Tuesday, 18 December 2018

WEEK 5 [17.12 - 23.12.2018] Multi-agent evolutionary systems for the generation of complex virtual worlds

Hi, today I would like to present to you the article about Multi-agent evolutionary systems for the generation of complex virtual worlds.

Link to the article:


Currently, movies, games and virtual reality applications depend to a large extent on computer graphics. Very complex models are often used to present many scenes and environments. Some works such as rendering, composition and animation have already been improved, but the modeling of complex models is still a laborious task. The authors of the article presented the use of an interactive genetic algorithm (IGA) for modeling computer graphics while compensating for user's program fatigue. The genetic algorithm is taught based on the choices made by the designer. Man chooses candidate solutions to be used in the evolution of the next generation of urban area. The training agent monitors the choices made by man. The training agent classifies new generations of sites based on several available classification algorithms and takes new decisions according to these classifications. The decisions made lead to actions, in this case the selection of new candidates. These choices are reintroduced into the genetic algorithm. The results indicate that the use of an Interactive  Genetic  Algorithm  with  a multi-agent implementation of a Human-Based Genetic Algorithm HBGA  allows for a greater exploration of the design space in a shorter time. Thanks to this, designers are less tired and can devote their attention to other details of created projects.

Questions:
1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
 

29 comments:

  1. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performed by appropriate algorithms?
    Yes, I agree. Of course, human will always be needed for the final assessment but the basic work will be done by Artificial Intelligence. In my opinion this issue applies to much wider area of our everyday life.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    I believe it is possible, why not? This issue boils down to the problem of choice. The choice is always dictated by some factors. Therefore, it is enough to create schemes that will form the basis for such choices (for example using deep learning). It can never be fully assessed because there is no objectivity in this aspect. Building any classifier on subjective assessments (such as colour or shape) is really bad.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    I hope it will happen. However, it will differ significantly from human work - at least at the beginning. In the future, it will be probably possible to reach a level close to human work. Certainly there will be situations when the user will not be able to distinguish who is the author of a given construction: AI or human.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    I really don’t know. There are certainly reasons for this and I am full of hope that one day it will come true. Personally, I think that it depends on the speed of AI development in terms of graphics, interest of industry and research in this area.

    ReplyDelete
    Replies
    1. Thank you for your answer. It's interesting how AI will affect our daily lives. It is possible that in many occupations the emphasis will be transferred from production to quality control of goods produced by artificial intelligence.

      Delete
  2. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?
    If the development of these algorithms will continue in this direction, then yes. Artificial intelligence will be able to create worlds that will be so great that man will not be able to repeat it in a reasonable time.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    Of course. Recently Artificial Intelligence algorithms started to paint paintings, so why they couldn't choose colors and textures in the generated worlds.
    https://www.theguardian.com/artanddesign/shortcuts/2018/oct/26/call-that-art-can-a-computer-be-a-painter

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    For very large projects I think that this type of solutions will be used. The results will be better and better, but always man will be the one who has to check the quality of what generated the computer.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    Honestly, I have never been more interested in graphics. I don't know how advanced it is in terms of hardware and algorithms used. But if there are scientists who are constantly developing these algorithms in terms of generating graphics, so maybe one day this will happen.

    ReplyDelete
    Replies
    1. Thank you for your answer. Artificial intelligence that paints images is particularly interesting, one can think about how well you can learn AI activities whose result is subjective. I think that you can find in art elements that resonate with the majority of society.

