Question - Is Inserter item drop position intuitive?

[Qon]

I think its pretty intuitive if you come from the belt perspective

Straight belt: Far side if orthogonal, right side of belt direction if parallel
Curved belt: inner corner

only parallel splitters look weird at first glance but if you think about that far-side thing it becomes clear^^

Wrong.

If the belt is parallel to the inserter, the item is dropped on the right side of the belt (from the belt's perspective).
If the best is perpendicular to the inserter, the item is dropped on the far side of the belt.
Corners are considered "perpendicular," and as such the item is dropped on the "far" side of the belt, which is considered to be the inside of the corner.

Wrong.

ousu(190606-160930-33).jpg (150.86 KiB) Viewed 5885 times

As soon as another case appears the "rules" don't work to predict what is going to happen. And you just have to remember the cases. So there's no rule? It's all about memorising a lot of cases?

More votes for "it's logical" and yet no-one was able to predict what was going to happen in U and V. You have to test and remember for many different cases. Votes ignored.

And we also get this:

EC left, AC right.

ousu(190606-163413-94).jpg (166.91 KiB) Viewed 5881 times

All 3 place on the same side (EC setup). That is really annoying when trying to make compact beaconed production with belts.

[jdtitans]

There's a difference between being logical/predictable versus how you want it.

The inserter/belt interaction always resolves the same way. For straight belts that are pointing toward or away from an inserter, the item will be placed on the right side of the belt, based on the belt's motion. This includes splitters and undergrounds. If not (i.e. perpendicular belts, curved belts, or perpendicularly placed splitters/undergrounds), the item is placed on the far side of the cell that can be reached.

I agree, this can make things difficult for a compact build, but it doen't make it illogical.

[Qon]

There's a difference between being logical/predictable versus how you want it.

The inserter/belt interaction always resolves the same way. For straight belts that are pointing toward or away from an inserter, the item will be placed on the right side of the belt, based on the belt's motion. This includes splitters and undergrounds. If not (i.e. perpendicular belts, curved belts, or perpendicularly placed splitters/undergrounds), the item is placed on the far side of the cell that can be reached.

I agree, this can make things difficult for a compact build, but it doen't make it illogical.

The far side isn't defined for curved belts like in M, N, O and P.

[Jap2.0]

If the belt is parallel to the inserter, the item is dropped on the right side of the belt (from the belt's perspective).
If the best is perpendicular to the inserter, the item is dropped on the far side of the belt.
Corners are considered "perpendicular," and as such the item is dropped on the "far" side of the belt, which is considered to be the inside of the corner.

Wrong.

Ah, that wasn't in the picture, so I missed it. In that case, a minor tweak to the ruleset for corners: inserters drop on what they consider to be the "far" side. To explain this concept of the far side, it can be described as either:

A. when the inserter is on the outside of the corner, it will drop on the inside of the corner
B. when the inserter is facing an end of the corner, it will drop on the outside of the corner

- OR -

A. the inserter will drop on the side of the belt that is located in the far half of the tile and (if both sides are on that half of that tile) whichever occurs least in the near side of the tile. (this is described in more mathematical terms in the next point)

- OR -

A. the inserter will drop on whichever side has the largest ratio of length (quadrants works too) on the far half of the tile to the near half. (in the first example, this would be 1:0 (∞) for the inside vs 1:2 (.5) for the outside; in the second, it would be 2:1 (2) for the outside vs 0:1 (0) for the inside.)


So I've given you three explanations for the 22 examples given, what am I missing now?
(I'm a bit curious, is it possible to via script or something take your examples of U and V and flip the belts so they're going away from the inserter? I'd test it, but I have no idea how.)

[mmmPI]

Quoting Twinsen from FFF 297

"As someone from Twitch chat noted "Inserters are so fast now, they even don't care about the side of belt". Remember that I fixed the rotation problem by finishing the Inserter movement if it's close enough. Well, what ended up happening was what now the Inserter would stop 0.0001 degrees short of perfectly vertical. That was of course closer to the other belt lane so the item would be dropped there. Previously it was always dropped perfectly vertical, and the lane decision algorithm would choose the right lane. The fix was easy and it's probably released by the time you are reading this. "


Changing the rotation of the inserter modified that " physical dropping point" which seems to exist, though "the lane decision algorithm" has the 'last word', in determining where the item will go on the belt and has been fixed in this case.

In situation M N O P I think we see this algorithm in action a bit more as it look at the belt before and after the drop point to decide where to drop the item. ( or it seems).

The problem in logic i see in situation M and N are : if you delete the belt before the curved one, both = J which changes the side of the belt going up in M, why ? this is not intuitive for me though it's a nice trick to use.

The belt before the curved one doesn't seems to have any link with the inserter, yet it makes the next belt being a curve , which in one case, means the inserter seems to 'adapt' but maybe it's actually the belt that 'adapt'.

