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

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

github.com

Post reply on HN
Print view View original

241–250 of 301 posts

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

#241

Post by quadcore »

TeMPOraL wrote: post #28
warent wrote: post #9

Earlier quoted context omitted.

I don't think this is "terrifying" to 3D artists any more than github copilot is terrying to us haha, I'm not sure why we engineers have this tendency of imagining every other industry as being populated by superstitious peasants from the dark ages who fear new tools. In my experience the vast majority of people are excited about GPT including artists.

I don't know, maybe I am one of those "superstitious peasants from the dark ages who fear new tools", but I'm increasingly terrified of GPT-4 and future iterations of it as applied to the industry I work with (i.e. software). It does seem to threaten to suddenly eliminate all the interesting parts of the job, a good chunk of openings, and to significantly reduce salaries for the openings that remain - all at the same…

Like Covid, it will change the world very fast. Talk to your representative about it.

As an example, I wont code at work anymore, I chatgpt all day. Just like that, overnight. And my productivity went 10x - though i was already 10x more productive than you.> What should have taken 1 week takes 1 day, sometimes less (it's new software im working on).

I would only code if I was doing great software, that is, for myself, we dont do that at work.

It's gonna build tension silently and then release: capital will be massively reallocated - that moment will be tsunami. Talk to your representative.

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

#242

Post by kranke155 »

TeMPOraL wrote: post #28
warent wrote: post #9

Earlier quoted context omitted.

I don't think this is "terrifying" to 3D artists any more than github copilot is terrying to us haha, I'm not sure why we engineers have this tendency of imagining every other industry as being populated by superstitious peasants from the dark ages who fear new tools. In my experience the vast majority of people are excited about GPT including artists.

I don't know, maybe I am one of those "superstitious peasants from the dark ages who fear new tools", but I'm increasingly terrified of GPT-4 and future iterations of it as applied to the industry I work with (i.e. software). It does seem to threaten to suddenly eliminate all the interesting parts of the job, a good chunk of openings, and to significantly reduce salaries for the openings that remain - all at the same…

You are absolutely right. The myth that automation does not replace Jobs is just that, a myth.

Huge numbers were left unemployed with industrial automation in the US and left unemployable. The technical term is structural unemployment. All it means is you cant retrain 10,000 factory workers to be front end developers, and that even If they find a job its often not as well paid.

The two greatest myths of modern capitalism are that free markets are good for everyone (they're not) and that automation doesnt lead to unemployment. Any reasonable assessment of the data will show both of these to be clearly false.

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

#243

Post by DrewADesign »

creativenolo wrote: post #172
DrewADesign wrote: post #126

Earlier quoted context omitted.

There's no mathematical proof for the correctness of graphic design, either, but that won't stop cheap AI-generated garbage from taking over the role of making commodity images and putting a lot of people out of work.

I'm curious about graphic design and if AI can do it to a passable standard. Generating fiction photographs, paintings, illustrations seems more flexible than constrained and balanced proper graphic design. I am sure AI can mash together templates and icons. Less sure about producing a solution to a client brief with a timely and timeless design.

I do freelance Graphic design. The whole conceptual thinking/visual communication aspect of it seems unlikely to be touched by these tools any time soon... But the commodity work that puts food on a lot of people's tables is likely toast. It's not like it will replace graphic designers, it just takes over the everyday jobs that drive most of the demand for their services and we all know what a drastically reduced demand does to a market.

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

#244

Post by DrewADesign »

anonzzzies wrote: post #201
DrewADesign wrote: post #133

Earlier quoted context omitted.

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.

There is not much to prompt engineer; the obvious trivial thing that is already happening and will make the glorious job of prompt ‘engineer’ completely redundant is AI doing the ‘engineering’ for you. One person (or AI in the future; this is already possible now for humans and gpt though) that the AI has been trained to ‘know’ and work with will blurb things into a mic and the AI will instruct millions of brains to…

I think you drastically underestimate the intellectual component of other people's work behind the tools they use to perform it.

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

#245

Post by Kaijo »

dagw wrote: post #229
waynenilsen wrote: post #226

Earlier quoted context omitted.

I think real CSI "enhance" is coming also although i'm sure all kinds of details will be imagined and interpolated

Yea, that is the real worry with this technology. You'll be able to zoom in on a blurry image of speeding car taken at night and get a perfectly clear license plate, but there is no way to guarantee that it is the correct license plate that was actually photographed. It'll just be the most 'probable' license plate based on some inscrutable probability distribution that is locked inside black box that even the develop…

Have you seen the leading upscalers? Things like Topaz Gigapixel are doing basically this.

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

#246

Post by chrisjc »

cube2222 wrote: post #233
iamflimflam1 wrote: post #183

I see a lot of people asking how does this work. The method he is using is one shot learning. He has a prompt and an example of what the interaction should look like. You can see the prompt here: https://github.com/gd3kr/BlenderGPT/blob/main/__init__.py It is really easy to build this kind of thing - I've got a very simple command line chatbot that should be very understandable and you can easily play with the prompt…

Did something similar as a natural language interface to DuckDB this weekend, and it's surprising how nice it works[0]. That said, through my testing I found it tended to trail off during longer exchanges if I only added the guidelines to the system prompt, so I ended up adding them to the first user message + adding a small reminder at the end of each further user message. [0]: https://github.com/cube2222/duckgpt

This is brilliant! Nice job!

Why didn't you do the same for OctoSQL? Are you considering DuckDB for your compute engine for OctoSQL?

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

#247

Post by csmpltn »

aiappreciator wrote: post #25
ChancyChance wrote: post #19

I'm glad it is not just me who thinks I need to "thank" chatGPT or encourage it, e.g.: "Great, now do .... " "Great, now do ... ".

Its not 'thanking', its positive signal that GPT's previous outputs were correct, so it should continue doing whatever it was doing. If you say no/bad etc, then GPT will try other approaches.

> "its positive signal that GPT's previous outputs were correct"

Is that somehow baked into the algorithms?

Are positive words of encouragement interpreted as "positive signals" by the inference pipeline? Or do they somehow influence the attention mechanism?

Because otherwise, you're just rationalizing completely random and unpredicted behavior.

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

#248

Post by csmpltn »

EamonnMR wrote: post #31

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

Why aren't they collecting feedback signals from people on the results produced by the model? "This was correct", "this was garbage", etc.

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

#249

Post by reportgunner »

csmpltn wrote: post #248
EamonnMR wrote: post #31

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

Why aren't they collecting feedback signals from people on the results produced by the model? "This was correct", "this was garbage", etc.

What if someone was asking to generate garbage in Blender though?

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

#250

Post by sinuhe69 »

iamflimflam1 wrote: post #177
iamflimflam1 wrote: post #176

Earlier quoted context omitted.

Typically, there's two ways to do this. The first is to just use the fact the GPT-4 will have seen a lot of blender code so just knows how to do it. The second way is to tell GPT-4 in the prompt what the API surface looks like and have it script against that.

You can actually just look in this file and see that the first method is being used. He feeds in a prompt with a basic example. https://github.com/gd3kr/BlenderGPT/blob/main/__init__.py

Oh, thanks. I thought somebody got a fine-tuning version of GPT-4. However, the author used the exact same prompts to demonstrate the ability of his interface, so I don't think it says much about its fitness for the purpose.
Post reply on HN