Pokemon Go setup, how I did it with Counters and how it could be improved

[aaron146]

Hi, first of all I'd like to say that I really like the app. I was fed up of other autoclickers with image and color recognition making my phone crash and with very complicated setups.

I previously had a setup with another autoclicker app with color recognition (down to the specific pixel) which worked until the phone crashed. I'm using PGSharp, a modified version of Pokemon Go, and when my phone crashes and the app is loaded up again, the interface is in a different position so I had to keep moving things around and remaking the tasks, every day. And it would never last more than a few hours before the phone crashed.

Firstly, I'd like to explain what the autoclicker is used for, how it is done with other autoclickers and how I've done it with Smart AutoClicker (and counters). In short, every pokemon has a small chance of being shiny, PGSharp provides a feed/list of pokemon to click on and teleport to them, and it also includes a feed/list of nearby pokemon, with a shiny scanner (little stars in the corner). So once a shiny is found it needs to click on it and catch it, but there is also a timer, and you can only catch pokemon when the timer is at 0, and it's green. If it's not 0 and it's red, then you have to wait until it's 0 and green. There are also a few other popups like weather alert and things that sometimes popup. There's also a situation where the pokemon don't load up sometimes, and it's commonly fixed by entering the battle menu for a few seconds, but that doesn't always fix it, and so it then needs a full app restart.

The simplest setup clicks to a pokemon, clicks where it should be, clicks where other popups might be and at the bottom of the screen to exit out of anything you may have clicked on that isn't a pokemon (a pokestop or gym), swipes to catch, ignores the timer that needs to be 0. The timer's max is 2 hours and it's unlikely to find 2 within 2 hours, but it is possible. This setup relies on timers and clicking too much makes some parts of the app screen unresponsive.

The setup I did with color recognition worked like if-else statements, being able to regularly check if the feed is loaded, if the timer is green, it would only click on pokemon if the shiny star was shown, it would only look for the star if the feed was loaded and if the feed didn't load it would enter the battle screen, and if that didn't fix it, it would restart the app. It still relied on lots of timers, and it constantly crashed.

With Smart AutoClicker I tried doing things like having the first image detection wait for 5 seconds, then the second one detecting the same image but clicking, this led to an infinite wait loop. Seems like it's not in order at all, but rather a priority system that starts from the top everytime it does anything. I used image recognition to detect the same things as with color recognition but the order was almost in reverse and I had to include several images for some of the actions. In order to wait for the feed to load, for example, I installed an app called Floating Counter, this allows you to have several counters on the screen at the same time which count up when clicked and also have a decrease and reset buttons (after long pressing the counter), I got the Smart AutoClicker to click these counters until it showed a specific number, then it took another action when it showed that number (like entering the battle screen), and then later I reset the counters. Now I can loop any task, especially simple waiting tasks that need to take an action if nothing has changed after a specific amount of time.

This would be much easier with 2 app functions. 1 is that an image detection (or not detected) triggers another task (another detection), like nested if statements. That would make it really smart. The second thing that would improve it is including variables within the app, like I am using counters. Similar to this, loops, an action that takes place X amount of times before moving onto another action.

As a bonus, image recognition anywhere on the screen. I wouldn't want it to just click anywhere on the screen, but, if the interface is moved around I'll have to mess with the interface and keep testing the autoclicker until it works, or remake the functions by taking new images. Taking new images is the easier option, however, for the shiny stars, I'll have to spend time to find one in order to take a picture. I can get the image from a screenshot, but then it won't be in the exact right place. Hopefully it never crashes and I won't ever need this. If it could detect anywhere on the screen, even if slower, it could be used to calibrate the previous setup by seeing where things are now.

It isn't clear in the app if the wait time is seconds or milliseconds, I now know it's milliseconds, but this isn't obvious, and I'm still not sure about the click time. I had to set short wait periods after each click to avoid repetitive clicks because it is so fast. However, you might consider adding a wait time to the clicks themselves, as optional. Also, if I need to change the outcome of a function, it's easy enough to add something on the end, but it seems impossible to add something to the beginning or change the order. I know you can change the order of the image recognitions, but not the outcomes of those recognitions. For some reason 0% tolerance didn't work with pixel exact screens, but that's fixed with a few % tolerance. Also, for this game, autoclicking on pokemon works best when zoomed out. There is no option for multi-touch gestures (pinching) so if I did reset the app, I can't zoom out automatically, I'll still have to do that manually, or have Smart AutoClicker start a different autoclicker just to do this task.

Overall, most of these things are not issues, but for it to truly be smart it should have loops and/or variables, and recognitions that trigger other recognition tasks. I see you can end a function with a specific condition, perhaps it could end and start up another function, this wouldn't be nested, but it would allow for tasks to flow in a specific order. The multi-touch gestures are also a must for many use-cases. Also, the simple single-pixel color recognition could still be a useful function, and I think it could be done by simply allowing the selection box to be resized or the screenshot to be zoomed in on even more, down to the pixel level. Also, when things weren't working as I expected, it would have been really useful to be able to test each part, for example clicking to test an image recognition instead of running the whole task. It would also be useful to have a notification of which task it is on. Either of these would help a lot with debugging, instead I had to mess with the wait timers and the order of recognition tasks.