可以把【甩钩、收杆人为操作】，【自动拉条】分离吗

[Sunyl0]

甩钩乱甩，瞄准半天，收杆过早； 而和鱼拔河的时候，拉条效果很好； 希望能把抛竿和拉条功能分离， 可以人为的抛竿、收杆，提高效率

[mwmi]

甩钩、收杆有些智障, 改一下 fish.py 中的代码就可以了

def start_fishing(predictor, agent, bite_timeout=20):
    ff = FishFind(predictor)
    env = Fishing(delay=0.1, max_step=10000, show_det=True)

    # winsound.Beep(500, 500)
    # keyboard.wait('r')

    while True:
        winsound.Beep(500, 500)
        keyboard.wait('r')
        # ff.do_fish()
        # winsound.Beep(700, 500)
        # times = 0
        # while True:
        #     if env.is_bite():
        #         break
        #     time.sleep(0.5)
        #     times += 1
        #     if times > bite_timeout:
        #         env.drag()
        #         time.sleep(3)
        #         ff.do_fish(fish_init=False)
        #         times = 0
        # winsound.Beep(900, 500)
        # env.drag()
        # time.sleep(1)

        state = env.reset()
        for i in range(env.max_step):
            state = torch.FloatTensor(state).unsqueeze(0)
            action = agent(state)
            action = torch.argmax(action, dim=1).numpy()
            state, reward, done = env.step(action)
            if done:
                break
        time.sleep(3)

[Vurks]

同请求， 而且改了貌似没什么用

[qilishidai]

甩钩、收杆有些智障, 改一下 fish.py 中的代码就可以了

def start_fishing(predictor, agent, bite_timeout=20):
    ff = FishFind(predictor)
    env = Fishing(delay=0.1, max_step=10000, show_det=True)

    # winsound.Beep(500, 500)
    # keyboard.wait('r')

    while True:
        winsound.Beep(500, 500)
        keyboard.wait('r')
        # ff.do_fish()
        # winsound.Beep(700, 500)
        # times = 0
        # while True:
        #     if env.is_bite():
        #         break
        #     time.sleep(0.5)
        #     times += 1
        #     if times > bite_timeout:
        #         env.drag()
        #         time.sleep(3)
        #         ff.do_fish(fish_init=False)
        #         times = 0
        # winsound.Beep(900, 500)
        # env.drag()
        # time.sleep(1)

        state = env.reset()
        for i in range(env.max_step):
            state = torch.FloatTensor(state).unsqueeze(0)
            action = agent(state)
            action = torch.argmax(action, dim=1).numpy()
            state, reward, done = env.step(action)
            if done:
                break
        time.sleep(3)

确实是，这样子效率还是可以的，不错

手动抛竿，然后拉杆，出现拉力条，正常拉一下，按一下r就自动了

[qilishidai]

要正常拉力，在中间

[Lycan1003]

@qilishidai 优化了一下您的修改，注释原有的def start_fishing函数，复制以下代码替换就ok了 1、现在只需要抛竿后，按r就行了，不需要等待鱼上钩，剩下的步骤可以自动进行。钓上后等待系统蜂鸣器响后再次抛竿按r，重复下去即可实现半自动钓鱼。 2、一般来说，上钩后200个step内一定能钓上，所以直接设置max_step=200 3、由于是手动抛竿，所以就不用考虑抛竿位置不对，鱼长时间不上钩的问题，所以调整bite_timeout=300 4、如果遇到IndexError: list index out of range的bug，可以参考https://github.com/7eu7d7/genshin_auto_fish/pull/145/commits/3e24afadd7dd60be19f58399eb027a47644fcd41 进行修复

def start_fishing(predictor, agent, bite_timeout=300):
    ff = FishFind(predictor)
    env = Fishing(delay=0.1, max_step=200, show_det=False)

    # winsound.Beep(500, 500)
    # keyboard.wait('r')

    while True:
        print("start fishing")
        winsound.Beep(500, 500)
        keyboard.wait('r')
        times=0
        while True:
            if env.is_bite():
                break
            time.sleep(0.5)
            times+=1
            if times>bite_timeout:
                env.drag()
                time.sleep(3)
                do_fish(fish_init=False)
                times=0
        winsound.Beep(900, 500)
        env.drag()
        time.sleep(1)
        state = env.reset()
        for i in range(env.max_step):
            state = torch.FloatTensor(state).unsqueeze(0)
            action = agent(state)
            action = torch.argmax(action, dim=1).numpy()
            state, reward, done = env.step(action)
            print("epoch:", i)
            if done:
                break
        time.sleep(0.5)
        print("finished")
'''

[sandboxdream]

感觉不自动开始拉条有点不智能。 而且判断是否咬钩还是挺准确的 不过我找了半天也没有吧判断咬钩的函数分离出来.......