Kind of looks like the top view of a Dwarf MulfiCannon, but I know it isn't one though :p
http://images2.wikia.nocookie.net/ru...warfcannon.gif
Printable View
Kind of looks like the top view of a Dwarf MulfiCannon, but I know it isn't one though :p
http://images2.wikia.nocookie.net/ru...warfcannon.gif
Damn nice. Did you find that NPC by searching all models for the model of the guard, then get the vertex co-ords, or just search for a random thing? Just wondering if you've gotten a "matchmaking" thing done yet.
And how will this be implemented into a bot? I'm guessing use some sort of custom client to parse the info, then pass these vars on to scar/simba which will do the script logic/clicking etc? If so, would it be able to be run minimised, like smart (which is now open source) can? Would it be possible to merge the two, so we have multibotting with traditional reflection, GPL reflection, and minimisable client all in the one thing?
And about drawing onto the screen: Didnt you say jagex could take a screenie of it and see? SMART uses a transparent window pane above the rs screen which we can debug on to get around this.
Very nice job finding the coords :)
Out of interest, how are you transferring information to scar? I need to find an efficient way of doing this... Windows messages? Also, I'm really interested in how you hooked RS. What/how were you hooking? I find this stuff really interesting :)
Yes, this will be a feature of the final release.
Hopefully this will simply provide scar with some extra functions, e.g. GetNPCCoords(X, Y, GuardID). This will mean no client modification. It will run with the standard RS client, or any modified client you have. It will also be able to run in minimised mode. We'll see ;)Quote:
And how will this be implemented into a bot? I'm guessing use some sort of custom client to parse the info, then pass these vars on to scar/simba which will do the script logic/clicking etc? If so, would it be able to be run minimised, like smart (which is now open source) can? Would it be possible to merge the two, so we have multibotting with traditional reflection, GPL reflection, and minimisable client all in the one thing?
You can haz sauce http://blanddns.no-ip.org:81/repos/public-smart/
This SMART stuff looks awesome. I will study when sober.
Here si a render off all NPC's detected and coords calculated:
http://i45.tinypic.com/2n1ar6t.png
NPC regonition cming tomorrow :)
Edit: MylesMadness: def remeber you from rscheatnet, right?
http://i46.tinypic.com/1t7i4l.png
if you mean rscheata I have posted there, but never scripted or got into it deeply
Hmm.. was a log time ago. Probably my bad memory :unsure:
Although not really related... first things you posted were cave walls.
Yeah, I'm not sure. There are a few animated models which aren't getting picked up... I will look into this later. This is still very much just a proof of concept :)
I see :)
Well.. Good job :D
Are you going to make it aim at the middle of the npc, and not at the first vertex?
How often does this break compared to the current form of reflection?
This will never break, unless they change their methods of rendering or change their models. A GetCoords(id) function would return the centre of the model :)
Here's a proof of concept ID system:
http://i48.tinypic.com/a2xlzc.jpg
Impressive. :)
I assume you're able to make an object recognition too? like trees, rocks and so on.
No breaking, perfect NPC (and object?) finding, no need to get colors of everything and (maybe) walking (somewhere in the future?). This sounds like the perfect system for macros... Only downside must be the need for HD = more CPU power needed
Yup, objects should be fine too. I will look into reading the map at a later date, but any implementation probably won't be as easy to use as hacked java clients. Probably something like passing scar the unrotated, clean map and your current x,y coords on said map. I'm not sure yet.
Running the game in HD will also reduce your CPU usage. HD simply means RS uses opengl, which passes large amount of processing to your GPU, so your CPU doesn't have to do it. Software rendering is slow.
Edit: well out of interest, I tested this and CPU useage seems pretty much the same in both, in low detail :o
Do those numbers represent anything specific or are they arbitrary?
How did you get those ID numbers?
And I just realized what's so great about this method too compared to reflection:
One single finding function for all grounditems/NPCs/Objects :)
E: And btw, are 2D models processed with OpenGL too, such as minimap and inventory items, or does the game-engine handle them separately? Cause if they aren't processed with OGL then you can't use this system to mapwalk.
Out of interest, have you been experimenting long enough to confirm they do not change their models/often or is it just a guess that they don't? As I'm pretty in the dark about this area, could they not alter the id's without changing the actual models used, a bit like with hooks for reflection?