Any idea?

[(s)okkr7]

My ape atoll agility script current is using a DTM for the tree that needs clicked on to glide down to the bottom (if anyone know what I mean by that..), and it fails a lot, finding the DTM way north, etc. Any suggestions/comments/help? The script is done besides that part, thanks in advance!

[thomas1116]

well im not sure, ask one of the mods or the pro's one :D

[riwu]

Did you use DTMrotated? Also DTMs are not very reliable for finding onscreen objects, try TPA.

[litoris]

Try TPAs, if it doesn't have a unique color, try comparing two TPAs of different colors that are both on the object to find close/overlapping points and divide up the resulting TPA into an ATPA. Beginner5 had a function for that. Runaway also had one that looks for a cluster of points, so if it has lots of colors on it, you could try that.

[VastlySuperior]

You could also try to bring down the tolerance on your dtm and see if it gets less hits for the object you are looking for, also look if something is unique near the tree and try to include that in your dtm to find the tree, Uniqueness it the key with dtm`s and as low as possible tolerance on the dtm you want to find.

Otherwise, go with the tpa which will be even more accurate.

[riwu]

You could also try to bring down the tolerance on your dtm and see if it gets less hits for the object you are looking for, also look if something is unique near the tree and try to include that in your dtm to find the tree, Uniqueness it the key with dtm`s and as low as possible tolerance on the dtm you want to find.

Otherwise, go with the tpa which will be even more accurate.
It's not totally correct to say that TPA finding is more accurate than DTM. DTM is in fact a more accurate form of TPA that tries to match both the color and the distance between the main/sub-points. Right because of the demanding accuracy it may not be reliable to find objects on mainscreen (not because it returns false positive, but rather it cant find any because of the changing color/deformed objects!)

And of course, another reason i prefer TPA is that DTMs take too much time to create and hog memory (unless you recall and free them after every use)

[VastlySuperior]

It's not totally correct to say that TPA finding is more accurate than DTM. DTM is in fact a more accurate form of TPA that tries to match both the color and the distance between the main/sub-points. Right because of the demanding accuracy it may not be reliable to find objects on mainscreen (not because it returns false positive, but rather it cant find any because of the changing color/deformed objects!)

And of course, another reason i prefer TPA is that DTMs take too much time to create and hog memory (unless you recall and free them after every use)

I stand corrected here about the accuracy, but the rest is all good advice for the dtm`s, and my second script I used dtm`s exclusively and that thing hogs memory like mad, but it is well accurate on what it is to find, just look out for memory leaks, release the dtm`s after each use. I can only run 2 instances of that script and sometimes my pc comes to a standstill because it runs the most times at 80% cpu capacity as I only have a 3ghz duelcore.

[(s)okkr7]

Did you use DTMrotated? Also DTMs are not very reliable for finding onscreen objects, try TPA.

Oh yes sorry, I was using DTM rotated.


Try TPAs, if it doesn't have a unique color, try comparing two TPAs of different colors that are both on the object to find close/overlapping points and divide up the resulting TPA into an ATPA. Beginner5 had a function for that. Runaway also had one that looks for a cluster of points, so if it has lots of colors on it, you could try that.

No unique color, as everything is the same shade of green, and same shade of tan (tree trunk). I'll look into ATPA, thanks :)