|
Multiple Processing
san from India [5 posts] |
15 year
|
Hello.
I am doing a project with 2 separate camera. Using Mosiac function I can view images, but I need to do separate image processing for these 2 cameras. In Mosiac the image processing is common for both.
I want to know is there any possibilities to do that in RR ?
|
|
|
|
Anonymous |
15 year
|
Yes, do NOT use the Mosaic module. Get one image to show in RR, process as needed and then use the Set_Marker module to save the current processed image and revert to another image/webcam available in the dropdown in that GUI interface. Keep in mind that RR keeps many copies of an image in memory but you only see the current one that you are working on. Switching from one image to another using the Marker module will provide the functionality that you are looking for.
http://www.roborealm.com/help/Marker.php
STeven.
|
|
 |
Hibou from France [8 posts] |
15 year
|
STeven, it would be good to give an explicit example of this script that I can upload. I still can't quite get the following to work :-)
1) Open an image (or camera), rotate and scale it arbitrarily, translate it by its width setting "Change image size as needed". Then set a marker "image1"
2) Open a second image (or camera) rotate and scale it arbitrarily, no need to translate it
3) Now display the two images as pasted onto the larger canvas size.
I get this to work if I only want to rotate and scale the first image, but can't quite see it if I also need to rotate and scale the second. There is something no doubt trivial that I am missing.
|
|
|
|
Anonymous |
15 year
|
Hibou,
That seems about right. Attached is a robofile that does that. Note that you will need to update the load image path's to something of your liking before you will see that work.
Note that while we use scale and rotate modules inbetween the image composition stuff you can use any modules you need in those areas. The results should be similar.
Note that the major difference is that we translate the second image to make room for the first. Seems to work better that way round.
STeven.
 program.robo
|
|
|
|
Hibou from France [8 posts] |
15 year
|
That's works perfectly STeven. Thanks. But when I try to apply it to two cameras instead of image files it doesn't quite work- file attached (with RoboRealm 2.0.9.7) I open the first camera under Options, do some transformations then put in a camera#1 marker reverting to the second camera, do some more transformations on that and translate it to make space for the first, before finally adding them together with the Math module. Individual statements seems to work correctly, except for the final Math module, which gives a bizarre result, with an image divided into 4 parts.
Maybe I have a problem because my cameras are 720x576 at 25 fps (PAL), but even if I halve the video output I still have a problem (apparently I can't reduce the frame rate, even though I would happily accept a lower rate).
If I save separate images from the two cameras, I can stitch them together after similarly transforming them using STeven's robofile, so why doesn't it work with the camera input ?
program.robo
|
|
|
|
Anonymous |
15 year
|
Hibou,
There were some issues with using two markers in the Math module instead of using the "Current" and a single marker. We've corrected this problem in v 2.11.5. Can you download and try again with this new version?
Thanks,
STeven.
|
|
This Thread