auto looter discussion
Posted: July 15th, 2009, 5:51 am
Finally got on Darkfall for the first time yesterday. Been waiting for NA-1 launch. Needless to say I have found most of what I needed to get a good start here, swimming, gathering, MM macros, etc. The biggest problem I've been having is looting. Right now I'm just beatin on goblin so its not too big a deal, but I'm lazy and all the click dragging is going to give me carpal tunnel.
I make my living as a software developer (mostly in web-based apps tho), and my interest in the field was actually sparked by macroing for UO back in the day. So now here I am come full circle and want to apply some of my l33t skillz (/sarcasm) to making an auto looter.
I would just like to get input on and discuss how we might approach making a good solid auto looter. What should it be written in? How should it be configured? How fast should it be?
My first thoughts:
I understand wyvernx is working with data taken directly from game memory for his teleporter/speed hack/etc. I have no experience in locating such memory values or anything like that, and I am unsure to what degree something like a loot window (and contained objects) would be exposed. I am thinking that any information we could access would be fairly unusable since any action we take needs to be in regard to A) the open loot window, B) the looters backpack. I doubt there is any way to take values from memory and convert them into screen coordinates which we could use for click dragging.
My first idea is basically a brute force method. I have not done much work with AutoIt so I have no idea what the speed implications would be of this approach. Basically, you would exhaustively check pixel colors of the loot window, looking for any that differ from the background of the window. I doubt you would need to check every pixel, possibly just 1 in 5, or 1 in 10 to still get good results (due to the size and orientation of most items i've seen in backpacks).
So if there is no real way to use a custom type app to make use of memory values, I think AutoIt is probably the next best approach. I believe its faster than AHK and seems like a more powerful platform in general.
I would like to hear others thoughts on this. I am open to other ideas because this is something I'd really like to put some time into and I want it to be useful for everyone.
EDIT:
okay a bit of preliminary testing makes think that this is feasible. I put together a little script to pixel scan an area and count all white pixels and count all other colored pixels. I ran it over ~500x500 pixel area, roughly 250k pixels. I ran the loop such that it scanned every other row and and every other column. It came out to about 66k pixels that were scanned and on my 2.0Ghz Core2 ran in 745ms. Next I'm gonna test dragging. Right now im testing just using windows explorer windows since I dont have access to DF right now. When I get something close to working I'll post it up.
I make my living as a software developer (mostly in web-based apps tho), and my interest in the field was actually sparked by macroing for UO back in the day. So now here I am come full circle and want to apply some of my l33t skillz (/sarcasm) to making an auto looter.
I would just like to get input on and discuss how we might approach making a good solid auto looter. What should it be written in? How should it be configured? How fast should it be?
My first thoughts:
I understand wyvernx is working with data taken directly from game memory for his teleporter/speed hack/etc. I have no experience in locating such memory values or anything like that, and I am unsure to what degree something like a loot window (and contained objects) would be exposed. I am thinking that any information we could access would be fairly unusable since any action we take needs to be in regard to A) the open loot window, B) the looters backpack. I doubt there is any way to take values from memory and convert them into screen coordinates which we could use for click dragging.
My first idea is basically a brute force method. I have not done much work with AutoIt so I have no idea what the speed implications would be of this approach. Basically, you would exhaustively check pixel colors of the loot window, looking for any that differ from the background of the window. I doubt you would need to check every pixel, possibly just 1 in 5, or 1 in 10 to still get good results (due to the size and orientation of most items i've seen in backpacks).
So if there is no real way to use a custom type app to make use of memory values, I think AutoIt is probably the next best approach. I believe its faster than AHK and seems like a more powerful platform in general.
I would like to hear others thoughts on this. I am open to other ideas because this is something I'd really like to put some time into and I want it to be useful for everyone.
EDIT:
okay a bit of preliminary testing makes think that this is feasible. I put together a little script to pixel scan an area and count all white pixels and count all other colored pixels. I ran it over ~500x500 pixel area, roughly 250k pixels. I ran the loop such that it scanned every other row and and every other column. It came out to about 66k pixels that were scanned and on my 2.0Ghz Core2 ran in 745ms. Next I'm gonna test dragging. Right now im testing just using windows explorer windows since I dont have access to DF right now. When I get something close to working I'll post it up.