      Delete
  3. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    In my opinion, the human participation in IT project also in creation of artificial virtual worlds will be decreasing in the future. The artificial intelligence is becoming progressively more popular and it becomes the essential part of our life. I am not very interested in creating the virtual worlds but I suppose that the development of technology and science will find the solution allowing to obtain the appropriate algorithms to automate the creation process. It does not mean that humans will become completely unnecessary in this process – a human will still create general conception of the virtual worlds. The role of end-user will still reserved for humans as well. The process will be realized by the specified algorithms.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    I think that it is more possible that the AI algorithms will help the graphic designers and designers to create elements of virtual worlds. The AI algorithms are widely applied in different projects for example where the decision should be made or something should be chosen. The specified and trained neuron network allows to analyze and to choose colors, textures and etc. The suitable process of feature selection allows to choose the best elements. If we apply a supervised classifier, we need to label classes – it can be subjective but the development of standards allows to avoid such problems. The appliance of unsupervised or reinforcement learning seems to be promising idea as well. To sum up, it is the challenge for actual and the future AI programmers.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    I think that it is possible. As I wrote, I am not an expert in computer games industry. Saving the designers time is important but saving time of the other people who create computer games is valuable as well. If we are able to automate something, why not to automate it? The artificial intelligence algorithms are constantly developed. They are becoming progressively more accurate and find application in solving newly arising problems. I suppose that there will be not semi-automatic but automatic solution which is presented in the article. It is likely that the specified algorithms will be able to imitate the human work.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    To my mind, it will be possible to create AI-based graphic engine. The 3D computer graphic requires high computing power and a lot of fast disk operations. I suppose that AI – based graphic engine will also need the mentioned parameters. It is challenge for the present times. The AI-based graphic engine will allow to save time of numerous people and enable creating of virtual worlds both by graphic and non – graphic designers.

    ReplyDelete
  4. 1.Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?
    Yes, I think is possible taking in to account progress that was made in this field. In the future algorithms will have to be less generic and maybe suited for specific user preference to avoid recurrence and multiplication of the same pattern.

    2.Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    Is highly probable, composing graphic can be translate to decision making, so it can be passed to the algorithm. I think that cooperation of algorithm and graphic designers will be crucial to this project success.

    3.Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    Now a day’s time is money especially in entertainment business. If time of creation computer games and movies can be reduced it will certainly bring the profits to production company. And this will significantly accelerate growth of that kind of algorithms.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    Yes, I think it will be possible, but I also think that we need some time to develop that kind of advanced solution. This is a song of the future.

    ReplyDelete
    Replies
    1. Thank you for your answer. Shortening the production time and increasing profits are always very tempting, or you can assume that the final products will be more refined and full of details, because people will not have to work on basics.

      Delete
  5. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performed by appropriate algorithms?

    Yes, I am sure of that. The person will perform the function of checking the result, submitting the desired parameters for the desired result.For the person will be more creative tasks, and re-checking the result

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    Yes, of course, rather it would recommend to a person based on statistics. Most likely the algorithm will be built on the feedback of people - which colors are more acceptable for them. But even here, the main word will be for the person, since deep learning will be based on past results, and the person will be able to introduce something new.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    Probably yes, it greatly facilitates the creation of film and games. I think the work of the algorithm will be very different from the work of a person.


    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    Honestly, I find it difficult to answer this question. I dont know. I would like to see how it will look and whether there will be such an opportunity for humanity. Looking at the speed of technology development, I think that sometime it will be possible, but the result itself is interesting to me

    ReplyDelete
    Replies
    1. Thank you for your answer. For me it is a very interesting topic and I will be interested in the development of these ideas and their results.

      Delete
  6. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    I do not think that the whole creation process will be performed by appropriate algorithms. The human factor is essential, and still human imagination seems to be better than algorithms based on machine learning. The basic work might be done by algorithms, or be challenged by algorithms but not done from start to end.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    I can see such possibility. I think that based on machine learning also algorithm can figure out more complex and advanced elements. It is a matter of learning data set and advancement of the algorithm itself.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    Of course it will. As we know the human part is very costly and it will be reduced whenever it will be possible, so also in design industry. If producent have opportunity to cut cost they will do so. I cannot predict how visible it will be for the audience yet, but sooner or later they will specialise them.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    One day for sure it will be done. It is like a wizard of WIX for website or Wordpress for non software developers. There will be a lot of businesses and technologies that will enable that, simply because there will be a demand.

    ReplyDelete
    Replies
    1. Thank you for your answer. Personally, I think that human imagination is limited by the world in which we live and artificial intelligence has the potential in the future to exceed the ability to connect facts on surfaces incomprehensible to people.

