It is generally possible to set this seed in programming languages in order to reproduce and check results that are based on RNG. If one knows the starting seed for a sequence of random numbers, it is then possible to reproduce this full sequence of "random numbers" resulting from this seed and everything become deterministic. On the note of non-deterministic, I know that all RNG in computing are actually pseudo-random in that they are generated through a random number seed. This more crude version of a full replay can be first implemented and be replaced with an accurate version of the replay at a later stage (if it can be done, that is). The worst case scenario one will see in the replay is some visual inaccuracies such as some bullets hit targets but does not result in any damage or damage being done by missed bullet. These are stored based on the results and timing during each player turn and just reapplied to each action in the replay irrespective whether a bullet hits or not. On the other hand, the results of the actions, such as damages done to mechs and buildings, are not re-simulated. ![]() The actions are still re-simulated by the physics engine, such as movements and bullet trajectories, but having these actions having no effect on the environment and mechs. Yes, I understand the non-deterministic nature of the physics engine, that's why I suggested doing the replay in a way where action and result is decoupled. ![]() This takes up memory and the more turns you save the more memory it takes to replay everything back. As such the devs basically are taking the state of the game almost every frame and saving it in memory. It basically plays out in real time and can't really be played back by feeding the inputs back into the system. If you've ever been in a strategic MP game and you get 'out of sync' with other players, what it means is that a bunch of players replayed the steps but for some reason your results are not the same as other players.įor this game the physics model isn't deterministic. I say 'should' because even in a theoretically deterministic model, you can have bugs, edge cases etc that can screw up the simulation. In a deterministic model you can basically queue up all the actions, and replay them and it 'should' replay everything exactly as it was intended. This will be way less memory intensive and I don't think many will take notice to the small discrepancies such as a few bullets not connecting yet are still doing damage in the replay. The result of each action, such as the damage done and hits taken, however are not re-simulated but updated according to what's stored in memory. On the topic of full mission replay, Is it possible to store just the orders and their results, such as the damage done to mechs and building each turn? If so then a full replay can be done in a semi-accurate way in that the actions of the units are re-simulated according to the orders they were given each turn. Originally posted by BigGreyWolf:Fantastic game, I am currently playing it through Epic store. I have already seen some of such comments in reponse to the question. ![]() Even your fans are going to use that as "proof" that BYG is "working on it and it will definitely be coming in a future update, so buy now!". If you are not ready to commit(to work on it or to give up on it permanently), then please don't tease us with such non-answers. I notice a lot of such "dodgy" answers to tough questions. It's important to clarify that point to avoid leading others on on false hope because "currently" seems to imply that there are plans to bring it in a future update. Even if the question is "is there?", it still behooves you to clarify as to whether there are plans in the works for it because that is the question that will invariably follow. So the PERTINENT question is - "Is full mission replay BEING WORKED ON for a FUTURE patch?". Yes, we know that CURRENTLY there's no full mission replay. ![]() The question is "will there be?" NOT "is there?". Pertaining to the "Will there be full mission replay?".
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |