it is very slow when working with buffer's pixel directly in RAM.

Question

Hi, All.It is very slow when working with buffer's pixel directly in RAM.for example:It is fast when implementing Blur by FastBlur function.But it is very slow when implementing Blur by buffer's pixel directly with Gaussian Algorithm4 http://blog.ivank.net/fastest-gaussian-blur.html .1. Is there any way to improve the speed when I use buffer's pixel directly in RAM?2. I created some functions that seems like "composite_rect" and "blend" by using buffer's pixel directly.But it is very slow than when using default composite_rect or blend function.I am not sure why it is very slow. do they use any different way to implement the effects or do they use GPU for calculation speed?Regards,Anthonie.

shachar carmi · Answer

generally speaking, working with direct RAM access to pixels should be very

fast. it all depends on your algorithm and architecture.

is your algo multi-threading? if not, then you're probably going at 1/4th

to 1/16th of what your machine is capable of.

is the access to each pixel wasteful? i.e, are you calculating the ram

address of each pixel from scratch, or are you getting the first pixel in

each row and then bumping the pixel pointer?

beyond that, some stuff is done faster on GPU and perhaps AE does some of

the stuff you're comparing your algo to on GPU. but it's most likely just

an optimization issue.

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.

Scanning file for viruses.

This file cannot be downloaded