      Delete
  7. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    I think that such algorithms can be co-creators of virtual worlds. Creating something imaginary can be done by artificial algorithms. It's exactly like with n-dimensional tables. When they were invented, no one could understand how they work. The entire digital world is n dimensional. Only people perceive (and not everyone ;-) )four dimensions.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    Of course. Already today, machines can do photomontages by adjusting the background using the algorithm. They will determine exactly what the background of something should look like in the same way. The same applies to the appearance (texture).

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    Such human-controlled algorithms will help in the implementation of large projects. However, human supervision is irreplaceable. He must look after the correctness of their functioning for the present moment. We can't leave a decision for machines. In the case of, for example, children's films, it could end with a slip-up.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    I don't think so. If someone isn't graphic, then the computer will not do it for him. There is always a taste different for different people. And what for one is nice for another nasty. Therefore, such algorithms will not solve this problem either.

    ReplyDelete
    Replies
    1. Thank you for your answer. Yes, it is true that people have problems understanding the n-dimensional world. I've heard about the game Miegakure, which is based on 4 dimensions and seeing the gameplay, it is difficult to know what's going on,while for the computer creating and understanding the game embedded in a ten-dimensional space would be child's play.

      Delete
  8. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    Maybe yes. This process could improve performance of creating a virtual world but the human must have a last word to say at the end of work. We have to control the process of creating a virtual world.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    I think yes but we will see what the future brings. For sure it will be a great help for designers.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    Propably yes but I am not sure. The filmaker could save a lot of time and do another things to make the film better. AI is a powerfull tool and we need to be carefully when we used it. Its possible that AI work coult be better than a human work. In some films effects could be very interesting. Today thanks to Nvidia we can see a virtual world created by AI. Link below:
    http://digg.com/video/nvidia-ai-rendering

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    I think yes. See the film in point 3.

    ReplyDelete
    Replies
    1. Thank you for your answer. The video that you have linked covers the interesting moral topic related to the identity. Advanced algorithms can be used in the future to create false evidence, or to create films that compromise other people without their consent.

      Delete
  9. Algoritms, eighter AI or rule based, can be successfully used in creation of virtual worlds as well as in desing in general. As a few examples here is a press article about a game with a possibly infinite universe (worlds are being generated when users explore them): https://www.technologyreview.com/s/529136/no-mans-sky-a-vast-game-crafted-by-algorithms/
    And here is a website with multiple examples of projects using algorithms for desing in general (no exactly desing of cities): https://algorithms.design/

    I'm not sure if the article tackled a real problem of designers. For me application of this method is totally made up :P I mean, if it was to help designers, there sould be some clear objective that users can set up and have it generated. If I wanted to have the highest buildings aggregated in a single spot then I would prefer to set this at the beggining and not guide the algorithm to a such goal iteration by iteration.

    I'm pretty sure that technology can support the whole process of desing to such extend that even non professionals are able to produce appealing works. They even can be unique and some sort of "artistic", but I think that still such solutions will be inferior to whatever professional designers create.

    ReplyDelete

  10. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    I think it's a bit hard to tell. So far, machines have proven useful in repetitive tasks. Once the task is defined, it can be optimized. Creativity is not based on such a model. It is also subject to fashion - we have enjoyed "computer generated worlds" since Pac Man or maybe better - NetHack. But then there is something new, a new trend and previous works of art, like computer games, fall into oblivion. Then the new paradigm can also become conventional, and new works are formulaic remixes of previous works - this could be taken over by computers.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    This part is, in my opinion, easier that the "general layout" design problem. And still there is no "right" and "wrong" in an objective sense ("user fatigue" is vague - what if our "metric" is not in the mood?).

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    That depends on the operator. If it goes through such a gatekeeper, the quality can be guaranteed at a level of the operator's choosing. But what if the algorithm never produces something as good? Maybe subpar products would be allowed, or the human would have to switch to manual control and move the creative process forward. Then maybe continue with the algorithms. However - it is much easier to be a critic, then to create on one's own. Just try painting.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    There would have to be some input - for games set in contemporary reality (like GTA) it would be easy to feed it real-life examples. But what about fantasy worlds? Like the Witcher? There would need to be some creative input before we sit 'anyone' in front of it, to give feedback to an algorithm and create a great game in the effect.

