Back by - Paul Hamilton & Dietrich Schultz Things that remain static should be in the background
layer (layer 0), everyting else should b in a layer higher if it
moves or displays momentarily. Do a lot of blank layers slow APW down? We want to
ensure the Glossary always appears in front of everything so
we may assign it to layer 9. No. Here is how it works internally. There are 2 offscreen buffers. One is called the "background"
and another is called "animation". All objects in layer 0 are
always drawn into the background buffer, and other layered
objects are in the animation buffer. So if an object is moved in the background buffer, then the
object is updated in the background buffer, then the area that
is changed (called the "inval" region) is drawn onto the
screen, and then that same area in the animation buffer (even
if it hasn't changed) is drawn onto the screen on top. so
everytime you change something in the background buffer
you will redraw the animation buffer too. If an object in the animation buffer changes, we only need to
copy the image from the background buffer that is exposed
and then draw the rest from the animation buffer. So why don't you just put everything into animation buffer? Imagine you have 20 overlapping objects in the same buffer.
everytime you want to move one in front, all objects behind it
need to be redrawn individually. If they are in a separate
buffer, then we just drawn the bitmap of what they look like
rather than asking them to redraw (which could take a while if
they are objects). I hope this helps a little bit. Paul.
-----------------------------------------------------
Paul Hamilton
Authorware Development Engineer
Macromedia
FW: Layers and screen redraw: Further clarification Segmenting a bitmap can be a benefit regardless of which buffer it's in.
I'll try to describe a simple test you can do to see what segmenting can
buy you. I encourage you to try this out for yourself. Load a full screen, 256-color bitmap into a display icon. Put a decision
after the display and set it to repeat 100 times. Attach a display to the
decision that contains a small rectangle in the middle of the screen (say
50 pixels high, or so). The idea here to to have a small object display
AND ERASE 100 times. Run this simple file and time how long it takes for
the decision loop to complete. Next, take the bitmap into Paintbrush and slice it into three horizontal
pieces. The middle piece should be at least as high as the square you
drew previously but not too much higher or the benefit of segmenting is
diminished. Now replace the display icon containing the single bitmap
with one containing the same bitmap sliced into three pieces. You have
to ensure that the foreground rectangle will overlap the middle slice of
the bitmap, but not the other two slices. Run the file now and it should be
considerably faster. On my machine the times were about 5:1. You can
take this one more step and break the middle segment into three vertical
pieces, with the center piece just big enough to sit under the foreground
rectangle. The actual benefit of this technique will vary based on your machine, the
relative size of the images, etc but should almost always provide some
improvement. An important point here is that the "foreground" object was also
being
drawn at layer 0, ie. into the background buffer. That meant that every
time the foreground object was erased Authorware had to restore that
portion of the background buffer that was affected. This redraw
operation is more efficient in the segmented case because a smaller
bitmap was involved. The redraw also requires less memory since a
smaller bitmap is loaded in the segmented case--that can make a
difference if memory is tight. You can get this sample to run even faster, with or without segmenting
the background, by moving the foreground object to the animation buffer
(ie. set it's layer to 1 or more). If you do that it doesn't matter whether
the
background is segmented or not because Authorware doesn't have to
reload the original bitmap to redraw anything. The foreground object was
never in the background buffer so erasing it doesn't disturb the
background pixels. Rules of thumb:
- This only matters if your background is relative slow to draw (a large
bitmap, complex object graphics, or text).
- Use layering when the foreground object will be displayed and erased
quickly.
- Segmenting a bitmap speeds up ERASING objects that sit on top of it
(and in the same buffer). It doesn't speed up the original display of the
bitmap. In fact, it may slow it down a little.
- It's usually best to break a bitmap into horizontal segments. Vertical
segments can work, too. In extreme cases you could break it into vertical
and horizonal pieces.
- This is all pretty tweaky and is best left towards the end of a project
when the screen layout isn't going to change. Also, see my answers to your specific questions below. Dietrich Schultz
Macromedia How do segmenting graphics and using different layers relate (if at
all)
to each other? They're different techniques and don't relate directly. We have our background at layer 0, and animations at layer 1. Does
segmenting the background have any effect on when Authorware
copies "the image from the background buffer that is exposed."
Is the
redrawn area just the exposed part of the graphic, or does this refer to
the whole graphic, as #3026 seems to imply? I'm not sure exactly which case you're referring to as I can't find that
phrase in the copy of TN3026 that's currently on our web site. However,
segmenting will only provide a performance benefit if the object being
erased is in the same buffer as the one behind it. If a large graphic is placed on the screen, another smaller graphic
placed over it, then a third smaller graphic placed over just the second
graphic, which graphics must be redrawn? Does using different
layers effect which graphics are redrawn? Good question. Assuming all the objects are in the same buffer and all
are rectangular and opaque and you're erasing just the top one, then
Authorware has to redraw just the middle one. In general, Authorware
recognizes opaque objects and doesn't needlessly redraw objects
behind them. My assumption (and resulting action) from Paul's posting was to place
graphics that stay on the screen for awhile at layer 0, and temporary
graphics at layer 1 or higher. TechNote #3026 seems to imply that by
not also segmenting the background, everytime a graphic in layer 1 or
higher changes, the whole background must be redrawn. This seems
to go against the benefit of two buffers, and the reason for my
confusion. Your actions are good. Don't bother segmenting the background if you're
using layering. Thanks for any insight and clarification on how the two statements above
effect each other. Clear as mud?
11015 - Authorware layers and graphic performance
Article appears courtesy of Danny Engelman’s Gems and Jewels of the AWARE list
Does placing displays in multiple layers slow APW down
or speed it up?
Back
Downloads
| Events | FAQs | Photos
| Humor | Articles | Links
| Search | Home