Yeah. I'm using it for some props that I will replace with paid assets or paying someone, I just need to pitch what I'm making so this is a pretty fast route for me. I tried bringing the 3D model into zbrush, sculpting more, baking onto a new low-rez with better UV and improving the baked texture giving the texture to leonardo, it was ok but not designed for it as the seam showed up more.
As a concept artist no. The main issue is that a.i art in the concept art scene is being used as a replacement to drawing. That’s the main problem. a.i should aid drawing not replace it, If you study a little about art and gain a better sense of design you’ll notice too how boring the design you’re showing in this video is. You can’t have ai do all of the work and have a good product. I hope that ai companies wake up and make better products that don’t try to make a full finished rendered image but help, the process of art.
Fair enough, and I'm sure that will happen. I think it's all just quick wins at this stage with these so called tools. Indeed the quality is very poor and if you are lazy like you see me in my video then yep, you will have pretty boring designs, but that might be all someone needs to fill a few gaps in whatever it is they are making, most likely some kind of game prototype. I'd happily use this to help with some assets that are not requiring special attention, background assets that don't require much pixels on the screen. When it comes to actual concept art for assets, props and characters, I would always go to a real person, an artist with feeling and expression learnt by many many years of study. My worry is, once you train A.I to learn correctly, and give it a memory and connect it to a kind of hive mind, then we can worry some more about that later!
My opinion about A.I is this. We use modern tools for efficiency. We collect data to improve products. We are already on this path of require digital assistance, a.i is just another level, the evolution of these things. Most people oppose a.i sure to 2 things: 1. The big scrape of data and 2. The environmental harm done by all these farms of computers processing the data to your prompt. We have similar issues all over the world with automation that are designed to improve costs (profits) for various businesses, such as self-checkout, these things make it easier to steal perhaps, but the benefit outweighs this because you save money on clerks. Electric cars are supposed to help the environment also, but where does the energy come from, renewable sources? Many companies are creating their own training data and a.i models, but who adjudicates this? The cons of using a.i are obvious: 1. you don't learn the proper work flow 2. your work is passing off on other peoples creativity 3. its basically garbage and will saturate various marketplaces that allow it The pros: 1. faster prototyping 2. visual communication aid 3. generate ideas without relying on one or several concept artists (especially where costs are related) You end up in a situation where bigger companies will aim to have their own trained models, ultimately reducing costs, enhancing their own workflow or pipeline. And smaller studios, indie companies will benefit from not having massive initial costs to create a prototype in order to gain investment, they will likely, and should, hire real artists once this investment is acquired. I like to think of a.i as a GPU able to solve small tasks fast, perhaps it helps fill gaps in photogrammetry where hours of man-power would normally be needed for this tedious work. Things of that nature where it could easily save costs with some human and a.i ingenuity. I guess there is much scaremongering going around about a.i as it explodes into all sectors, Chat-GPT especially. Replacing or reducing the workforce, and we're seeing many lay-offs, who knows, could be partly to do with a.i, could be COVID and lock-down related as more people worked from home, could be a bit of both. The purpose of this video is to show the non-skilled how a simple approach like this could enable them into the world of 3D, where they should question things. I will be making a video on this soon to show why this is not a good thing to assume these are good enough for production, that really depends on the production. Even turning these into simple icons like you see in the video requires more work, you can see I'm basically showing the laziest approach possible which is happening everywhere now, this reduces standards in most cases, we will see some pretty bad games appear where it's blatant how much badly generate a.i is used, that is without any extra thought or work gone into it. Over the last 2 years I've both enjoyed and hated using a.i, I hate the quality, the dead-ness, the sameness, the flaws, the plagiarism of it all, this is all in the mindset of 'I need to do extra work on this'. With the mind of knowing there is more work to be done, is could act as a gate for very young people to get creative, after that it makes sense to learn the right ways. When we are young we tend to copy everything we see, comic book characters, cartoon characters and so on, we might learn something from this, similar to how datasets work, they may limit us and we only draw the same things until we actually learn fundamentals. A.I has yet to learn anything like this, anything about fundamentals, understanding structure, light theory, anatomy and so on. It has no real memory, it is simply and algorithmic process of noise, prompts and de-noising on the data it has access to. When A.I learns fundamentals, then we are all pretty much replaceable, the monster has already been unleashed, it is simply in its infancy at this stage whether you choose to feed it or not, I find it reasonable to try to adapt it and use it if possible.
Interesting process but as a small game dev I don't think I'm ever going to use something like this (yet). One of the major issues AI has for things like this is changing minor details. For example if we decided we wanted the lock thing on the front of the chest to be a square instead of a diamond, there wouldn't be an easy way to change that without just doing the entire modeling process.
I'm sure the next iteration will have this for 3D models, just like we have after 1-2 years of the 2D image generators. Yes, indeed, please learn and change the models if they are considered final for your game, nobody really wants to make a game with a.i generated content. It's wise to make the adjustments as you polish your game.
1 year ago we said this will be only possible after 5 years but here we are with uv'ed ready 3d model creations. the development process speed is very scary.
Looks fine as an asset for some small game. Are you tried to bake the texture and ask Leonardo to redraw it? Good video btw.
