ACME Crafting BOT

Working great on DX9.

I pruned some more search trees yesterday and had another idea to make things even faster :D, maybe new version today.

I'm gonna remove showing icons on the left side of the bot, just going to use labels, for a few different reasons. (one of them includes I don't wanna deal with shit)

Those icons are a double bladed sword.
If they look fine to let you know how far the bot is doing with the actual craft (making you breathe again during a 3* with hasty touches), on the other hand the continuous update of that list makes the program sometimes go berserk and shows the "icon error" (the square with the small image on the top right side) for missing images. That lasts only a second but yet... xD

I guess (and that's just a wild guess as i'm no coder) that removing those icons and anything related to "showing the human in front of the monitor what i am going to do" might as well improve the overall performance and reduce the memory used.

Another good option u could come up with is an "Always on top" option.
Most ppl is playing "Borderless windowed" to use their parser (ACT). Now if there was a chance to have the bot always on top unless you use the "Reduce to icon" button that would be amazing. Or you could use the idea the fisher bot creator had, the "mini bar" that remains on top until the bot is fishing (in this case crafting) with an icon showing what it's the bot doing at that time. That would be great in an easy readable form like:

Ofc this ASCII doesn't really looks like a finished work, but gives a rough idea.

As usual i don't know if this is doable with this bot, but throwing ideas never killed anyone, so... why not xD

06/17/2016
- Faster algorithm - not sure whats the % but seems considerably faster, have not reduced how good the searches are, just pruned some more pointless subtrees.
- Fixed some bugs .
- Small UI changes.

Highly recommend you guys download it!

Quote:

Originally Posted by F4Bz Those icons are a double bladed sword. If they look fine to let you know how far the bot is doing with the actual craft (making you breathe again during a 3* with hasty touches), on the other hand the continuous update of that list makes the program sometimes go berserk and shows the "icon error" (the square with the small image on the top right side) for missing images. That lasts only a second but yet... xD I guess (and that's just a wild guess as i'm no coder) that removing those icons and anything related to "showing the human in front of the monitor what i am going to do" might as well improve the overall performance and reduce the memory used. Another good option u could come up with is an "Always on top" option. Most ppl is playing "Borderless windowed" to use their parser (ACT). Now if there was a chance to have the bot always on top unless you use the "Reduce to icon" button that would be amazing. Or you could use the idea the fisher bot creator had, the "mini bar" that remains on top until the bot is fishing (in this case crafting) with an icon showing what it's the bot doing at that time. That would be great in an easy readable form like: Code: +-------+-------------------------------+-----------------------+-----------------------------+ | ACME | Crafting: [Qty]x [Recipe] | Status: ------------ | | _ | [] | X | +-------+-------------------------------+-----------------------+-----------------------------+ Ofc this ASCII doesn't really looks like a finished work, but gives a rough idea. As usual i don't know if this is doable with this bot, but throwing ideas never killed anyone, so... why not xD

Generally speaking, everything is possible with programming, the question is always will/skills.

That's another great idea I've actually thought of this before, I actually really hate the ACT overlay, overlay needs to be unclickable and auto-minimize/move with the game window, and also transparent(obv), I've done this for a small project on another game, the code was in WPF not winforms though... anyways.. I'll see.. maybe, eventually...

Yesterday, I saw the bot confidently (ignoring a good, so it believed it had a 100% solution, I think) get to 98% on quality. Here's my logs in case it was a rounding error with quality formula.

I noticed that too.

Made it twice during two crafts for the last crafter tools available in the game (these 3.3 tools) so it's repeatable... At first i thought it was too many fails at Hasty but then i noticed that, starting off HQ materials, it NEVER used a single hasty touch, so it went straight with regular touches and all of them was 100% so it never failed one. He went twice to 98% (apparently same rotation for both crafts) and closed the craft (with 10 durability left AND some leftover CPs (nothing more than 10 ofc).

Ya being off by 1 point of 11k quality is 98%, I can very very slightly lower the quality formula...

Unfortunately, FFXIV has a terribad RNG where a 98% has the same weight of a 1% (how many items did you actually HQ while Quick Synthesizing? And how many times did that 98-99% items come out as NQ?).

Ok i've been lucky too by getting some HQ crafts when at 10% (like when i failed EVERY single hasty touch in a craft and still managed to get to that 10%) but reaching 100% is the only way to be sure you're not wasting materials on a (potentially) NQ craft.
Or at least that's what crafters usually try to do...

Also, as far as i could notice, it's not actually giving track of how many quality points u're missing from the cap in order to define HQ\NQ items (something like 9999\10000 means only one chance on 10000 to go NQ). It's just the % that rules that chance and 98% has a potentially high 2% chance to go awfully wrong xD

This is the xiv curquality/maxquality vs HQ% plot, plotted with ACME data. (200k+ data pts)

[Only registered and activated users can see links. Click Here To Register...]

The rng works fine IMO, I was just complaining about 1/11k being 2% which is not it's 0.0001~... something completely negligible, but the formula they use anything short of 100% is 98%.

What would be nice to do is use all the goods possible, if you can get a tricks/pt off it will definitely give more room for the algorithm to adjust itself incase of a bad excellent...

Quote:

Originally Posted by AcmeCrafting What would be nice to do is use all the goods possible, if you can get a tricks/pt off it will definitely give more room for the algorithm to adjust itself incase of a bad excellent...

