这节我们研究下pygame的几种碰撞检测模式:
如上图,左侧是默认的检测模式:基于矩形的检测(这也是性能最好的模式), 右侧是基于圆形的检测(性能略差于矩形检测)。
矩形检测法虽然性能好,但是缺点也很明显:检测不准确,上图中"飞机与目标"从视觉上看,根本没碰到。
为了改进,pygame给这二种模式,新增了xxx_ratio的方法,允许指定检测时,指定二个目标的叠加程度,只有达到指定的叠加值,才认为是真正碰撞到了,参考下图:
注:ratio的值越大,表示叠加的部分越少!
我们写一个小程序来测试一下:
1 import pygame 2 from os import path 3 from xml.dom.minidom import parse 4 5 SIZE = (WIDTH, HEIGHT) = (265, 320) 6 FPS = 45 7 8 BLACK = 0, 0, 0 9 WHITE = 255, 255, 255 10 11 pygame.init() 12 pygame.mixer.init() 13 14 screen = pygame.display.set_mode(SIZE) 15 pygame.display.set_caption("My Game") 16 clock = pygame.time.Clock() 17 18 spritesheet_image_file_name = path.join(path.dirname(__file__), "../img/spritesheet_jumper.png") 19 spritesheet_xml_file_name = path.join(path.dirname(__file__), "../img/spritesheet_jumper.xml") 20 spritesheet_image = pygame.image.load(spritesheet_image_file_name) 21 spritesheet_image.set_colorkey(BLACK) 22 spritesheet_dom_tree = parse(spritesheet_xml_file_name) 23 root_textures = spritesheet_dom_tree.documentElement 24 sub_textures = root_textures.getElementsByTagName("SubTexture") 25 dic_image = {} 26 27 28 def get_image_rect(img_name): 29 if dic_image.get(img_name): 30 return dic_image[img_name] 31 for texture in sub_textures: 32 name = texture.getAttribute("name") 33 if img_name == name: 34 dic_image[img_name] = pygame.Rect( 35 int(texture.getAttribute("x")), 36 int(texture.getAttribute("y")), 37 int(texture.getAttribute("width")), 38 int(texture.getAttribute("height")) 39 ) 40 return dic_image[img_name] 41 42 43 def get_image(img_name): 44 rect = get_image_rect(img_name); 45 image = pygame.Surface((rect.width, rect.height)) 46 image.blit(spritesheet_image, (0, 0), rect) 47 image.set_colorkey(BLACK) 48 return image 49 50 51 class Demo(pygame.sprite.Sprite): 52 def __init__(self, image, pos): 53 pygame.sprite.Sprite.__init__(self) 54 self.image = image 55 self.rect = self.image.get_rect() 56 self.rect.x = pos[0] 57 self.rect.y = pos[1] 58 59 60 running = True 61 pos1 = (20, 5) 62 pos2 = (125, 110) 63 64 demo1 = Demo(get_image("bunny1_walk1.png"), pos1) 65 demo2 = Demo(get_image("bunny2_stand.png"), pos2) 66 67 all_sprites = pygame.sprite.LayeredUpdates() 68 all_sprites.add(demo1, layer=2) 69 all_sprites.add(demo2, layer=1) 70 71 group2 = pygame.sprite.Group() 72 group2.add(demo2) 73 74 while running: 75 clock.tick(FPS) 76 77 for event in pygame.event.get(): 78 if event.type == pygame.QUIT: 79 running = False 80 if event.type == pygame.KEYDOWN: 81 if event.key == pygame.K_LEFT: 82 demo1.rect.x += -3 83 if event.key == pygame.K_RIGHT: 84 demo1.rect.x += 3 85 if event.key == pygame.K_UP: 86 demo1.rect.y += -3 87 if event.key == pygame.K_DOWN: 88 demo1.rect.y += 3 89 90 screen.fill(BLACK) 91 92 all_sprites.draw(screen) 93 94 pygame.draw.rect(screen, (0, 255, 0), demo2.rect, 1) 95 pygame.draw.rect(screen, (255, 0, 0), demo1.rect, 1) 96 97 font = pygame.font.SysFont("Menlo", 25, True) 98 99 # 默认的检测模式(rect)100 if pygame.sprite.spritecollide(demo1, group2, False):101 pos_txt = font.render("hit:true", 1, (255, 255, 128))102 else:103 pos_txt = font.render("hit:false", 1, (255, 255, 128))104 screen.blit(pos_txt, (150, 10))105 106 pygame.display.update()107 108 pygame.quit()
这是默认的Rect检测模式。把100行换成:
# 矩形检测(至少要重叠1-0.7=30%才算发生了碰撞) if pygame.sprite.spritecollide(demo1, group2, False, pygame.sprite.collide_rect_ratio(0.7)):
再看下效果:
继续,换成圆形检测试下:
# 圆形检测(至少要重叠1-0.7=30%才算发生了碰撞) if pygame.sprite.spritecollide(demo1, group2, False, pygame.sprite.collide_circle_ratio(0.7)):
那么,有没有一种方法能做到精确检测呢?当然有,有一种基于mask(在绘图软件中,也称为遮罩或蒙版)的检测方法,类似把背景去掉后,像素级的碰撞检测,当然性能也是最差的。
首先要在Sprite的类上,指定mask:
1 class Demo(pygame.sprite.Sprite):2 def __init__(self, image, pos):3 pygame.sprite.Sprite.__init__(self)4 self.image = image5 self.rect = self.image.get_rect()6 self.rect.x = pos[0]7 self.rect.y = pos[1]8 #指定蒙版9 self.mask = pygame.mask.from_surface(self.image)
然后把检测代码改成:
# 基于mask的检测 if pygame.sprite.spritecollide(demo1, group2, False, pygame.sprite.collide_mask):
好了,利用上面学到的知识,把的遗留问题:"player与enemy的碰撞检测不准确" 解决一下,效果如下:
优化前 | 优化后 |
源码示例: