3 ms·

You can definitely do that right now in midjourney 4, go on their discord and search for "character sheet" or "character design" and you'll find high quality ex

You can definitely do that right now in midjourney 4, go on their discord and search for "character sheet" or "character design" and you'll find high quality examples with coherent front/side/back view Once you know what tokens trigger the right answers it's relatively easy to get where you want, as usual with these tool you should be flexible when it comes to details, but the overall design can be defined

Sakos 4y ago
So, uh, is every artist going to be out of the job in the next 5 years?
- lm28469 4y ago
  I personally don't think so, it made incredible progress very fast but it stagnates now. The more precise design you want the less likely you'll be able to generate it. It's usually very nice looking at first but once you look at the details it completely breaks apart, it's a bit like in a dream.
- idopmstuff 4y ago
  Honestly, I expect the number of art roles at gaming companies to drop precipitously over the next couple of years. Right now you have a creative director who defines the style of art and a bunch of early-to-mid-career artists/designers who then create all the assets you need (and there are a lot of assets, so there's plenty of work to go around). Soon that'll be a creative director defining the style, AI producing most of the assets, and a few designers cleaning up what the AI produces. In a few years, those few designers probably won't be needed. I think a lot of artists who take commercial art/design type jobs to pay the bills are going to be out of work soon, and that's a serious problem that we should be looking to get ahead of as a society. On the plus side, this will be a huge boon for indie game devs - they'll be able to create AAA-level assets for their games at extremely low cost. For better or worse (for better and worse, really), the times, they are a-changin'.
  - simion314 4y ago
    >I think a lot of artists who take commercial art/design type jobs to pay the bills are going to be out of work soon, and that's a serious problem that we should be looking to get ahead of as a society. They can learn this new tools and be 100x more productive and make more money. The AI tool will not replace the artistic taste and creativity , some boring low quality work will be automated but artists can be more productive with this new tools and say finish the artwork for a novel or game much faster, so making more money.
    - sebzim4500 4y ago
      Yeah but if you make artists 100x more productive the natural consequence is that there will be far fewer employed artists. The senior ones will probably be ok (as you say, they might even get a payrise, given how much impact a single individual will be able to have) but people just joining the industry now are likely screwed.
      - mentos 4y ago
        I think it could mean games have the same art budget but make 100x the art content to compete with studios that use the innovation to slash their headcount..
        mike_hearn 4y ago
        Automation and better technology massively reduced the fraction of the population that works in farming, whilst simultaneously we ended up with a food surplus. Machine translation definitely had an impact on the human translation industry. Seems likely that this is what will happen for art.
      - simion314 4y ago
        No, there will be more art. Say I would like to make a Visual novel but it will take an artist 100 hours to illustrate it, if he can do it 10x faster then I can have him illustrate 10 of my novels with the same money. Nobody loses and my VN readers can enjoy 10x more content. Btw this AI tools work a lot better if you sketch something first, so I can see a very productive artist doing a rough sketch then using AI then manually touching stuff after. Also an artist could train the AI with his style and with his characters and keep the model private and use it over and over again.
- simion314 4y ago
  >So, uh, is every artist going to be out of the job in the next 5 years? No, probably artists also ask if ChatGPT will replace developers and they can just ask the AI write me the code for a cool RPG game, make it in CoolLang and use NN for NPC AI, also make it super efficient and optimized and similar keywords. This tools will be used by artists in their workflow, they will be more productive so more high quality art will be created .
- numpad0 4y ago
  The author seems to be a 3D artist, and claims to have spent 18 hours for this work, that he estimates as 5 days worth, that he thinks not copyrightable. To me it seems to be just raising the bar/granting more productivity for artists with prior skills and abilities to read the room.
  - Multicomp 4y ago
    My understanding re: copyright was that the raw output of the AI cannot be copyrighted. But his rigging and mesh maps and otherwise creatively-directed work to take that raw material and make it look good? Surely he has copyright on those? It's just that in theory, anybody else can take the raw output of the AI and do all that extra work on it themselves with no copyright violation.
- harrylove 4y ago
  Short-sighted employers will probably cut staff. The other more likely scenario is that teams will be asked to produce more work faster by letting AI handle the boilerplate. Creative directors and producers will run a lot more experiments with fresh ideas and then use the expert skills of their teams to produce a cohesive, polished product. The end result may be that we get a ton of mediocre, high-fructose-corn-syrup art at first. The hype for that will last about three months. We’ll also see an increase in typical fare from big studios as they figure out how to monetize long-tail content. And then occasionally there will be a true work of art or a new form of art we haven’t seen yet. Supply and demand.
crazygringo 4y ago
I'm mostly just shocked that the model can produce what appear to be entirely consistent models of the same character from different views. I can be convinced, but I was also equally skeptical. Because that implies a kind of cohesive conceptual understanding of a single object, that I didn't think the current architecture was able to model that. So many image outputs of humans are perfectly fine in local areas, but then they have three arms or a hand where a foot should be or something. I mean, the chosen character isn't perfectly consistent -- e.g. the side view shows a kind of metal badge on the vest that doesn't exist in the front view, while the back view shows a kind of mid-thigh holster (?) that doesn't appear on the front view, plus the vest on the back view is a few inches shorter. Similarly the collar is popped on the front view, but not the other two. But nevertheless, it's basically about 99% right. And I simply don't understand how that's possible, unless there author did about ten thousand tries and is showing the only ones that happened to be nearly entirely consistent, somehow just mostly by chance. (Maybe because they're particularly close to the real-life characters used for training?)
- minxomat 4y ago
  > kind of cohesive conceptual understanding of a single object It doesn't, it just requires character sheets in the training set, the rest is basically a style transfer.
- YeGoblynQueenne 4y ago
  >> Because that implies a kind of cohesive conceptual understanding of a single object, that I didn't think the current architecture was able to model that. It works because it's all in one image, you still can't get the same consistency across different images. Someone did the same thing with sprite sheets the other day: https://www.reddit.com/r/StableDiffusion/comments/yj1kbi/ive_trained_a_new_model_to_output_pixel_art/ https://www.reddit.com/r/StableDiffusion/comments/yj1kbi/ive... Which again works because you're asking for a single image that is similar to other single images (i.e. other sprite sheets). I thought that was very clever. So I guess the next big thing will be to train on datasets of entire graphic novels where all the pages of each novel are stitched together into "one image". Then we will finally achieve the graphic singularity and comic book artists will have been replaced, too :P

