根据3D Developments John Carmack, 。 “...... 人类感官系统可以发现,视力部分,特别是听力领域的相对延迟很少,但当绝对延迟低于约20毫米秒时,它们一般都是不可接受的。
That is to say, 20ms is a good rule of thumb, even though flicker may be noticeable quicker. So 10ms (150 vs. 160ms) shouldn t be perceptible. However, you re asking about whether the user will notice the difference between 150ms and 160ms. If it s a one time event (as opposed to something blinking at 15 or 16 kHz), I don t see how people could notice the difference.
如果人们仔细看看,就会看到30-40张图像的负荷,而且没有缓冲。 然而,还有许多其他内容正在网页上。
- Network plus OS latency varies far more than 10ms, so your 150ms load time will vary-- even if you re talking to localhost.
- The browser will typically have several connections open at one time to load files more efficiently. So files will seem to load simultaneously or out-of order. If there s any file size to pay attention to, it s the size of one Ethernet packet. If the complete response fits in one packet (typically 1500 bytes, including packet headers), making it smaller shouldn t improve performance.
- The browser may delay displaying images for a few milliseconds to minimize frame buffer (window) re-writes.
- The browser needs to parse the CSS rules for displaying the image. Browsers start loading images before they have completed the CSS parsing, and CSS parsing may take longer than image loading.
- If you are testing animations, your knowledge and anticipation of an event makes it seem slower. Other users won t notice things that are glaringly obvious to you.
长期以来,图像档案的规模很可能是你最担心的事。 如果它很重要,就不依靠霸权主义,而是依靠自己的测试。 所有现代网络浏览器都有巨大的时间工具,尽管谷歌最侧重于相对性。 (尽管如此,在任何时候都可能出现飞跃。)
此外,对不同类型的互联网连接(千里的当地网络、移动电话网络)和不同的浏览器和装置进行测试,
