掲示板 Forums - Mnemonics - visual upgrades

That would be brilliant! Are you learning French?

That looks very interesting. Although I already know that for some of my mnemonics it won't work.
How exactly does it work? I sometimes chose parts that don't appear in the list. Can the system recognise and connect those?

I just needed about 15 questions to see that it's worse. In the case of the French defs, I'm pretty sure there was a problem at the import, maybe not so with the Russian.
First, there are fewer defs on the database you use (than in French). For example it switched to English for 頬, or めぐまれる. At the very least, it's incomplete.
Then the def for ぶつける contains "ぶつける”
And the def for 品(ひん）starts with "counter of dishes" (should not be first!), and then elegance (OK), and then rough, crude (uh? Oh, OK, that's for 品のない), and then dignity (should be next to elegance). Oh, and it ends with "ひん”
There are decent Russian Japanese dictionaries out there... But maybe not available for download and use on a different website? In any case, this one is rather weak, if my little quiz was representative of the whole dictionary.

Sorry for the mix up, it was just a reminder of something I mentioned before. For the Russian, it's mostly as you say. But as I explained, for the French, in some cases the beginning of the open definition (which I check with the Android app Tenjin) is simply missing, such as with ける. The rest of the defs are the same as on renshuu. The other difference being that Tenjin also picked up from tatoeba the many thousands of example sentences translated into French and other languages.
And if I check Russian defs available with Tenjin, I see it's a different database (Open WordNet) as the defs are different, but only slightly better. E.g. there's a def for 頬, but not for まぐまれる.

To make my life easier, I'll show off some screenshots.

The simplified kanji for ほたる features a katakana ツ which is listed as a hiragana つ. You corrected the same issue for 桜 already, but I suppose more kanji are affected by that issue.
After doing some research, I found that the kanji is based on this 熒 part. Since it is easily recognizable especially in the more complicated kanji and makes a lot of sence meaning wise, I used that part for both kanji mnemonics.

Next thing I noticed is that some meanings are missing.

As for 蝶 I always had a feeling that there is a connection to "leaves" and was kinda dazzled when the corresponding part didn't appear among the choices, so I added it.

And lastly there are 4 or so kanji using that "sound" + "halbert" compound, but only two of them feature it as a part (missing the meaning, though)...

...and the other two don't.

Just a thought : the visual improvement you are working on might also be interesting when you get to show the most similar kanjis (as in Aedict: "do not confuse with"), to highlight the part that is different. Say 持／待, if you put one on top of the other, with the left parts highlighted, it might help differentiate them, even better than simply listing the similar-looking kanjis. Those 2 in my example are so frequent they're not so prone to confusion, but for many other pairs, such direct comparison could help.
Edit: not sure why, but here is how I see your post reply, line breaks aren't working correctly:
 

A - Most of the parts have names now.
B - Although I cannot look at all of them as a whole, I took the examples you had to see what was going on:
"And lastly there are 4 or so kanji using that "sound" + "halbert" compound, but only two of them feature it as a part (missing the meaning, though)..." --> The way the character was built in the dictionary for the stroke diagrams prevents the site from being able to reasonably guess how these pieces come together in the kanji that do not already have the piece listed. 
C - looking at 蝶, you actually put a piece in there that isn't in the kanji itself. the leaf kanji 葉 has a few extra strokes on top, and I have not yet been able to find anything that suggests the piece on the right side of the kanji exists as a single unit.
D - I also tried to make the mobile version of the constructor a bit more space-conscious.

と言いました。

Yes!

hmm...I swear that's not what I saw the last time I looked at it, but I was pretty sick when I checked it yesterday. Let me look again and get back to you.

Ok, so I took a closer look at it. My guess, without having actually asked the people in charge of the list, is that they did not put any characters like 枼 that are not normally used by Japanese in the kanji data file. 

So here is what I think I could do. It would be a bit of work, but if you really want to get that colorful thing down, then I would need to extend the mnemonic tool. Not when you make the mnemonic, but when you "bind" it (similar to sentences), it would say "hey, this other piece you are using, it isn't in my data. Can you tell me which lines belong to it? You could click on a bunch of lines to turn it on, save it, and it would then know how to color it properly.