• Which the release of FS2020 we see an explosition of activity on the forun and of course we are very happy to see this. But having all questions about FS2020 in one forum becomes a bit messy. So therefore we would like to ask you all to use the following guidelines when posting your questions:

    • Tag FS2020 specific questions with the MSFS2020 tag.
    • Questions about making 3D assets can be posted in the 3D asset design forum. Either post them in the subforum of the modelling tool you use or in the general forum if they are general.
    • Questions about aircraft design can be posted in the Aircraft design forum
    • Questions about airport design can be posted in the FS2020 airport design forum. Once airport development tools have been updated for FS2020 you can post tool speciifc questions in the subforums of those tools as well of course.
    • Questions about terrain design can be posted in the FS2020 terrain design forum.
    • Questions about SimConnect can be posted in the SimConnect forum.

    Any other question that is not specific to an aspect of development or tool can be posted in the General chat forum.

    By following these guidelines we make sure that the forums remain easy to read for everybody and also that the right people can find your post to answer it.

P3D v5 Run multiple instances of scenPROC on same PC?

"Are you sure your imagery has a NIR channel included"

Please excuse my ignorance but I am not sure. As far as I thought NIR, meaning near infrared, was to mean the image does not have an true infrared channel? I thought your NDVI compensates for not having an IR channel. Am I way off on this assumption? When I pull up the image in Photoshop I see 3-bands (RGB). The imagery is from the USGS NAIP data. I also was able to download a separate IR image of the same area but when I placed it into TFE, it seemed there was no filter/processing needed as it was already infrared?

If my image does not have NIR, how is TFE able to produce a preview of it with the NDVI step? I am more than happy to provide you with the image+ samples+IR image if that would help? Since the image is large (1GB), I would post it on a cloud for you or upload to your cloud if that is more convenient. I have pretty fast upload speeds.
 
Hi,

When I use the NDVI step in my imagery I load 4 band images (red/green/blue/NIR), this is indeed NAIP data as well. But if you load only RGB data the NDVI step will not work, the NDVI index requires a band with infrared data to give useful output.

When you write you loaded infrared images, that is probably colored infrared images, these are a bit harder to process, but it might work if you know what each band is.

If you have only RGB data you can better use the BackProject step instead of NDVI.

If it helps I can take a look at the data you have.
 
I think it could be a good exercise to check the data and see if that has any impact with the way I am using the TFE. I have the data uploaded, I just need your email address to get you access. Not sure if that is public or if you want to PM me, or email me with it?
 
Hi,

I had a look at the data. The normal imagery is indeed only 3 bands (red, green, blue). If you use the NDVI texture filter from the manual on that you will never get good results. You will get many false detections. You can use the colored infrared image though. This images contains a near infrared, red and blue channel (in that order). So you need to change the NDVI step attributes. The NirChannel needs a value of 0 and the RedChannel a value of 1. Then you will see that the vegetation detection is uch more accurate. Be aware it will also select the crops, not only the trees. The screenshot below shows a merge of the results of the filter on a piece of your imagery using the CIR version.

Capture.PNG
 
That is amazingly accurate. But I am also surprised at how well the normal image produced vegetation. Maybe that is why some images to so long to detect. I think my next step is to process one of each type. See if using the infrared image reduces processing time but allow keep PROCESHOLES intact. Tell me, did you need to create sample 1000x1000px using the infrared? Thanks for the feedback.
 
I just made one sample of roughly 1000 x 1000 pixels to show how the filter works with the CIR image. I did not run it over the entire image, but I'm sure it will work.
 
I had surprising results testing the regular RGB image against its infrared counterpoint. Actually, the opposite from what I was expecting.
I created an infrared TFC using the 1 & 0 values you suggested. The RGB image produced a more "fuller" area of vegetation (188,203 veg detections), while the Infrared was sparse in reproduction (7378 veg detections). Both tests were processing holes. RGB took 14 seconds while the infrared took 2 minutes.

Another result which surprised me was the the IR image tended to filter out all vegetation on top of crop areas?!? I had no filter steps running either. Top image is RGB results while lower is using the infrared image and TFC.

RGB results.jpg



Infrared results.jpg
 
Hi,

Looking just at these pictures it is hard to say what happens. What kind of threshold value did you use in the threshold step for example? How much extrude and dilate did you apply to the output? This can all effect how much vegetation you get in the end. You will need to study the sample images in the texture filter editor to tune your filter to work well on your imagery. There is no magic setting that works on any imagery.
 
Gotcha. Basically I was still using the basic TFC in the manual and only changed the NDVI settings. Did not touch any other steps. As you say, I need to mess around with the settings and experiment more. As a suggestion, it would be cool in Threshold binary (maybe NDVI too?), had a slider for the value settings rather than having to type in the values. That way you could see the changes happening as you slid one way or the other. Just a thought as I mess with it. :)
 
Hi,

The sample image I posted above was also generated with the settings from the manual, except for the band settings of the NDVI step. But I did not make actual autogen and check how it looks in the sim.
 
You must log in or register to reply here.
Back
Top