Ther erbi (二笔) IME has multiple versions, the most popular is called qingsong (青松).

The advantage of this system in contrast to wubi, is that is easy to memorize.
The drawback is that you need to know at least the first pinyin character.

There are five basic strokes, these are in order:
- horizontal (and not-really-horizontal strokes going from left to right)
- vertical (with an optional hook at the end)
- slanted down-left
- slanted down-right (and dots generally)
- other (mainly strokes with breaks in them)

A character is built up as follows:

(a) characters with only one part:
1. the first pinyin character
2. first and second strokes
3. third and fourth strokes
5. fifth stroke

(b) left-right / top-bottom / enclosure characters:
1. the first pinyin character
2. first and second strokes of the first part
3. first and second strokes of the second part
4. third and fourth strokes of the second part

The keyboard is divided into 6 regions:
1. QWERT (down-left)
2. YUIOP (down-right)
3. ASDFG (vertical)
4. HJKL; (horizontal)
5. ZXCVB (one stroke)
6. NM,./ (other)

In each region the second strokes are encoded from left to right,
so e.g. Q : down-left + horizontal; D : vertical + down-left,
so in this way we can write two strokes with one key.
If we want to write only one stroke, the "one stroke" region should be used.

Examples:
- 田: TGJZ (T[ian] + G [vertical+other] + J [horizontal+vertical] + Z [horizontal])
- 规: GHGT (G[ui] + H [horizontal+horizontal] + G [vertical+other] + T [down-left+down-right])

There are words with multiple characters, here the codes of all characters are distributed.
Example:
- 万岁: W;SG (W[an] + ; [horizontal+other] + S[ui] + G [vertical+other])
- 毛主席: MQZX (M[ao] + Q [down-left+horizontal] + Z[hu] + X[i])
- 中华人民共和国: ZHRG (Z[hong] H[ua] R[en] G[uo])

Note: my implementation does not support multiple characters.

Some common roots are mapped to keys:
- Z: metal
- X: wood
- C: water
- V: earth
- B: grass
- S: sun
- D: moon
- F: man
- L: mouth
- U: hand
- I: fire

Examples:
- 木: MX (M[u] + X [wood])
- 增: ZVIG (Z[eng] + V [earth] + I [down-right+down-left] + G [vertical+other]

For some characters you don't have to write out the whole code,
e.g. ZVIG above can be shortened to ZVI, but you have to "accept" it with the space key.

Note: in my implementation, you should write a code only until the right character appears,
e.g. writing TGJZ, you see the character changing like this: 他团町田,
so if you wanted to type "团", you need to stop at the second character by typing a space.
Also, when there are no continuations, the character is accepted automatically,
so e.g. ZVI writes 增 without the need of a final space keypress,
because there are no other codes starting with ZVI.

With only the first character + space key, you can write the following:
起为而人他一大有偶平
安是的分个和就可了
在学成这不你们
[According to the variant, 偶 may be replaced by 哦, 成 by 产, etc.]

It should be noted, that there are some keys that don't start a valid pinyin, but there are
also characters mapped to these (though they can be written by their normal code, as well):
- U: 大 (DKV)
- I: 有 (YKD)
- V: 这 (ZYPV)

Also, the first character of a code cannot be punctuation,
so these (;,./) can be used normally.

These extra keys are also used in another way (though this may not be standard):
some frequent characters can be written by U + one key:

区我很认同以角首头排
则四第化由于及科里倒
总系重种吧建母放难呀

In most cases the mapping is based on the first pinyin character,
but there are quite a lot of inconsistencies, to write some common characters.

Note: in my version, I preferred this mapping:

却王怎认头以手西种吧
当事都放工很己科里老
主行重同便女名知于四

In the same way, the I and V keys also have mappings - mainly symbols or compounds,
it is interesting to experiment with them.

Note: in my version there is no multikey combination mapped to I or V.

Let's see a sentence:

二笔输入法非常有意思，对吧？
EH BQ S;FRR FC FA CFL IYYPSGP,D. UP?

The main advantage of this method is that you don't have to choose from lists.
This is true most of the time, but there are difficult cases.
For example, the 遂 character cannot be differentiated from the 送 character,
as both are written by the SIPV code.
My implementation handles this by mapping the more frequent 送 to SIP,
and the less frequent 遂 to the SIPV code.

Even with this, there are problems, e.g. 奶/努 (N,,).
Here you still have to select from a list, but fortunately there are not so many of these,
and in most cases at least one of the colliding characters is very rare.

Here's a list of those characters whose code collide,
not counting the combinations with multiple characters:

BL;T	哔跸
BTTN	鲍飑
CGHK	蝽帱
CJS	刺曹
CUQA	捶攒
CX.V	杈柽
CYFF	卒痤
CYGH	衰廛
DGJ	盯町
DJJ	酊耵
DKJA	碟耷
DLJA	喋蹀
GGGH	蝈帼蜾
GGJL	蛄罟
GGS	刚刿
GJGF	贡贾
GJT.	毂彀
HCGJ	溃涸漶
HCJA	洪潢
HGJA	蚶蟥
HGJL	蝴岵
JBYR	荠蒺茭
JGFW	蚧骱
JGFY	瞿雎
JJV	基墼
JKS	劂剞
JLKI	蹶噘
JLYR	霁跻哜
JOHS	蹇謇
JTGH	飓龟
JTL	句咎
JTYR	饺鲛鲚
JYM	郊疖
JYST	竟竭
KOGH	窠悝
L/GJ	蠡缧
LCGJ	漯澧
LCXX	淋漤
LCYR	浏漓濂
LGFP	囹蛉
LGIJ	髅蝼嵝瞵
LKTV	硌尥
LL,	另叻
LLK,	雳呖
LYIJ	麟瘘
LYJN	癞瘌
LYYG	凛廪
M/	马乜
MBGH	苜瞢
MBSK	莫墓
MBYC	蘑蘼
MGBG	瞄蠛
N,,	奶努
PCJN	浦溥
PYTN	庖疱
QCX	渠柒
QGHA	蜻圊
QGIH	蜷蜣
QJT	欺壳
QXFY	樵檎
SBMJ	蔬荪
SCJA	澌滠
SCJN	漱涑
SFJ	什隼
SQJN	簌穗
SUJA	撒撕揲
TUKO	挞闼
WJL	吾軎
XBYI	薪莘
XL,	吸勋
XLA,	嘘噱
XXWN	皙檄
Y/.H	绎驿
YCGK	泱洇
YG.L	贻眙
YG/P	黝蚴
YGKU	崦罨
YGKV	崖睚
YGTP	鸯鹦
YJM	耶鄞鄢
YLM	郢郧
YLYI	噫喑
YTKG	颖颍
YVZS	垣堰
YYW;	庾瘐裒
Z/TN	绉驺
ZEGJ	舯舳
ZGI;	赚嵫
ZGQK	蜘帙
ZGYI	幢蟑嶂幛
ZIMZ	籽孳
ZLAA	嘴趾呲
ZQJK	筑箸
ZTS	旨昝
ZU.;	摺擢
ZUGJ	蛰蜇
ZXYI	梓樟
ZYKG	斋颤
ZYPO	恣戆