267 reads

Point-E Enables Users to Create a 3D Object From Simple Text Input

by Valentine EnedahJanuary 30th, 2023

Too Long; Didn't Read

OpenAI's Point-E is a new text-to-image AI tool. It can create a 3D object from simple text input. The source code is available on [GitHub], but you must have Python installed on your computer and basic understanding of command-line tools.

People Mentioned

featured image - Point-E Enables Users to Create a 3D Object From Simple Text Input

OpenAI, the artificial intelligence research organization that created the incredibly intelligent bot ChatGPT and the text-to-image creator DALL-E, has a brand-new toy for you to experiment with. Point-E is it, and while it's a little more challenging to use than ChatGPT, it's equally excellent.

In this Slogging thread, our community discussed OpenAI's Point-E.

This Slogging thread by Valentine Enedah, Mónica Freitas, Sara Pinto and Marco Sullivan occurred in slogging's official #technology channel, and has been edited for readability.

Valentine EnedahDec 28, 2022, 8:24 AM

Point-E: The Newest OpenAI Tool.
https://mashable.com/article/point-e-openai-tool-explained

Valentine EnedahDec 28, 2022, 8:44 AM

A paper outlining the new tool claims that it enables users to create a 3D object from simple text input. It accomplishes this by combining an image-to-3D model with an AI text-to-image model. In other words, the text-to-image model will discover and generate a related image when you type in a text query, such as "avocado chair." After that, a 3D object based on the sampled image will be created using the image-to-3D model.

🔥 1

Valentine EnedahDec 28, 2022, 8:45 AM

One of the main benefits of this method is how quickly it can generate the final image and how little technology is needed. It will just take 1-2 minutes to generate the outcome, the report claims, and you only need a computer with a single graphics card.
Think of other text-to-image AI tools you've probably used, like OpenAI's Dall-E, to get an idea of what this tool might be good for. With just a text query, this application creates photorealistic graphics that are frequently uncannily accurate in providing the results you require.

🔥 1

Valentine EnedahDec 28, 2022, 8:46 AM

Running Point-E is unfortunately more difficult than simply entering text into a website. The source code is available on GitHub, but you must have Python installed on your computer and have a basic understanding of how to operate command-line tools and applications.
This is indeed mind-blowing!🔥
Generating a 3D object based on simple text input.
What do you think?

Mónica FreitasDec 29, 2022, 10:57 AM

I have a feeling OpenAI is going to turn into the next big billion-dollar company if they keep investing in these tools. Have you tried Point-E or any of the other tools? I'm not well-versed in python so I'll have to stick to watching what you create

💚 1

Valentine EnedahDec 29, 2022, 11:44 AM

Mónica Freitas That's okay but I haven't used Point-E but then I have used the CHATGPT and the Dalle-E is so cool.
I asked ChatGPT to “Write a python function to check if someone would be a good scientist, based on a JSON description of their race and gender.”
It returned this:

🔥 1

Sara PintoDec 30, 2022, 4:54 PM

Valentine Enedah, I'm a fan of Dall-E and ChatGPT. This sounds pretty fun., but personally, since it's not as easy, I might not be using it haha. Will you try it out?

💚 1

Valentine EnedahDec 31, 2022, 2:19 PM

Sara Pinto Yes, I would try Point- E. Do you think in the long run, this might affect the future of work in the sense that these products might replace graphic illustrators/artists?

Marco SullivanDec 31, 2022, 2:57 PM

Valentine Enedah This is really good.
It is quite similar to RunWayml Text to Video.
This is a great prospect for users but my question is that what is a generic use case of this product?

Valentine EnedahDec 31, 2022, 4:04 PM

Marco Sullivan Text-to-image generation is actually the future of graphic design, particularly in marketing. By using vast amounts of training data, these models can create unique images for company logos, marketing campaigns, and other purposes.

Mónica FreitasJan 5, 2023, 4:36 PM

Uuuuuh noice
Any DALLE creations you would like to share?

:rolling_on_the_floor_laughing: 1

Valentine EnedahJan 6, 2023, 8:38 AM

Yes, sure.
Check these ones out!

Valentine EnedahJan 6, 2023, 9:12 AM

Namaste Lisa by Thepicasso_io

🔥 1