Sakos 4y ago

So, uh, is every artist going to be out of the job in the next 5 years?

lm28469 4y ago
I personally don't think so, it made incredible progress very fast but it stagnates now. The more precise design you want the less likely you'll be able to generate it. It's usually very nice looking at first but once you look at the details it completely breaks apart, it's a bit like in a dream.
idopmstuff 4y ago
Honestly, I expect the number of art roles at gaming companies to drop precipitously over the next couple of years. Right now you have a creative director who defines the style of art and a bunch of early-to-mid-career artists/designers who then create all the assets you need (and there are a lot of assets, so there's plenty of work to go around). Soon that'll be a creative director defining the style, AI producing most of the assets, and a few designers cleaning up what the AI produces. In a few years, those few designers probably won't be needed. I think a lot of artists who take commercial art/design type jobs to pay the bills are going to be out of work soon, and that's a serious problem that we should be looking to get ahead of as a society. On the plus side, this will be a huge boon for indie game devs - they'll be able to create AAA-level assets for their games at extremely low cost. For better or worse (for better and worse, really), the times, they are a-changin'.
- simion314 4y ago
  >I think a lot of artists who take commercial art/design type jobs to pay the bills are going to be out of work soon, and that's a serious problem that we should be looking to get ahead of as a society. They can learn this new tools and be 100x more productive and make more money. The AI tool will not replace the artistic taste and creativity , some boring low quality work will be automated but artists can be more productive with this new tools and say finish the artwork for a novel or game much faster, so making more money.
  - sebzim4500 4y ago
    Yeah but if you make artists 100x more productive the natural consequence is that there will be far fewer employed artists. The senior ones will probably be ok (as you say, they might even get a payrise, given how much impact a single individual will be able to have) but people just joining the industry now are likely screwed.
    - mentos 4y ago
      I think it could mean games have the same art budget but make 100x the art content to compete with studios that use the innovation to slash their headcount..
      - mike_hearn 4y ago
        Automation and better technology massively reduced the fraction of the population that works in farming, whilst simultaneously we ended up with a food surplus. Machine translation definitely had an impact on the human translation industry. Seems likely that this is what will happen for art.
    - simion314 4y ago
      No, there will be more art. Say I would like to make a Visual novel but it will take an artist 100 hours to illustrate it, if he can do it 10x faster then I can have him illustrate 10 of my novels with the same money. Nobody loses and my VN readers can enjoy 10x more content. Btw this AI tools work a lot better if you sketch something first, so I can see a very productive artist doing a rough sketch then using AI then manually touching stuff after. Also an artist could train the AI with his style and with his characters and keep the model private and use it over and over again.
simion314 4y ago
>So, uh, is every artist going to be out of the job in the next 5 years? No, probably artists also ask if ChatGPT will replace developers and they can just ask the AI write me the code for a cool RPG game, make it in CoolLang and use NN for NPC AI, also make it super efficient and optimized and similar keywords. This tools will be used by artists in their workflow, they will be more productive so more high quality art will be created .
numpad0 4y ago
The author seems to be a 3D artist, and claims to have spent 18 hours for this work, that he estimates as 5 days worth, that he thinks not copyrightable. To me it seems to be just raising the bar/granting more productivity for artists with prior skills and abilities to read the room.
- Multicomp 4y ago
  My understanding re: copyright was that the raw output of the AI cannot be copyrighted. But his rigging and mesh maps and otherwise creatively-directed work to take that raw material and make it look good? Surely he has copyright on those? It's just that in theory, anybody else can take the raw output of the AI and do all that extra work on it themselves with no copyright violation.
harrylove 4y ago
Short-sighted employers will probably cut staff. The other more likely scenario is that teams will be asked to produce more work faster by letting AI handle the boilerplate. Creative directors and producers will run a lot more experiments with fresh ideas and then use the expert skills of their teams to produce a cohesive, polished product. The end result may be that we get a ton of mediocre, high-fructose-corn-syrup art at first. The hype for that will last about three months. We’ll also see an increase in typical fare from big studios as they figure out how to monetize long-tail content. And then occasionally there will be a true work of art or a new form of art we haven’t seen yet. Supply and demand.

crazygringo 4y ago

I'm mostly just shocked that the model can produce what appear to be entirely consistent models of the same character from different views. I can be convinced, but I was also equally skeptical. Because that implies a kind of cohesive conceptual understanding of a single object, that I didn't think the current architecture was able to model that. So many image outputs of humans are perfectly fine in local areas, but then they have three arms or a hand where a foot should be or something. I mean, the chosen character isn't perfectly consistent -- e.g. the side view shows a kind of metal badge on the vest that doesn't exist in the front view, while the back view shows a kind of mid-thigh holster (?) that doesn't appear on the front view, plus the vest on the back view is a few inches shorter. Similarly the collar is popped on the front view, but not the other two. But nevertheless, it's basically about 99% right. And I simply don't understand how that's possible, unless there author did about ten thousand tries and is showing the only ones that happened to be nearly entirely consistent, somehow just mostly by chance. (Maybe because they're particularly close to the real-life characters used for training?)

minxomat 4y ago
> kind of cohesive conceptual understanding of a single object It doesn't, it just requires character sheets in the training set, the rest is basically a style transfer.
YeGoblynQueenne 4y ago
>> Because that implies a kind of cohesive conceptual understanding of a single object, that I didn't think the current architecture was able to model that. It works because it's all in one image, you still can't get the same consistency across different images. Someone did the same thing with sprite sheets the other day: https://www.reddit.com/r/StableDiffusion/comments/yj1kbi/ive_trained_a_new_model_to_output_pixel_art/ https://www.reddit.com/r/StableDiffusion/comments/yj1kbi/ive... Which again works because you're asking for a single image that is similar to other single images (i.e. other sprite sheets). I thought that was very clever. So I guess the next big thing will be to train on datasets of entire graphic novels where all the pages of each novel are stitched together into "one image". Then we will finally achieve the graphic singularity and comic book artists will have been replaced, too :P