Live data from Hacker News

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

github.com

211–220 of 301 posts

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

#211

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.

> I know this tech is terrifying to actual 3D artist who don’t want be a “prompt engineer” My predication: in the next 5~10 years, most artists won't be "prompt engineers". Instead they'll focus on fix small details on AI-generated art. It's still kinda sad tho, because it's usually the most tedious and boring part of the process. Now AI is taking the fun part and leaving the unfun part to humans.

Yeah, but it's kinda like making business software. Most working people need to do boring work to sustain their lifestyles.

I hope that these task-specific implementations of AI can reduce the tedium in these fields, like the way PCs did. Certainly, this advancement is leading to the ability for practically anyone to program a computer, in the general sense. Things will shift, but there will be opportunities to exploit those abilities for personal gain.

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

#212

Earlier quoted context omitted.

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.

If the machine is that clever I would worry about other things. You definitely wouldn't need to ask anything of it anymore.

Besides maybe.. "make all these entrepeneurial types obsolete." Poof!

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

#213

Until We started to see LLMs, and the tools that can be created with them, I doubted the possibility of Star Trek's Voice command system. Asking for the computer to clarify some concept, or filter and reduce data sets based on arbitrary data was pure science fiction. Seeing something like this makes me think that the arbitrary holodeck commands "Paris, 1950's, rainy afternoon" is suddenly not a challenging part of th…

The image AIs are much more capable too, and I find it interesting that every technically inclined person used to always make fun of those “enhance” moments in TV shows and movies where they would zoom into some area of a photograph or security footage. The fact that now this is actually possible (to some extent at least) is pretty wild.

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

#214
post #196

Earlier quoted context omitted.

From authors website: "I graduated high school in May of 2022..". Elsewhere, he writes he placed first at a university level drone programming challenge at IIT-Bombay Techfest. Very impressive.

The JEE exam in India is very advanced, I doubt most Europeans or US students could pass it, even after a year or two at university. So those who enter IIT are already quite good very early.

India's high-stakes exams are also plagued by cheating.[1]

So just because someone ostensibly "passed" an exam doesn't mean they actually did.

Of course, India's education system is far from the only one where cheating is rampant, which is just one reason to look on credentials and accolades everywhere with a skeptical eye.

[1] - https://www.economist.com/asia/2022/05/26/indias-exams-are-p...

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

#215
post #196

Earlier quoted context omitted.

The JEE exam in India is very advanced, I doubt most Europeans or US students could pass it, even after a year or two at university. So those who enter IIT are already quite good very early.

IIT colleges in themselves are not even in top 100 Universities of the world. If you train a lion in a circus, it won't be the king of the jungle. So if we were to believe your statement: > I doubt most Europeans or US students could pass it, even after a year or two at university. than this allegedly miraclous students are wasting there talents studying in IITs!

USA and European university rankings have a strong bias against universities in the Global South (especially India), in part because of the stigma against Indian programmers and in part because they're generally wanting to promote US and European universities as the global leaders. If you're going to disparage a university based on nothing but ranking lists run by USA and Europeans, you're missing out.

IIT students are amazing, i've worked with them. Far more creative and competent programmers than the average Ivy League grad in my cohort.

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

#216

Earlier quoted context omitted.

Junior software developers will still have careers, they'll just have to learn a slightly different skillset that involves more reviewing of code, like a senior developer already does. Before sometime hits merge, they need to understand if the code meets all requirements, and it doesn't really matter who or what wrote the code in the PR.

They will be reviewing the unit test written by TestGPT for the code written by CodeGPT.

Until ReviewGPT takes over.

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

#217
post #59
post #52

Earlier quoted context omitted.

While I agree, I don't think this is what Dijkstra was speaking about. A lot of the "AI interfaces" are just underpinning the real technical representation. For example, if you say "Which customers recently bought product #721?" this might generate a SQL query like "select * from customers where ...". The interface is not the ChatGPT text box; it's SQL. The ChatGPT text box is just an assistant to help you do the cor…

Your example isn't very comparible to tasks involved in creating, texturing, lighting, animating and rendering 3D models using spline and mesh topologies.

All of that seems very plausible, right?

"Make a juvenile elephant."

"Make his ears comically large."

"Bigger."

"Give him a little floppy hat."

I don't think it's out of the question for these kinds of commands to result in the correct outcomes. Now, maybe I can adjust the ear size more precisely with my mouse, but it probably saved me a bunch of work.

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

#218

"The generated code might not always be correct. In that case, run it again lmao" is the best documentation I've read all week.

We also use GPT to perform actions in the software I build at work and we hit the same issue of inconsistency which lead me down a long rabbit hole to see if I could force an LLM to only emit grammatically correct output that follows a bespoke DSL (a DSL ideally safer and more precise than just eval'ing random AI-produced Python).

I just finished writing up a long post [1] that describes how this can work on local models. It's a bit tricky to do via API efficiently, but hopefully OpenAI will give us the primitives one day to appropriately steer these models to do the right things (tm).

[1] https://github.com/newhouseb/clownfish

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

#219
post #96

Earlier quoted context omitted.

People in the textile industry were well paid and given a place to live at one point in history. Then the auto loom showed up and they were kicked to the street to starve. I wouldn't want to try to live in any major US city on a blue collar salary.

Maybe one day AI will allow completely unskilled laypeople, ignorant of software, to build and deploy software end to end with no involvement from any software engineers, testers, SREs, anything. That would be the power loom of our time - the power loom allowed unskilled workers to completely replace skilled hand loom weavers with no drop in quality of output. When that's possible, the world will be very different. R…

> When that's possible, the world will be very different. Right at this moment, AI is still useless for unskilled workers trying to write software, it's just a productivity multiplier for skilled engineers.

Depends on the software, and how much mediocrity the end user is willing to put up with.

A trivial prompt can spit out a web page with functioning JavaScript for a mediocre-but-playable version of Pong.

This may not be of interest to us, but our standards are not necessarily shared by normal people: in the wild, I've seen websites where the thumbnails were all loaded as full-sized images and merely displayed smaller, bottles on supermarket shelves whose labels had easily visible pixelation and JPEG artefacts.

Infamously, there's a lot of stuff done in Excel that really shouldn't be. Some genes had to be renamed because scientists kept using Excel, and Excel kept interpreting the gene's names as dates.

I get SMSes whose sender ID has obviously involved someone somewhere trying to record phone numbers as floats.

Even in places with high standards, the UI of the Calculator app on iOS still gets confused if I tap buttons too fast (before animations finish playing?).

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

#220
post #61

Until We started to see LLMs, and the tools that can be created with them, I doubted the possibility of Star Trek's Voice command system. Asking for the computer to clarify some concept, or filter and reduce data sets based on arbitrary data was pure science fiction. Seeing something like this makes me think that the arbitrary holodeck commands "Paris, 1950's, rainy afternoon" is suddenly not a challenging part of th…

Here's an image result from MidJourney for "Paris, 1950s, rainy afternoon". No additional editing, and I intentionally avoided adding any text to the prompt beyond your own. https://i.imgur.com/IYuh29H.png Not perfect but man, we're getting pretty close.

It's a nice image, but I find it's easy to spot IA generated images when trying to make images of very specific existing hardware. Here you can see all the cars are generic with 50's design look, none are models that existed. Try to ask an IA to draw you a Boeing 747-400 for example, see what I mean. Btw, Have you noticed all the Youtube vids thumbnails made by IA now ? Easy to spot.
Post reply on HN