Yeah. I'm using it for some props that I will replace with paid assets or paying someone, I just need to pitch what I'm making so this is a pretty fast route for me.
I tried bringing the 3D model into zbrush, sculpting more, baking onto a new low-rez with better UV and improving the baked texture giving the texture to leonardo, it was ok but not designed for it as the seam showed up more.
As a concept artist no.
The main issue is that a.i art in the concept art scene is being used as a replacement to drawing.
That’s the main problem. a.i should aid drawing not replace it,
If you study a little about art and gain a better sense of design you’ll notice too how boring the design you’re showing in this video is.
You can’t have ai do all of the work and have a good product.
I hope that ai companies wake up and make better products that don’t try to make a full finished rendered image but help, the process of art.
Fair enough, and I'm sure that will happen. I think it's all just quick wins at this stage with these so called tools. Indeed the quality is very poor and if you are lazy like you see me in my video then yep, you will have pretty boring designs, but that might be all someone needs to fill a few gaps in whatever it is they are making, most likely some kind of game prototype. I'd happily use this to help with some assets that are not requiring special attention, background assets that don't require much pixels on the screen. When it comes to actual concept art for assets, props and characters, I would always go to a real person, an artist with feeling and expression learnt by many many years of study. My worry is, once you train A.I to learn correctly, and give it a memory and connect it to a kind of hive mind, then we can worry some more about that later!
My opinion about A.I is this.
We use modern tools for efficiency.
We collect data to improve products.
We are already on this path of require digital assistance, a.i is just another level, the evolution of these things.
Most people oppose a.i sure to 2 things:
1. The big scrape of data
and
2. The environmental harm done by all these farms of computers processing the data to your prompt.
We have similar issues all over the world with automation that are designed to improve costs (profits) for various businesses, such as self-checkout, these things make it easier to steal perhaps, but the benefit outweighs this because you save money on clerks.
Electric cars are supposed to help the environment also, but where does the energy come from, renewable sources?
Many companies are creating their own training data and a.i models, but who adjudicates this?
The cons of using a.i are obvious:
1. you don't learn the proper work flow
2. your work is passing off on other peoples creativity
3. its basically garbage and will saturate various marketplaces that allow it
The pros:
1. faster prototyping
2. visual communication aid
3. generate ideas without relying on one or several concept artists (especially where costs are related)
You end up in a situation where bigger companies will aim to have their own trained models, ultimately reducing costs, enhancing their own workflow or pipeline.
And smaller studios, indie companies will benefit from not having massive initial costs to create a prototype in order to gain investment, they will likely, and should, hire real artists once this investment is acquired.
I like to think of a.i as a GPU able to solve small tasks fast, perhaps it helps fill gaps in photogrammetry where hours of man-power would normally be needed for this tedious work. Things of that nature where it could easily save costs with some human and a.i ingenuity.
I guess there is much scaremongering going around about a.i as it explodes into all sectors, Chat-GPT especially. Replacing or reducing the workforce, and we're seeing many lay-offs, who knows, could be partly to do with a.i, could be COVID and lock-down related as more people worked from home, could be a bit of both.
The purpose of this video is to show the non-skilled how a simple approach like this could enable them into the world of 3D, where they should question things. I will be making a video on this soon to show why this is not a good thing to assume these are good enough for production, that really depends on the production. Even turning these into simple icons like you see in the video requires more work, you can see I'm basically showing the laziest approach possible which is happening everywhere now, this reduces standards in most cases, we will see some pretty bad games appear where it's blatant how much badly generate a.i is used, that is without any extra thought or work gone into it.
Over the last 2 years I've both enjoyed and hated using a.i, I hate the quality, the dead-ness, the sameness, the flaws, the plagiarism of it all, this is all in the mindset of 'I need to do extra work on this'. With the mind of knowing there is more work to be done, is could act as a gate for very young people to get creative, after that it makes sense to learn the right ways. When we are young we tend to copy everything we see, comic book characters, cartoon characters and so on, we might learn something from this, similar to how datasets work, they may limit us and we only draw the same things until we actually learn fundamentals.
A.I has yet to learn anything like this, anything about fundamentals, understanding structure, light theory, anatomy and so on. It has no real memory, it is simply and algorithmic process of noise, prompts and de-noising on the data it has access to. When A.I learns fundamentals, then we are all pretty much replaceable, the monster has already been unleashed, it is simply in its infancy at this stage whether you choose to feed it or not, I find it reasonable to try to adapt it and use it if possible.
Interesting process but as a small game dev I don't think I'm ever going to use something like this (yet). One of the major issues AI has for things like this is changing minor details. For example if we decided we wanted the lock thing on the front of the chest to be a square instead of a diamond, there wouldn't be an easy way to change that without just doing the entire modeling process.
I'm sure the next iteration will have this for 3D models, just like we have after 1-2 years of the 2D image generators. Yes, indeed, please learn and change the models if they are considered final for your game, nobody really wants to make a game with a.i generated content. It's wise to make the adjustments as you polish your game.
1 year ago we said this will be only possible after 5 years but here we are with uv'ed ready 3d model creations. the development process speed is very scary.