Відео

@@plur_ndbn Привет. Спасибо за вопрос. Как я говорил ранее, это независимые агенты и это заметно во второй части этого демо, да и в других так же. Так же скажу, что кроме интеграции в игровые проекты, одна целей продукта это реализация балансировки нагрузки на маршруты на лету. Ещё раз спасибо за вопрос и если вы знаете возможного спонсора или инвестора, пожалуйста, дайте ему знать об этом проекте. Заранее благодарю.

@@mx.s Привет. Сразу скажу что не знаю инвесторов, ютюб посоветовал другое видео, где есть твой комментарий. ну типа есть же магазин юнити и эпик стора попробуй оформить в виде модуля и продавать там, но это очень всё зависит от реализации и быстродействия, как бы в стратегиях даже 90-начала 2000х была возможность управлять огромным числом юнитов. ну и вторая часть мне не показывает ничего, я вижу что идёт поиск пути до цели, а если на пути агент - до агента, типа убери проверку на коллизию агентов друг с другом и они все в одну точку сольются же. Если вдруг ты решил проблему поиска пути и сделал лучше чем а-стар, то напиши научную работу на это, люди то бьются с проблемой)

@@plur_ndbn Спасибо за Ваш комментарий и советы. Создание аддонов не является приоритетом. Тем не менее мне очень нравится концепция свободных (условно свободных) инструментов разработки после выпуска основного продукта. Пожалуйста, извините, но я не могу обсуждать Ваше мнение о демо в контексте какого-то угадывания внутреннего устройства по причине поиска инвестиций. Касательно публикации научной работы, это не является моим приоритетом Спасибо за диалог. Было очень приятно с Вами общаться.

@littlesaigon1042 Thank you for your question. These are independent agents. They do not follow each other and there is no trick of this kind. I am very sorry that I cannot say more, as I am looking for an investor to continue the development. Once again, thank you very much. Feel free to write to me.

@GamouPlays Thanks for the comment. I think my solution is amazingly fast, and it's still an alpha version. So there is no direct comparison. I think in production the performance will be even better, and 2 of the 3 optimization steps are not implemented yet. But you can dive into the topic of 3D pathfinding or, for example, evaluate the demo with 10,000 agents to preliminarily evaluate the performance of the solution. As for the possibility of purchase, I really like the concept of free development add-ons after the release of the game. It's just a question of finding investments. If someone can help, including information - I will be very grateful. Very grateful. Hugs.

​@@mx.s Do you do software development as your job or education?

@GamouPlays Thank you very much for the communication. I can gladly answer your questions about me in direct communication. But still, I can say that there are no employers, organizations or any persons who can claim the rights to my code. This is only my development. P.s. I looked at your channel. I can say that in a game like ZeroSpace, the integration of my solution, in addition to other advantages, could give, you know, at the moment when units meet for battle. They just stand and shoot. Just a crowd of units that shoot, although I myself love RTS, and ZeroSpace will be very cool. So at this moment, my solution allows for many options for unit actions. This is the use of tactics, for example, encirclement of a group of enemies, or the use of attacks from the flanks, or retreat with fire cover, or the use of game objects as shelters, and so on. And all this is not a scripted action, but work in dynamics. I can write a long time and a lot on this topic. That is, it can be much more interesting than a crowd of units that just stand and shoot at each other. And for the player, it will be more exciting, because there, behind the screen, combat units that demonstrate intelligence capable of processing commands of a higher level than "go and shoot". Can you imagine what RTS will turn into in this case? P.p.s. Or you know, when you play RTS and in the battle you understand that you are losing, because a couple of units are stuck, or can't take a couple of steps, or turn, and you have to be distracted. Thanks again for your comment. Write to me without hesitation.