Sometimes you can remove one such belt not realising immediatly whan will happen next.

Yet i can't picture what would be a more intuitive behavior to me, like for underground belts, EG and FH, both outputing on the same side creates the situation that 3 position offload same side vs 1 for the other side, but the change would be to decide arbitrarily that E and F would choose a different side than G and H for the sake of providing 2 positions to load on each sides vs 1-3, so it would also create a situation where you'd have to remember a rule like when belts goes up and inserter is facing up, when belt goes up and inserter is facing down, when belt goes down and inserter is facing up, when belt goes down and inserter is facing up. BUIU=> Right side BUID=> Left side BDIU=> left side BDID=> Right side.

you can tell it is not straightforward for me right ?

[Mango]

I see two straightforward rules:

1) Select farther side - if this is not enough, then

2a) for straight belts select right side
2b) for curved belts select inner part of the curve (if possible - must still be on farther side)

That's all.

[Greybeard_LXI]

The belt position is decided based on the "input" side of the belt. In U the input side is to the left while in V the input side is to the right. That determines the far side of the belt. How the belt curves after the far side is determined does not change which side is the input side.

[Qon]

The belt position is decided based on the "input" side of the belt. In U the input side is to the left while in V the input side is to the right. That determines the far side of the belt. How the belt curves after the far side is determined does not change which side is the input side.

Wrong.

turn.jpg (37.17 KiB) Viewed 5831 times

[Qon]

So I've given you three explanations for the 22 examples given, what am I missing now?

Fairly Qonfusing descriptions. I understood the first one.

(I'm a bit curious, is it possible to via script or something take your examples of U and V and flip the belts so they're going away from the inserter? I'd test it, but I have no idea how.)

They are called "Long handed inserters".

wx.jpg (97.34 KiB) Viewed 5830 times

[FuryoftheStars]

So I've given you three explanations

I see two straightforward rules:

1) Select farther side - if this is not enough, then

2a) for straight belts select right side
2b) for curved belts select inner part of the curve (if possible - must still be on farther side)

That's all.

That’s actually 3, too, but...

Wouldn’t it be easier if there was just one rule? Inserter always drops to the far right quadrant relative to itself?

[Greybeard_LXI]

The belt position is decided based on the "input" side of the belt. In U the input side is to the left while in V the input side is to the right. That determines the far side of the belt. How the belt curves after the far side is determined does not change which side is the input side.

Wrong

Guess I missed a case.

But with curved belts a simple rule would be pretend it goes straight across the inserter and put it on the far side. Then adjust for the curve.

[Jap2.0]

So I've given you three explanations for the 22 examples given, what am I missing now?

Fairly Qonfusing descriptions. I understood the first one.

Well, the intention was that at least one would get across.

(I'm a bit curious, is it possible to via script or something take your examples of U and V and flip the belts so they're going away from the inserter? I'd test it, but I have no idea how.)

They are called "Long handed inserters".

Yeah, I suppose they would work. (thanks) Nice to see that my test holds true.

[Jap2.0]

So I've given you three explanations

I see two straightforward rules:

1) Select farther side - if this is not enough, then

2a) for straight belts select right side
2b) for curved belts select inner part of the curve (if possible - must still be on farther side)

That's all.

That’s actually 3, too, but...

Wouldn’t it be easier if there was just one rule? Inserter always drops to the far right quadrant relative to itself?

Sure it would be easier. No one's debating that. This thread has devolved into providing a simple and consistent set of rules to explain how it currently works. It's not going to change because:
1. It would break a lot of factories - and while major breaking changes have been made in the past, they never were frivolously (and this could arguably be harder to fix than most due to the ubiquity of inserters and the spaghetti nature of some factories).
2. It's development time on something that has been tweaked to perfection in its current state (sure, sure, sunk time fallacy me all you want), and would have arguably little benefit.
3. This would have the greatest benefit for those who are worried about the intricacies of inserter dropping logic, which I'm guessing is a similar group of people to the groups who try to make compact designs, have large factories, and exactly who changing this would cause most harm to. New players (who a decent amount of development work is going to at the moment) simply don't often try to make beaconed setups that have to rely on inserters putting items on the correct lanes of underground belts and corners.

[mrvn]

Quoting Twinsen from FFF 297

"As someone from Twitch chat noted "Inserters are so fast now, they even don't care about the side of belt". Remember that I fixed the rotation problem by finishing the Inserter movement if it's close enough. Well, what ended up happening was what now the Inserter would stop 0.0001 degrees short of perfectly vertical. That was of course closer to the other belt lane so the item would be dropped there. Previously it was always dropped perfectly vertical, and the lane decision algorithm would choose the right lane. The fix was easy and it's probably released by the time you are reading this. "


Changing the rotation of the inserter modified that " physical dropping point" which seems to exist, though "the lane decision algorithm" has the 'last word', in determining where the item will go on the belt and has been fixed in this case.

