On 27/02/2005, at 4:51 AM, GSR - FR wrote:

[EMAIL PROTECTED] (2005-02-26 at 1805.22 +1100):
In order to get past any background processing photoshop may have been
doing, I also tried adjusting the curve in the curves dialog then
quickly hit OK - this came out taking the same time as if I moved the
curve and waited 10 seconds before clicking OK. Closing the image
immediately afterwards was also instant, so there seemed to be no
tricky background processing happening over the whole image area.

For anybody trying to guess about background tricks, best would be a system monitor and see the CPU load. If app says it has ended, but the monitor reports the app is eating near 100%, there are some kind of trick there, or interface oriented optimization, it depends the point of view. So fire up top, gkrellm, activity monitor, task manager or whatever you have in the OS and see what is really going on.

Aha! - That has revealed it I think - from the results it seems there may be background completion of each task, and perhaps some form of queuing as well.


Doing one Levels change across the image in Photoshop came out as in my earlier post - with control coming back to me within a second - but top showed CPU use was above 70% for another 4 seconds before it dropped down to 1-2% idle again.

Doing eight operations quickly, one after another - a combination of adjusting curves, levels and color balancing - left the cpu use peaking up high for about 16 seconds after finishing the last op. I'd expect to see that if each operation was being queued. Whatever it's doing in the background is transparent enough that zooming around and drawing over the image works OK as it's happening - prioritising what needs to be quick & interactive, and playing catchup with the rest a bit later whenever the user pauses.

I imagine many of the benefits of an interface optimisation like that would be lost for heavily scripted processes involving those colour based functions ( levels, color balance, curves, brightness/contrast, posterize etc).

dana
--
http://www.danaquarium.com/

_______________________________________________
Gimp-user mailing list
Gimp-user@lists.xcf.berkeley.edu
http://lists.xcf.berkeley.edu/mailman/listinfo/gimp-user

Reply via email to