It does. Last 3* i crafted it went the usual rotation that would take me to 98%, got a Precise Touch (after 2-3 Goods) out of SH during the flawless synthesis phase and it went up to 100% ;)

Btw about the Reclaim thing...
Technically the bot creates a "shadow" image for the craft, creating a rotation on the fly to cope woth what he's got in terms of Craftmanship (to define how many progress skills it has to use), Control (to define how many touch action it has to use to fill the quality bar) and CPs (to know how many "actions" it can use to achieve its goal). Now what if Reclaim rule is something like:

- Check a baseline to define, before breaking the wall of 55 CPs left, how much quality it should have and how much it have and if the remaining IQ stacks will fill that gap
- IF Quality is, let's say, 6% or more lower than expected, Reclaims and hasty touches to fail
- IF Quality is, let's say, 5% lower than expected it keeps going (we wish in a precise touch?) and in any case the 90% chance to reclaim is similar to the 90% ish chance to HQ that, so it can just finish the craft
- IF Quality is above that value, it just keeps going as the craft is gonna be 100% no matter what

This could be a pretty good way to define Reclaim usage even tho that would mean one more cross class skill, but considering those days when you can't really hit a single hasty touch... u know... having a way to skidaddle could be nice.

As usual i'm just throwing ideas based off what i see and what i would modify if i was the coder :)

Nah, it doesn't use all of the possible goods. Once it decides on a 100% solution it won't be (very?) reactive anymore. For example, during a SH->GS->BB finisher, if you get a good before GS, it will still use GS then BB. Not only is it a waste of 32CP, you could get two chances to bait another good or excellent and gain 20cp by doing SH-ToT-GS-(Observe)-BB.

Quote:

Originally Posted by jelio Nah, it doesn't use all of the possible goods. Once it decides on a 100% solution it won't be reactive anymore. For example, during a SH->GS->BB finisher, if you get a good before GS, it will still use GS then BB. Not only is it a waste of 32CP, you could get two chances to bait another good or excellent and gain 20cp by doing SH-ToT-GS-(Observe)-BB.

It's always reactive up to a point - it's VERY reactive vs excellents we always try to fix bad excellents, even after you get a 100% solution,it's why you keep getting all the different answers on easy items.

On your example good modifier is 1.5 which is less than 2.0 so it's definitely not a waste to GS, but yes we could tricks gs bb and this is something maybe I can improve, I probably can't make it use ALL goods on tricks/pt but I can make it use some more, this will mostly help with futureproofing vs bad excellents, but will also help with the 98% thing.

Btw, I've prob crafted well over 150 3* end items since patch, with mostly 2* gear, all HQ, not even close to a NQ.
Here's the last 3 batches
[Only registered and activated users can see links. Click Here To Register...]
[Only registered and activated users can see links. Click Here To Register...]
[Only registered and activated users can see links. Click Here To Register...]

:D

Plz don't undercut me!

Hmm, I wonder why I have trouble sometimes. I NQed 2 out of 35 one star crafts using the recommended cross class skills for 2 star. Guessing stats are too low. 837 Craft, 760 Control, 444 CP with old food. Way too scared to try to make myself i190 gear (my specialists have enough with good food for the minimum).

Quote:

Originally Posted by jelio Hmm, I wonder why I have trouble sometimes. I NQed 2 out of 35 one star crafts using the recommended cross class skills for 2 star. Guessing stats are too low. 837 Craft, 760 Control, 444 CP with old food. Way too scared to try to make myself i190 gear (my specialists have enough with good food for the minimum).

I've mentioned this before 1*'s are actually slightly harder than 2* due to MaMa being broken as fuck, not only that but things you sell/use that are 2* are usually end-items with craftable mats, so you have a shot at HQing the mats, which really fucking matters and compounds the difference between 1* and 2*, full HQ mats means your craft starts from 50% quality even though the bar shows 14% (look at the plot I linked) , 1* items ppl craft is usually food with full NQ mats so thats like really really hard.

Also you're at 444 CP I've ran a SIM for 2* at some points increasing your CP by 20 points made a huge HQ% difference (dont recall exact number), CP is king, control comes 2nd, craftsmanship just get the minimum.

It's absolutely silly watching it craft 3* with some HQ mats, even with 2* stats I've finished quite a few IW using 0 hasty touches.

ACME will only 90-95% 3* with 2* gear and full NQ mats, because you need a fuckton of hasties to land , but since every end 3* item has subcraftable mats that HQ rate goes to something between 99% and 100%, since all you need is to get a few goods or excellents and/or a few hasties to land for it to finish the item for you with a 100% rotation.

I've NQ'd a ton of 1* food (my cul is sort of shitty tier geared too), but 2* I've only NQd 1 item out of hundreds, and 3* I'm at 0 NQ's out of god knows how many, only NQ'd that 2* because I was mostly using serpentskin gear with NQ mats and it literally missed 80% of hasty touches


Edit: Also when I say 2* stats I mean fully overmelded, CP & Control are important.

Yeah, you're right of course. Just easily made my first ironworks piece starting out with about 4200 quality. I've been trying to catch up from lvl60 melded white gear. It seems possible if you supplement with HQ mats that you can't craft, and get the i200/i180 tools from scrips.