There’s no way any of us can predict what the future holds. I have my own personal theories about your questions but I’m not going to share them here because it would be irresponsible. The only thing you can do is your best, and let the future worry about itself
Sometimes yes, mostly not ![]()
And if it doesn’t work, you often can’t have GPT fix it, because the error usually gets worse instead of better.
You can also request PBR all at once, but you have to specify that you need it separately ![]()
Your table looks great ![]()
Yeah… Try 6 months maybe. Or maybe say “Planning to start learning”…
I don’t use substance, but then it’s not only about the software. You need some time to learn to see textures… it’s gonna take some time. No reason to have unreasonable expectations. Unless you enjoy failure.
Sure, you can do all this with Photoshop and similar programs, but the use case shown here is particularly useful when you don’t yet have a texture to edit or improve. GPT essentially creates the desired texture (which I would otherwise have to search for or buy) out of nothing. So it’s not about editing/extending, but about having and creating the basic texture. Within 10 seconds, I can describe and create a texture that I didn’t have before.
But that’s only useful for some quick images, not for productive or real-world projects.
And I’m usually more of a fan of procedural textures anyway.
You can use almost your “prompt” as a search query for some non-AI search engine:
texture image wooden table with signs of wear and tear, a few cup marks (water stains), nicks and scratches :
..properly it’s “better” to use any of those texture image collection websites.
For profesionals: they might even use some paid site to be on the “safe” side according licenses.. (that’s usually and initally the idea about specializing: there is someone who can do it very good, so to save time one does pay for it.. i do not make/build my cloth, furniture or groceries for example for a reason
)
Also: where is als the fun in this.. ??? ![]()
I think that in the not-too-distant future, such AI-generated textures will eventually become real competition for texture/material shops like Poliigon and others. But I can’t really imagine it in modeling at the moment.
Absolutely not. Because AI.
Also Polligon has been selling AI textures for several weeks now.
Wow, I didn’t know that. Weird.
I strongly doubt it. Your efforts to texture should be centered on what story the textures tell about the object you are painting. That story informs what you do with it - going overboard or being very clean as far as dirt/damage/detail etc. Artists are the only connection to do that imho.
Yeah, no, the AI doesn’t create something from nothing. Please be aware that the AI only mixes other results from human artists to get that product you decide is usable to you - not from nothing.
To create seamless textures or channels I do not use ai, the only reason I use it is when the designers/clients give me a very low res image, for those cases I use gigapixel locally (I managed to buy a permanent license back in the days) to upscale it and then my go to tool is materialize https://boundingboxsoftware.com/materialize/. It is so easy to use and it saves the texture as a project, so you can go back and re-adjust parameters if need it.
I always wondered when it comes to AI so many people do use this (or some similar) phrase.. for example: “if this get better in a few month then..” totally ignoreing that this is some logical contition. If and only if.. but some already pay for this.
On the other side some people tell about Blender: “It’s not there yet”..
Que Sera, Sera.. ![]()
![]()
When I see people thinking about using AI to offload the creative part of what they do, it always makes me think of this:
I use AI to refactor, debug, and patch problems in code- the boring, tedious, time-wasting stuff, so I have more time to do the creative stuff- texturing included- that brings me joy. Why would I want a machine to do all the joyful parts? What’s left for me to do then? The tedium? The drudgery? Why would I automate away the fun parts of life?
Putting ethical problems aside for this… I don’t see the point. Between polyheaven, ambient cg, adobe, marmoset, quixel mixer (even if it’s no longer maintained it ships with huge library), etc. you can basically have any texture you want. If I want something more unique there are tools for that (painter, designer, mixer, toolbag, instamat and hopefully blender soon), or I can take a photo and turn it into material. Also, as was mentioned in this thread: where is the fun in just prompting for that (and getting uncontrollable/unpredictable results)?
I don’t know about that.
It would be nice if that was the case, but I think for many people and many situations it isn’t. I think creating textures is something most people in these forums do a whole lot. I think AI is useless for this in vast majority of cases, but I wouldn’t write it off completely. It just doesn’t work the way people imagine. At all
No, it cannot texture a character, that’s bs. No, it cannot create a texture for archviz. No, you cannot just write a prompt for anything you can imagine and get something useful, that just doesn’t work. ![]()
I think there are thousands of little tricks to make textures for thousands of situations - Photoshop, painting, procedural stuff… all have use cases and there seem to be a few tricks AI can actually do for this job sometimes. But only sometimes like with everything else, only specific use cases, specific little tricks and they still require thinking, still require working and still require knowing what one is doing. Of course people are gonna ignore the reality here and try to use it for stuff that it absolutely doesn’t work for in ways that have no chance of working… but that’s a trend with AI.
It’s easy to hate that. But I think if I find a little trick I can use AI for in making textures that works, I’m gonna use it. It’s tricky to find though.
Most things AI are just a waste of time.
I occasioanlly use it due to time contraints, but mainly I will take a photo of for example a rear car light and run it through a filter of sorts to make it slightly illustrative looking
But minimal is the way to go, used kria to do this. would be just as easy to use a photoshop filter or paint over the photo to remove the highlights etc
also I sometimes use them as a modelling guide, cutting in to the texture with the knife tool and then giving them flat colors
these ones are a bit more glaringly obvious, but I have done others that are a lot less ai looking}
also it just an albedo, so it will be pretty much disguised by specular etc
FYI guys, Image to Materials in software like Substance used genAI for years even before chatGPT became a thing
genAI is not only prompts
Well, we can’t predict the future, it’s very likely that AI is going to be part of the toolset but that doesn’t mean artists will become obsolete.
However, as image making get simpler, we have access to a lot of assets through online stores, tools get better, and a lot of people want to work in CG…
As a result the entry level is quite high and projects are likely to involve less people.
I think 3D artists aren’t going to be replaced but it can be much harder than before to find your place in that industry, at least currently we are in a major crisis in the field it’s also unclear how we’ll recover from that…
Generally what I told students / people willing to get into this business is how much you want to invest in getting good at this, and how much you’d accept uncertainty to live your passion ?
It’s getting a risky business like acting, music … some people manage to make that work, but this implies some amount of sacrifices to get there and risking to fail also…
Obviously it’s similar to everything in live, some stuff are just riskier than others…
wait what, that’s so ironic. i wonder if that intentional or just something that get overlooked by reviewer. though i don’t know how they work (managing/publishing texture)
I am quite sure that it is a deliberate decision. Andrew seems to have quite a pragmatic approach to AI (kinda: use it or perish)