Dear @gasstationjunkfoodie. Thanks for the comment, I see you wrote to me in telegram. I really want to implement my concepts in pathfinding, including in a solution for game developers. I like it when development tools become freely available after the release of the game, at least for small and medium developers. But I didn't think that investments in such projects are now at rock bottom. It's sad. I'm still looking for an investor. Even if I immediately implement an open source project, I still need resources and assistants, since releasing a raw or limited in functionality open source product is a waste of time. If you have ideas, criticism or comments, feel free to write to me.

@homosapien5684 Thanks for the questions. I still think my solution is something new. Check out the other videos on the channel, it's not machine learning, it's just the algorithm at work. However, your question raises an interesting point, using my solution with machine learning will give amazing results in solving some high-level problems. At the moment, I hope to continue working on the dynamic pathfinding solution, and start working on solving the route load balancing problem.

@mx.s thank you for replying. I will check out your channel. Subscribed.

@DasAntiNaziBroetchen Yes, of course, agents cannot step on each other. In addition, it is now possible to use various tactics of approaching the target and surrounding. Also, there is a theoretical solution for balancing the load on routes in dynamic conditions. Based on the example of this video, it is likely that the balancing solution would not leave a chance for the player, even taking into account his passage through walls. I hope I will find resources to continue the work.

Me too

@ltkdt @bossjules2785 Thanks for the comments. The code is still in development. The project is looking for a sponsor to continue the work.

raylib is a cool library. I really liked it and it left a pleasant impression

@bossjules2785 Thanks for your comment.

@dandymcgee Thanks for the comment. Green agents try to go along paths clear of hunters (Red) to shelters that are located at a distance from the hunters. When the paths to such shelters are blocked by hunters, the green ones stop, although the behavior can be anything, for example, they can attack or run in panic. A clear path to the shelter is a path along which you can walk at some distance from the hunters. The logic for selecting cover and the hunter's avoidance distance can be customized. The freeze occurs at 0:25, at this moment the exit is blocked by one hunter at the top, and a crowd of hunters at the bottom right. Further, the path to the upper shelters is clear and the green ones begin to move at 0:26, at the moment of 0:27 the green agents freeze again, since the exit is again blocked. At the 0:28 mark, the greens begin to move, as the red one on the lower right has moved a sufficient distance from the exit. In fact, slowdowns were introduced into the actions of the greens, otherwise they would have been out of reach. You can evaluate the speed of the algorithm in the video with the maze. Thanks again for your comment.

Amazing!

Ответил в другом вашем комментарии

@user-cy7yt8gm7z thanks for the comment. Sorry, but I didn't understand the meaning of the question about "fast enough". Can you explain?

@@mx.s I didn't mean much. What I want to say is, how much performance does it have compared to the fastest known "astar" and what are the advantages and disadvantages

@user-cy7yt8gm7z Thanks for the question. If you are talking specifically about the performance of a software solution, then I can answer that at the moment 2 out of 3 planned optimization steps have not even been started. At the same time, you can already preliminarily evaluate dynamic navigation for multiple agents in 3D space as an indicator of the performance of the software solution. The speed of video with 3D navigation has not been changed, the reaction speed of agents in 3D space is quite fast. Well, given the state of development, I'm not ready to talk fully about the advantages and disadvantages, however, for example, you can handle some reasonable non-static set of dynamically moved targets, and you can also use different prioritization logic for each of such targets.

@skycitrice6146 Dear friend. Thanks for the comment. A solution is in development. The concept that I am forming implies many interesting things for developers. Of course there will be a tutorial.

@benjaminweeg9684 Thank you very much for your comment. Give me the video please, it's very interesting.

@@mx.s well, now I have to build a video for it. It's been something I've been meaning to do for a while anyway! 😁 I'll have one thrown together shortly...

@benjaminweeg9684 Wonderful. I'll be waiting. Contact me here or via telegram.

OK, I finished the semester and passed all my classes! Yay. I'll get this together, or at least started, this weekend! What do you recommend for screen capture software? I used zoom share screen and record session for my school projects and it's really poor quality.

