript: Jaeden Schafer So there's three different templates you can choose from you can do a with their with their like one click story thing you can do a character vlog you do a music video or you can do an explainer So for a character vlog basically um you get upload an image of the character and then you enter a prompt um about what you want to basically happen which i think is kind of interesting. If you upload a song, it basically understands the lyrics and it creates animations that align with the song's themes. So, you know, it'd be illustrating like flowers blooming in a garden if the song is singing about that. So I think this is pretty interesting. As a side note, as someone who my wife is a musician and has done a lot of work to promote her music recently. And so she's actually experimented with doing things like running meta ads, targeting people that are interested in your style of music towards your Spotify cam, your Spotify artist Page. But making the meta ads is a lot of work to basically make all of the ad sets that you use to send people. So anyways, honestly, a tool like this that creates some of those animated graphics that you basically put clips of your song up, and then have some sort of animated graphic in the background, And then a caption on there. I think a tool like this would be really, really cool for that. So you then are able to edit individual clips by going to they have something called the editors storyboard mode, you can tweak your prompt if you want to refine your results a little Bit. ([Time 0:05:22](https://share.snipd.com/snip/877e8dd9-cb89-4b41-854c-37d6db3d2ea7)) - **Liability Gets Pushed To Foundational Models** - Platforms often push IP liability onto the foundational models rather than policing uploads themselves. - This shifts legal and moderation risk down the stack but leaves platform-level screening limited. Transcript: Jaeden Schafer So overall, I'm really excited. I think a lot of people have kind of criticized when you're doing these video generation, like the concerns around IP. I think their CEO talked about this specifically he said we try to be cautious around ip infringement when you upload some ip characters by default the models we use will reject them And it's not able to produce ip characters but sometimes it slips it's kind of interesting that a company like this doesn't have basically they they put all of the liability onto the Foundational models right they're like the models we use by default won't let you upload IP, right? Like OpenAI and everybody else. However, it's interesting because them as a platform, they don't have any way to screen or look at or stop that. Anyways, I'm not criticizing that. I just think that's an interesting thing you'll see with most of these AI companies. They'll try to push the liability out to the foundational model. Like, oh yeah, they don't accept this kind of content when it's, I don't know. It's just interesting. The buck is getting passed for sure. ([Time 0:06:53](https://share.snipd.com/snip/37c02606-7020-4508-811a-865c9f29cfa5)) - **Character Consistency Improves Immersion** - Character consistency across images and videos is a key differentiator that improves storytelling immersion. - OpenArt claims to solve this for images and is working to extend it to video for more coherent narratives. Transcript: Jaeden Schafer Okay, one thing that I do think is interesting is basically the fact that OpenAI believes something that they think really sets them apart is that they can do character consistency. This is something I remember when the first models, the first image models came out of OpenAI. My wife was like, oh my gosh, I'm going to make like these illustrated kids stories. I'm super excited. And she tried using them. And it was impossible to get a consistent character throughout any sort of basically image model. And she gave up on that idea. And we've since moved on with our lives and have other things going on. So I don't know if that particular idea is gonna get revisited anytime soon. Today, things like OpenAI, I think, and their GPT image generator is amazing because you could basically get it to generate an image of a character. You could upload a picture of that character and say, have this character doing XYZ things. So you basically can do the illustration for an entire book with a consistent character i think through chat gpt that way um open art also is able to do that uh this is what they said about It though they said a problem that a lot of ai couldn't really handle well is to have a character consistent in the same video if you don't have the same character then it's hard to get immersed In the story so this is something interesting right opening, I solved that problem for images. And it looks like the next step is to solve this for video, which is really interesting. So I'll definitely keep you up to date on what they roll out. ([Time 0:07:54](https://share.snipd.com/snip/b3dbc8dc-f416-4e4c-a3f5-e8cea67cd1cb)) - **Try OpenArt If You Produce Videos** - Try OpenArt if you create videos or images and want fast, AI-driven production tools. - Jaeden recommends testing the platform for art and video workflows though he notes it's not sponsored. Transcript: Jaeden Schafer I think this is a very interesting tool worth trying. If you're in the art space, if you create any sort of video, or even images, but really, I think this is a great video platform, go try out open art. Very interesting, not sponsored. I just think it's an interesting tool worth taking a look at. ([Time 0:09:14](https://share.snipd.com/snip/45921b1b-5a47-4be0-ae52-b62a154c70d5))