While we are belaboring...
Web, centered on the target, takes extra 10'. The other 10' was going to be spent moving in that direction anyways.
Gust takes an extra 5' side step out and a 5' side step to get back to you = 10'. Of course, that only counts if you are their target. But you can reposition it for next turns.
If you know which direction it's going, you don't center the web on it.
Still irrelevent. Dex scaling protects you exactly the same amount without it as much as it does with it.
With 1 dex, it can absorb 3 attacks.
With 100 dex, it can absorb 3 attacks.
With 1 Dex, it absorbs 3 attacks no matter how inaccurate and weak. With 100 Dex, it absorbs three accurate, probably very powerful attacks. That is scaling, my friend. Without that, you would lose images to attacks that would have missed you anyway.
Which is the biggest downside. The fact that it can be wasted.
I'm sorry, but this really feels like grasping at straws. The biggest downside of mirror image is that it provides the most defense against the earliest attacks? With as brief as combat usually is in this game, that is a strength, not a weakness.
Enemies see you protected and can simply ignore you until it's gone.
Mirror image just reduced the damage the sorcerer takes by 100%. In practice, I have never seen a DM do this. In fact, I once had a couple of the thugs we were fighting become so frustrated with how difficult I was to hit that they kept targeting me. In practice, I have also only rarely seen a combat last more than a minute.
Or you can cast it, kick down a door, and fine no one on the other side.
Then I have misjudged the situation. I've seen players do it with dozens of spells.
Nope.
You spent 1 action and a level 2 slot to block block 3 attacks.
They spent 1 action and a level 2 slot to make 3 attacks.
Their level-2 spell slot bought them 4.57 damage. Mine prevented 10.58, and will most likely have an effect in at least one subsequent round. This is not an equivalence, and it only has the appearance of one because I set up the combatants to be equal. Mirror image will continue to be a 2nd-level spell when it is deflecting multiattacks from a CR 5 cambion, a CR 10 aboleth, or a CR 17 death knight.
However, there's a good chance at least one of their first-turn sorching rays will land an extra 0.5 damage compared to your firebolt (a bit pedantic, yes. But magic missile won't work because you both have shield).
Which is why I would cast scorching ray right back. Even if he uses shield, he will be half dead when I'm still at 75% health. What does he do then? Another scorching ray? The same casting of mirror image will reduce it to 7.66 average damage, a 49% reduction, even if I don't use shield. 54% of the time, I will still have at least one duplicate left, and the total damage reduction done by that one spell slot is now 18.07. Because it scaled in value with the number of attacks directed at me. (If 18.07 seems impossible because the damage of the average ray is 5.05, bear in mind that that latter number includes the chance to miss. Rays which would have hit me--the rays which the duplicates intercept--average 7.77 damage.) On average, 2.1 rays have now missed, 2.33 have hit illusions, and the remaining 1.57 have hit me.
Cast shatter instead? His spell DC ought to be about 15, while I've got a +4 to my Con save, and that gives me a 50% chance to succeed. 3d8 on a failed save, half on a successful one makes an average damage of 10.125. Mirror image just reduced the damage I took by 33% compared to his unaltered scorching ray, and I still have as many duplicates as I started the round with.
Also, mirror image only works if you know when the gates will open ahead of time. If you cast it, and there's a minute delay, you're out your slot.
There are so many "if" clauses you can use with any number of spells.
Sleep only works your targets have low HP. If you cast it and don't roll high enough to match anything's HP, you're out your slot.
Suggestion only works if the target can understand you. If you cast it and the target doesn't speak your language, you're out your slot.
Fireball only works if you can see a space more than 20 feet from your party members. If you cast it and there's not enough room--well, you'll still do damage, but you might cause a TPK.
You can do it for virtually any spell in the game. Does mirror image still have value if not cast at the optimal moment? Yes. Is the optimal moment reasonably frequent anyway? Yes.
If mirror image is effective the next round, that means it wasn't the first round.
Horse hockey! (Yes, I went full Sherman T. Potter.) In the demonstration, it reduced the average incoming damage by 70%. That it will most likely continue reducing damage in subsequent rounds is hardly evidence that it was not effective. Yes, you will sometimes take damage; that's the nature of the game. Show me the 2nd-level, non-concentration spell that protects you from 100% of attacks for the entire combat. Show me the one that is more effective than mirror image against the first attack, the second, the third, the fourth, or the fifth.
Blue said:
Reading about Mirror Image and how much it protects you against the first attack leaves out part of it. If the attack would miss you but would hit your mirror image, then you have chances to lose an image from attacks that would have a 0% chance to hit you. Ignoring the ablative nature of it vs. those types of attacks makes it appear more powerful than it is.
At the least, assume Dragon Bloodline or Mage Armor, which means +3 AC for the caster over the images. That's 15% of all attacks that can take out an image that would have missed in the first place. And it's easy to boost AC from there.
I won't "assume" the +3 from mage armor, because it is the caster's choice whether to accept that antergy, though you are quite right that it exists. Still, I will be happy to do the math--I long ago built a calculator in a spreadsheet that accepts all these variables.
Baseline 65% chance to hit or crit both the caster and the illusions:
Mirror image is worth 9.75 AC on the first attack, 9.11 on the second, 8.39 on the third, 7.35 on the fourth, 6.14 on the fifth, and it averages 5.62 over the first ten attacks. 23% chance at least one image remains after ten attacks.
With mage armor, 50% chance to hit or crit the caster, 65% chance the illusions:
Mirror image is worth 7.5 AC on the first attack, 7.01 on the second, 6.45 on the third, 5.65 on the fourth, 4.72 on the fifth, and an average of 4.32 over the first ten. 23% chance at least one image remains after ten attacks (no change).
Mage armor: 3 AC
Mirror image: 5.62 AC, averaged over ten attacks
Mage armor + mirror image: 7.32 AC, averaged over ten attacks
For a wild or a storm, I would recommend keeping one or the other but not both, since the diminishing returns are significant and you can always use another spell known. Use your judgment based on your campaign, but I generally prefer mirror image. For a dragon, there is less opportunity cost to keep mirror image, so it depends how much defense you feel you need.
Now, for fun, boss fight! 80% chance to hit or crit both caster and images:
Mirror image is worth 12 AC on the first attack, 11.04 on the second, 9.91 on the third, 8.10 on the fourth, and 6.12 on the fifth, and it averages 5.86 over ten attacks. The chance of at least one illusion surviving is 10%.
Boss fight, with mage armor, 65% chance to hit or crit the caster, 80% the illusions:
Mirror image is worth 9.75 AC on the first attack, 8.97 on the second, 8.05 on the third, 6.58 on the fourth, and 4.97 on the fifth, and it averages 4.76 over ten attacks. (Still 10% chance of a dupe hanging around.)
Mage armor: 3 AC
Mirror image: 5.86 AC, averaged over ten attacks
Mage armor + mirror image: 7.97 AC, averaged over ten attacks
There is less loss from the combination of the two effects here, suggesting that dragons might be best served by holding on to mirror image and only using it against especially powerful opponents. Still the player's judgment call.