In situation M N O P I think we see this algorithm in action a bit more as it look at the belt before and after the drop point to decide where to drop the item. ( or it seems).

The problem in logic i see in situation M and N are : if you delete the belt before the curved one, both = J which changes the side of the belt going up in M, why ? this is not intuitive for me though it's a nice trick to use.

The belt before the curved one doesn't seems to have any link with the inserter, yet it makes the next belt being a curve , which in one case, means the inserter seems to 'adapt' but maybe it's actually the belt that 'adapt'.

Sometimes you can remove one such belt not realising immediatly whan will happen next.

Yet i can't picture what would be a more intuitive behavior to me, like for underground belts, EG and FH, both outputing on the same side creates the situation that 3 position offload same side vs 1 for the other side, but the change would be to decide arbitrarily that E and F would choose a different side than G and H for the sake of providing 2 positions to load on each sides vs 1-3, so it would also create a situation where you'd have to remember a rule like when belts goes up and inserter is facing up, when belt goes up and inserter is facing down, when belt goes down and inserter is facing up, when belt goes down and inserter is facing up. BUIU=> Right side BUID=> Left side BDIU=> left side BDID=> Right side.

you can tell it is not straightforward for me right ?

You could take the belt out of the decision. Simply drop on the far left of the tile in the direction the inserter is facing. So if it is facing an underground belt and is facing north it drops on the left side of the tile. If it is facing south it drops on the right side of the tile. The direction of the underground belt doesn't matter.

Same would work for all belt cases. Only complex thing would be splitters. An inserter facing north on a splitter going left would unload after the splitter. An inserter facing south would unload before the splitter. Swap that around for splitters going right. But it's easy to see why by remembering "far left of the tile in the direction of the inserter".

[coppercoil]

It’s a good topic. I needed to think to get the answer. The rules are:

1. If the belt is perpendicular, the item is dropped on the far side. That’s simple and intuitive.
2. If the belt is parallel, the item is dropped on the right-hand side. That’s simple and intuitive too.
3. Curved belts acts like perpendicular and starts to turn past drop point. That’s simple, though not so intuitive, and you may need to think a couple of seconds to apply it.

3rdrule.png (670.93 KiB) Viewed 5747 times

I think it’s a good rule because perpendicular movement is very common case, and it allows to make a turn at the last assembling machine/chest output without breaking “far side” rule.

I voted for #3 but now I have changed my mind.

Edit: striked through wrong part of the rule.

[SkiCarver]

Some great posts on this thread ..... I would say that the simple fact that there have been multiple posts discussing the different situations is conclusive proof that many do not find the inserter drop position to be 'intuitive'. I expect that many beginners will (as I did oh so many game hours ago!) have issues sorting out what the rules ... a change to something more predictable would be good ... but is it worth the hassle - that is the question!

[Qon]

3. Curved belts acts like perpendicular and starts to turn past drop point. That’s simple, though not so intuitive, and you may need to think a couple

Wrong.
In O & P you can't say it is perpendicular and that it starts to turn past the drop point. Because it starts out parallel before the turn, which is the far side where the inserter will drop off and neither side is the far side.

[coppercoil]

3. Curved belts acts like perpendicular and starts to turn past drop point. That’s simple, though not so intuitive, and you may need to think a couple

Wrong.
In O & P you can't say it is perpendicular and that it starts to turn past the drop point. Because it starts out parallel before the turn, which is the far side where the inserter will drop off and neither side is the far side.

I agree. Here's corrected definition: 3. Curved belts acts like perpendicular at drop point.

[csduff]

I believe the rules are actually much more simple than some of these arguments.

1) If lanes are straight and...
a) run across the front of inserter, place in the farthest lane.
b) run toward/away from inserter, place in the right lane of travel.
2) If curved, place in the lane that is more often farthest from the inserter.

This seems to cover all instances that have been illustrated that I can see.

Regarding the original posted question "Is Inserter item drop position intuitive?", sometimes, not exactly.

Regarding the original poll response, there is clearly logic which defines the results.

Either way, I vote to not change anything, as it does work in a predictable and consistent way.

EDIT: The only instance that I do believe needs any attention is splitters. It would be nice to see them placed either before the split happens, or after the split in all cases. Not before the split (south and west travel) and after the split (north and east travel).

[Qon]

Either way, I vote to not change anything, as it does work in a predictable and consistent way.

Good joke!

No one managed to predict what U, V, W & X would look like. People constantly come up with wrong rules even for cases that they can check against the examples posted in the thread. It seems to be impossible to predict what will happen. You would have failed too if you didn't have the images to check before you posted.