@@benjaminweeg9684 Great. There are many programs and they are all mostly good, try movavi or ShareX

I changed the title of the video. Everything is fine now. Please write what features of this solution you would like to see in the release for developers.

@timkuligfreemusic Exactly. Thank you very much for your music.

Thanks a lot

​​@@mx.s Hi, I'm a hs student and have been interested in creating pathfinding algorithms. I've done some research on a*, d*, hpa* and other stuff like that but I think your channel has a really impressive implementation that I haven't seen before. What should I look into since I want to learn how to create a similar implementation to improve my skills?

​@@jellyfish5121Thank you very much, dear friend. I cannot recommend that you study this topic, since there are several features that will demotivate you. For example, the decision to automatically distribute loads in order to avoid congestion is also unnecessary for anyone. Show my videos to your teachers, they will be a little nervous.

Sorry friend. In fact, simply releasing the code to Git is not the same as giving life to the project. Someone will simply take it (the code) to pieces. In addition to releasing my games, my goal is to make a tool for game developers, where pathfinding will be raised to a higher level. If this is open source, then I need to create a community of developers. Any similar open source project has a group of like-minded people. However, today, I cannot even get conversations with experts, for example, about the need to find a path in 3D space in games. For example, are you ready to express your opinion on this issue?

As far as I understand, my computer does not support UE installation, it is too weak. However, the code is initially written in CPP and only then ported. Thus, there is no reason to say that the solution will not work for the UE.

​@@mx.s You should try, for me it much more easy to get better performance with unreal, and if you really end up making a good overall pathfind solution i bet many many people would be interested, i would definitely pay for such solution.

@cnnhean Thanks a lot. I agree that such porting is necessary. It seems that UE can be installed on 1650. Now I'm working on a system for managing movement tactics and agent interaction. There is also a ready-made theory about the automatic distribution of agent congestion on the map; this is when agents, seeing a congestion, choose a less congested path, if it exists.

@cnnhean Give me the timecode of the moment please.

@@mx.s 00:30 Although it doesn't seems like the same algorithm you used in unity. Here it looks like the resulting path from a*, while the path from theta*/anya looks better, and is shorter, due to being able to move in any angles

Thanks a lot

@@mx.s You are a big talent i hope you'd be keep it up brother !

@cnnhean Bro, thank you very much, this is motivating!!!

Thank you very much for the music. Your musical creativity is amazing and inspiring. If you're interested, I can talk about this demo in general detail. Let me know please.

Thank you. There will be several more emulation models

subscribe to the channel to be in the forefront when the solution is released

@@mx.s subscribed :)

​@@jedi_minecraft1​Thank you. I think and hope there will soon be a demo of dynamic agent tactics using various covers from the player's attack, such as tracking, attacking, defending, etc.

I hope it will be soon

This is a new pathfinding solution. I hope we see this in games.

@@mx.s That doesn't answer the question. How did you achieve this? Could you explain or give a link to the code/project if there is one? Thanks

​ @paulblart7378 Thank you for your comment. Sorry, but the project code is still under development. I must achieve my goals, and then the code must be revealed. If you are working on pathfinding projects please contact me.

Not sure if they have their own system, but you can look into algorithms like astar or flow fields

oh it's not a game. This is a pathfinding solution that I hope will be used in games. How do you like NPCs with such intelligence?

@@mx.s oh okay thanks for the message;)

@@mx.s im more than a year late :( as a programmer they really interest me.

@0x9D99 Hello. Thank you for your comment. I really appreciate it. You are not late. Now I am solving organizational issues of project development. I have a developed concept, several implementation plans, there is a roadmap. In any case, if you are a programmer, I have questions for you, and I think our conversation can be interesting. I will be glad to receive your answer.

Thank you. Your comment is motivating. Watch the new code optimization video.

Now it's c++. In fact, it doesn't matter what programming language. The solution will be available for all popular languages. Now we hone the logic and go to 3D space. Further, these are addons for UE and Unity.

Thank you. A new video has been released with a large number of agents.