Live data from Hacker News

BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

github.com

131–140 of 301 posts

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#131
post #67

Earlier quoted context omitted.

Sure but GPT certainly cannot actually do this. It can iterate, but in GPT-world that means "give me another response" not "learn from what i'm saying, and give me another response" The issue is, sometimes those two things seem to be the same thing!

Eh? GPT definitely uses the past comments as context, so as you clarify, its responses get much closer to the target. If you haven't tried GPT-4, you absolutely should before having strong opinions about it.

parent is only half wrong; given a sufficiently long iteration process any of the chatGPT versions would surely start losing coherency regarding far past requests -- this may be less evident when used within a well confined and (somewhat) easily self-referenced system (like Blender, for example) -- but it's especially evident when trying to prompt GPT to write a fiction story from basics or otherwise work entirely on its own without a place to store outputs and then refer back.

tl;dr: it's easier to tell chatgpt to "Rewrite this story: " and then feed back previous outputs when writing a story than it is to get to an acceptable output from massively detailed prompts or long chains of iteration; this trait has far-reaching consequence rather than just writing fiction.

I do understand , however, that 'long-term memory' is a very active point of discussion and development.

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#132

Famous quote: "Wouldn't it be nice if our machines were smart enough to allow programming in natural language?". Well, natural languages are most suitable for their original purposes, viz. to be ambiguous in, to tell jokes in and to make love in, but most unsuitable for any form of even mildly sophisticated precision. And if you don't believe that, either try to read a modern legal document and you will immediately s…

Meanwhile, Star Trek has been using natural language for a few decades now, and it works pretty well.

Considering they also predicted iPads, we might want to take it at face value.

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#133

I know this tech is terrifying to actual 3D artist who don’t want be a “prompt engineer”, but as someone who has never used Blender, I think its cool that I can create something using tools like this and use them in my projects, ex background animation on a website hero section.

Most people aren't worried about being a prompt engineer-- they're worried about being a Doordash driver if it turns out that one prompt engineer can replace many professional, well-paid people in their industry. That's pretty rational.

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#134
post #57

Can we get someone in the comments who has used it and can speak thoughtfully on how well it works?

I have Blender installed, but haven't taken the time to figure out how to use it. This is how BlenderGPT went for me:

I started with a scene that had a camera and a cube. I told it to make the cube red. Result: "Error executing generated code.."

I told it to delete the cube. It failed again.

I deleted the cube and told it to make a cube at origin. It made a cube.

I told it to make the cube bigger. It made the cube bigger.

I told it to make the cube red. Fail.

I told it to make an animation of a spinning cube. Fail.

I told it to make a car. It made a rectangular cube with 4 cylinders, somewhat resembling a toy car made of wood blocks, laying on its side.

I told it to turn the car upright and make it more aerodynamic. It failed and I returned here to HN to see if anyone else had some advice.

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#135

Famous quote: "Wouldn't it be nice if our machines were smart enough to allow programming in natural language?". Well, natural languages are most suitable for their original purposes, viz. to be ambiguous in, to tell jokes in and to make love in, but most unsuitable for any form of even mildly sophisticated precision. And if you don't believe that, either try to read a modern legal document and you will immediately s…

"Wouldn't it be nice?" "But it isn't so."

An is-versus-ought style mistake.

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#136

Famous quote: "Wouldn't it be nice if our machines were smart enough to allow programming in natural language?". Well, natural languages are most suitable for their original purposes, viz. to be ambiguous in, to tell jokes in and to make love in, but most unsuitable for any form of even mildly sophisticated precision. And if you don't believe that, either try to read a modern legal document and you will immediately s…

Meanwhile, Star Trek has been using natural language for a few decades now, and it works pretty well. Considering they also predicted iPads, we might want to take it at face value.

star trek is also fictional where things happen as the writers wish it to… the real world is not subject to such restrictions

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#137
post #95

Earlier quoted context omitted.

These systems sure hallucinate text.

Yes, because models like midjourny tiny compared to LLMs like GPT. I'm pretty sure there's a good hackernews discussion on this that occurred recently, but with all the AI talk I can't find it. But really we need a lot less information to make a reasonable city, then the amount of information we need to make billboards and signs make sense. I don't think Midjourny wants to pay 10+ million dollars to have their model…

if you ever find it i’d love to read!

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#138

Earlier quoted context omitted.

In (language model-backed) natural language interfaces the loss of precision can be made up via iteration. If it were about getting it right on the first try this would be a dead end but there's no need for that restriction.

What makes us think that encoding all functionality using natural language is somehow more compact than using code? Sure, describing a website with a button is easy, but I don’t quite see how you would describe Kubernetes or a device driver without writing a book in “legalese”. Or try formulating a math proof with natural language.

Who said you need to need to describe Kubernetes? "Make my programs work at any scale". Boom

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#139

Earlier quoted context omitted.

In (language model-backed) natural language interfaces the loss of precision can be made up via iteration. If it were about getting it right on the first try this would be a dead end but there's no need for that restriction.

What makes us think that encoding all functionality using natural language is somehow more compact than using code? Sure, describing a website with a button is easy, but I don’t quite see how you would describe Kubernetes or a device driver without writing a book in “legalese”. Or try formulating a math proof with natural language.

The idea is you wouldn’t formulate the proof, you’d describe what want proven and it would generate the proof.

Re: BlenderGPT: Use commands in English to control Blender with OpenAI's GPT-4

#140

Famous quote: "Wouldn't it be nice if our machines were smart enough to allow programming in natural language?". Well, natural languages are most suitable for their original purposes, viz. to be ambiguous in, to tell jokes in and to make love in, but most unsuitable for any form of even mildly sophisticated precision. And if you don't believe that, either try to read a modern legal document and you will immediately s…

If you want something unambiguous in a single pass, sure. But interaction with some back and forth can dig into the ambiguities that matter quickly, ignoring those that don’t. Dialog is a whole separate model. So maybe natural language is a terrible specification language but a great interaction language.
Post reply on HN