as a developer of industrial vision applications, I often get pretty awkward images like 6000x4000 pixels or more.
While the camera and image processing work with a constant stream of new images (and this processing is the main task), I would like to allow the user to comfortably view another image in parallel.
Doing this on a processor (GDI, etc.) leads to excessive performance. For example, we need 0.2 seconds to analyze the image, but 0.8 seconds to display it with one zoom (resized to fit some control), not to mention letting the user move on and dive into it.
Since Photoshop allows very fast memory and processing to be shown and scaled using a graphics card, I wondered if anyone could give me an idea if and how I can experiment with this in my own code: click on the data on the graphics card (how How long can it take for my 76 MB of rgb data?), and let it be displayed in some control without much effort to scale and move it by the user to the control / map to interact with the user it.
There is no need for 3D viewing, just moving and resizing in a 2D rgb image. The goal is to provide fast and convenient viewing with low processor load.
==> Is it possible (like texture or something like that)? ==> Are there any limitations with current low-level 3D cards> = 256 MB? ==> can someone suggest some expectations (copy data, scaling)?
claus
source
share