    ReplyDelete
    Replies
    1. Thank you for your answer. The unfolding of roguelike games on the gaming industry as well as the increasing use of randomly generated maps, and the selection of opponents from ready-made elements, may indicate that game developers sometimes like to create "random" games and the worlds presented in them.

      Delete
  11. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    Yes, I agree with previous opinions. Artificial intelligence will be constantly developed and created for "simpler" tasks (for example: calculations, schematic tasks). The remaining part requiring the development of the application and supervision will be managed by human.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    In my opinion, artificial intelligence algorithms can help designers. Examples include filters in graphic programs, search algorithms, automatic camera settings when You taking a photo. Photoshop has the function of adding a background after cutting the layer. As for the choice of color, the computer will choose it more precisely than a human, but the final decision should be always remain with the designer.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    I believe that the artificial intelligence algorithms will be used by filmmakers on a large scale. Films with special effects are very popular and require a large budget. In addition, the use of algorithms can have an impact on reducing film cost production (you do not have to pay graphics for hundreds of hours). Of course, algorithms should be used with reason, and the whole should be supervised by a human.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    It can be stated that it was already possible to use AI in the AnvilNext engine used in Assassin's Creed games. The engine is responsible for creating the city. Such engines enable the creation of games much faster and with not expensive. The presented algorithms in the article are the first step in this type of solutions. Maybe it would be worth paying attention to and getting some ideas from the engines already available and tested for example Unity.

    ReplyDelete
  12. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?
    I think there will always be a human in the loop, to provide feedback. After all, it's all about human perception of the virtual worlds. It might be possible to use target audience for feedback, but I guess the generate-evaluate-update loop would be too long then.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    Yes, I've stumbled upon this: http://vcc.szu.edu.cn/research/2018/TexSyn "Non-stationary Texture Synthesis by Adversarial Expansion" - a paper on synthesizing large textures using small example as base. Also generating new materials based on couple of photos, without tedious fine-tuning of layers and parameters could be possible.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    I'm quite sure it will, if not used now. I even suspect there is something that goes one step further: the Youtube channels of algorithmically-generated children nursery rhyme videos, constantly gathering data on views and optimizing for maximum engagement.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    Of course, one can use stock texture packs and models, and create something that looks really good. Such worlds would be utmost boring though, as it takes some practice, knowledge and imagination to create world that captures attention.

    ReplyDelete
  13. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performed by appropriate algorithms?

    In my opinion, yes. Men will be needed only for final product quality assurance, or maybe even that will be done by Artificial Intelligence. Who knows:)


    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    I think it's possible, but I do agree with my previous speakers that it's all about solving the problem of choice and selection. But in the end, if everything will be done on the basis of intelligence algorithm, graphic designers will lose all the fun and satisfaction they get from their work.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    I guess, it's only a matter of time. When it comes to quality of work done by AI and human being, at first, there might be some difference, but in the end, I'm sure we won't be able to distinguish between pieces done by AI and men.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    I don't know, but anything is possible, so let's wait and see.

    ReplyDelete
  14. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    I think at some point human will need only to provide the idea of the world and at the end evaluate the results, the rest will be performed by AI. Currently some GAN deep networks are able to sell art at auction, so generation of virtual worlds sound like a near future.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    As I mentioned in point 1, if the AI is able to generate art, then choosing textures or colors doesn't sound complicated. Of course currently i believe the first step should be creating good training data for such algorithm and after that try and error method to find working AI soution.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    If the quality in real application will be good then I'm sure, that filmmakers will be interested in saving money on designers:)

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    If it will be possible, then it will be a hugh challange for graphic disigner to compete with such engine, but it will be good and will raise level of today's 3D graphic, so I hope it will be possible. Next problem could be a performance of such engine, normally creating of high-resolution graphic means to use a lot of GPU power, so the question is, if non-graphic designer will have such powerfull PC.

    ReplyDelete
  15. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?
    I'm not sure of that. I know that this process could be automated in first part when designer create typical elements of the generated world (e.g. houses, trees...), but the really important in Computer Graphics are details. They differentiate the good game from the bad. Nobody likes to see an objet which don't match to generated reality or more which breaks it's law (e.g. penetrates through the walls).

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    Of course. Moreover I heard that similar algorithm was created before. It could create quite easy objects (consisting of several elements) and generate suitable good parameters. I heard about it near 2 years ago so possibly now this algoritm could evolved to create much more complicated objeccts.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    As I said before that algoritm (not only this one from article) was made at least 2 years ago. I don't think that they aren't using by filmmakers or in computer games designer. I can't say anything about comparasion between human and algorithm work, but I think that the human would be much more better in it,but he would need more time...

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    I think that it is absolutely possible, but we have to remember that this world wouldn't be as good as created by professional which will be assisted to that algorithm.

    ReplyDelete

  16. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    No, I don’t think so that it will be the way it works in the future, because always the most important element is he concept, created by the human. It can’t be replaced fully effectively by appropriate algorithms. From the beginning there were methods of automatic generation of the worlds in the computer games, but eventually there weren’t as good and effective as worlds designed by humans. For example in the game Minecraft the world, which was automatically generated, still was reorganized by the human player. So I think that there won’t be replacement of the human interference and work by appropriate algorithms.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?

    I am sure this is possible. In such cases the algorithm should be able to compare our work and expected result with the ones currently existing. But in my opinion the work of concept graphics and their original projects are irreplaceable.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    Some of the predictable effects for sure can be generated by the algorithms and then used by filmmakers, but only if it won’t cause the situation, where all of the movies look alike. After the CGI became popular, special effects were very similar in different movies, but still it was profitable to use them, because they draw the attention of the audience.


    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    I am not sure, but I don’t think it s highly unlikely. Nowadays every technologies are developing really fast, so we basically cannot exclude anything. But as I mentioned above, I think that in this case the created worlds would far more similar to each other.

    ReplyDelete
  17. Hello,
    thank you for an interesting article!
    I'm a fan of an approach to AI which is not a "replacer" but rather a "supporter" for human in a task and problem-solving. Especially, in challenges which are time-consuming yet not so creative and simply dull. I do not know the craft of artists, but I can imagine that some operations, such as texture and colours choosing and filling could be an example. Furthermore, a selection of proper scene enlightenment composition and camera action could be proposed by AI, and after that- adjustment by a human, to give it finishing touch. However, the filmmaking is such a heartfelt and unobvious process that the last word will belong to man. At least- for now.

    ReplyDelete
  18. Dear Monika,
    Thank you for interesting article.
    1. In fact, definitely I deeply believe that in the future human interference in any kind creation process can be negligible. Building the elements of artificial world can be a bit easier, because after all, if you don’t have to map something existing very precisely, then you can follow your imagination embedding that in technical possibilities. There are already many creations proving using AI algorithms to such purposes, so I look optimistically into this future, because it can only be more interesting and closer to quasi-human effect.
    2. Consequently, as I mentioned above, I have to maintain my opinion, because such details like textures, colors etc., will always take us several steps further in this designing process, comparing to previous effects. All those small elements, taken together, will give us results unexpected now. Why? Because still, as humans, we cannot believe we can be really replaced. Our ego cannot believe it. That’s why, in my opinion, it is better, that it goes through small details, because it helps to concentrate on them rather than on the complex effect, which will be worth seeing, I’m sure.
    3. Again, as for first question – yes, because it will let designers develop their other skills instead of those known and used before. And as for the second question – no, because there are different people, with different level of creativity. I guess there will be a lot of high quality effects created by AI, which only few very smart humans could beat.
    4. Definitely, yes. It only depends on the level of complexity of the elements which will be used by the user. If they are simple, then the user has to be more professional. If they are more sophisticated, then they can be used practically by everyone. Interesting issue, and impressing idea :-)

    ReplyDelete
  19. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?
    I think that it is technically possible but it won't happen. More possible is that some boring parts of creation process will be automatized but not the full process. People still like to be creative, want to do this kind of job and still perform better than a machine which can only follow some kind of algorithm.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.? I am sure that AI can halp to choos textures colors, shapes, composition and many more but we still need human in this process. For example if we have some shape the algorithm can sggest the proper texture or set of textures that are feasible for this shape but the final decision is up to the desinger. AI can learn how to create some kind of realistic sceen but sometimes we want to do someting different and then humans are better.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?

    I think that presented solution may speed up the process of making a movie or computer game. The quality may differ from the products made by humans so then someone should be aware and make the proper corrections. I mean that this methods is good to do some standard sceens and more sophisticeted should be directed by human or at least check by human and corrected.

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?

    I think yes it is possible. According to historry at the beggining in every kind of software we had only simple programms that were possile to be used only by proffesionals. Now programms are more complex but easier to use by users.They heve more options but better desing and a lot of automatic functions for instance to make photo postprocessng extremely easy and fast. The same may happen with 3D computer graphic.

    ReplyDelete
  20. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?

    I think this is the nowadays. Scientists from Nvidia proposed a similar Technology. Nvidia's new technology uses a deep AI neural network to analyze existing films and then apply visual elements in new 3D environments. This new technology can be a revolutionary step forward in creating 3D worlds. In addition, this technology allows you to train AI models with video to automatically render buildings, trees, vehicles and objects in new 3D worlds, instead of requiring the usual tedious process of modeling scene elements, the results are still nice. It is also possible to generate scenes in real time. Nvidia scientists also used this technique to model other movements, such as dance movements, and then applied the same movements to other characters in real time. It can also be used to create new levels and content in older games. Over time, the company expects this technology to spread and become another tool in the game developers' toolbox.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    Yes, because graphic designers can enter the outline of the scene (indicating where different objects should be), and the algorithm will fill the outlines with realistic details. At this stage, the graphics generated by the algorithm are not perfect - because some objects may warp slightly, almost like in a dream - but the output is already very advanced.

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    I think it is already the case that artificial intelligence algorithms are better than images painted by man. Rutger's Laboratory of Art and Artificial Intelligence explored a computational system for the art generation that does not involve the human artist in the creative process, but engages human creative products in the learning process. An important element of art-generating algorithms is the connection between their creative process and art that has been created by human artists at all times. Researchers believe that this is important because the human creative process uses previous experiences and exposure to art. Deep neural networks have recently played a transformative role in advancing artificial intelligence across various application domains. Several generative deep networks have been proposed that have the ability to generate novel images to emulate a given collection. Generative Adversarial Networks (GAN) have been quite successful in achieving this goal. GAN learns to generate images through a game between two players.,


    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    Such an algorithm already exists, I described it in the first part of the answer. https://www.theverge.com/2018/12/3/18121198/ai-generated-video-game-graphics-nvidia-driving-demo-neurips

    ReplyDelete
  21. 1. Do you think that in the future human interference in the creation of artificial virtual worlds will be negligible and the whole creation process will be performer by appropriate algorithms?
    In my opinion human involvement will be necessary for achieving reasonable results. Parts of creation process could be automated but still somebody has to have clear vision and should control the creative process.

    2. Do you think that it is possible that artificial intelligence algorithms (not necessarily genetic algorithms) can help graphic designers and designers to create not only schematics of virtual worlds composed of simple elements, but also to choose colors, textures etc.?
    Yes this could be done , but the real question is, is it efficient do these approach help in any way. This task seems to be eligible for automation but is it worth it?

    3. Do you think that the semi-automatic solution presented in the article (part of the work is done by genetic algorithms, which is supervised and evaluated by a human being) will be used by filmmakers or computer games to save designers time? Would the quality of such a solution differ significantly from human work itself?
    If it will be cost efficient, it sure could be used by professionals to optimise daily work

    4. Do you think that there is a possibility to create an AI-based graphics engine that would allow anyone, even a non-graphic designer, to create a virtual world that does not diverge from today's high-resolution 3D computer graphics standards?
    Once again I do think this is possible but why any company should invest large budget to do this

    ReplyDelete