Half-width katakana, a couple kanji, and a lot of revisions

[bluetoad07]

I actually saved the changes this time :]

Anyways, more JP stuff from me!

This PR includes:

• Half-width katakana (U+ff61-ff9f)
• As suggested and improved on by @PhMajerus:
• • Basic kanji common in early computing (万中円分千土大小年日時月木水火生百秒金人区市村町)
• • The postal mark (〒)
• A LOT of consistency and readability changes to existing characters, including completely new designs for many different kana and punctuation (see commits)
• Various bugfixes
• Removal of U+3040

Here's a map with all the updated kana designs, as well as the new kanji:

And the new half-width katakana:

The kanji were surprisingly fun to make, and given how derivative they are of each other I will absolutely do a lot more soon!

Thanks! <3

[PhMajerus]

The rationale to include the 万中円分千土大小年日時月木水火生百秒金 kanji and 〒 is explained in #147 (comment)

Also @bluetoad07, did you include the 人区市村町 as well? They are in your image of the new kanji but not in your list.

[bluetoad07]

Yes I included those. I think I copy-pasted the wrong list there.

[PhMajerus]

If you ever feel like doing a lot more, you could support JIS X 0208, which is the standard used by pretty much all legacy Japanese computers such as NEC PC-98 and IBM/MS DOS/V.

Here is the list of JIS X 0208 Level 1 kanji:

亜唖娃阿哀愛挨姶逢葵茜穐悪握渥旭葦芦鯵梓圧斡扱宛姐虻飴絢綾鮎或粟袷安庵按暗案闇鞍杏以伊位依偉囲夷委威尉惟意慰易椅為畏異移維緯胃萎衣謂違遺医井亥域育郁磯一壱溢逸稲茨芋鰯允印咽員因姻引飲淫胤蔭院陰隠韻吋右宇烏羽迂雨卯鵜窺丑碓臼渦嘘唄欝蔚鰻姥厩浦瓜閏噂云運雲荏餌叡営嬰影映曳栄永泳洩瑛盈穎頴英衛詠鋭液疫益駅悦謁越閲榎厭円園堰奄宴延怨掩援沿演炎焔煙燕猿縁艶苑薗遠鉛鴛塩於汚甥凹央奥往応押旺横欧殴王翁襖鴬鴎黄岡沖荻億屋憶臆桶牡乙俺卸恩温穏音下化仮何伽価佳加可嘉夏嫁家寡科暇果架歌河火珂禍禾稼箇花苛茄荷華菓蝦課嘩貨迦過霞蚊俄峨我牙画臥芽蛾賀雅餓駕介会解回塊壊廻快怪悔恢懐戒拐改魁晦械海灰界皆絵芥蟹開階貝凱劾外咳害崖慨概涯碍蓋街該鎧骸浬馨蛙垣柿蛎鈎劃嚇各廓拡撹格核殻獲確穫覚角赫較郭閣隔革学岳楽額顎掛笠樫橿梶鰍潟割喝恰括活渇滑葛褐轄且鰹叶椛樺鞄株兜竃蒲釜鎌噛鴨栢茅萱粥刈苅瓦乾侃冠寒刊勘勧巻喚堪姦完官寛干幹患感慣憾換敢柑桓棺款歓汗漢澗潅環甘監看竿管簡緩缶翰肝艦莞観諌貫還鑑間閑関陥韓館舘丸含岸巌玩癌眼岩翫贋雁頑顔願企伎危喜器基奇嬉寄岐希幾忌揮机旗既期棋棄機帰毅気汽畿祈季稀紀徽規記貴起軌輝飢騎鬼亀偽儀妓宜戯技擬欺犠疑祇義蟻誼議掬菊鞠吉吃喫桔橘詰砧杵黍却客脚虐逆丘久仇休及吸宮弓急救朽求汲泣灸球究窮笈級糾給旧牛去居巨拒拠挙渠虚許距鋸漁禦魚亨享京供侠僑兇競共凶協匡卿叫喬境峡強彊怯恐恭挟教橋況狂狭矯胸脅興蕎郷鏡響饗驚仰凝尭暁業局曲極玉桐粁僅勤均巾錦斤欣欽琴禁禽筋緊芹菌衿襟謹近金吟銀九倶句区狗玖矩苦躯駆駈駒具愚虞喰空偶寓遇隅串櫛釧屑屈掘窟沓靴轡窪熊隈粂栗繰桑鍬勲君薫訓群軍郡卦袈祁係傾刑兄啓圭珪型契形径恵慶慧憩掲携敬景桂渓畦稽系経継繋罫茎荊蛍計詣警軽頚鶏芸迎鯨劇戟撃激隙桁傑欠決潔穴結血訣月件倹倦健兼券剣喧圏堅嫌建憲懸拳捲検権牽犬献研硯絹県肩見謙賢軒遣鍵険顕験鹸元原厳幻弦減源玄現絃舷言諺限乎個古呼固姑孤己庫弧戸故枯湖狐糊袴股胡菰虎誇跨鈷雇顧鼓五互伍午呉吾娯後御悟梧檎瑚碁語誤護醐乞鯉交佼侯候倖光公功効勾厚口向后喉坑垢好孔孝宏工巧巷幸広庚康弘恒慌抗拘控攻昂晃更杭校梗構江洪浩港溝甲皇硬稿糠紅紘絞綱耕考肯肱腔膏航荒行衡講貢購郊酵鉱砿鋼閤降項香高鴻剛劫号合壕拷濠豪轟麹克刻告国穀酷鵠黒獄漉腰甑忽惚骨狛込此頃今困坤墾婚恨懇昏昆根梱混痕紺艮魂些佐叉唆嵯左差査沙瑳砂詐鎖裟坐座挫債催再最哉塞妻宰彩才採栽歳済災采犀砕砦祭斎細菜裁載際剤在材罪財冴坂阪堺榊肴咲崎埼碕鷺作削咋搾昨朔柵窄策索錯桜鮭笹匙冊刷察拶撮擦札殺薩雑皐鯖捌錆鮫皿晒三傘参山惨撒散桟燦珊産算纂蚕讃賛酸餐斬暫残仕仔伺使刺司史嗣四士始姉姿子屍市師志思指支孜斯施旨枝止死氏獅祉私糸紙紫肢脂至視詞詩試誌諮資賜雌飼歯事似侍児字寺慈持時次滋治爾璽痔磁示而耳自蒔辞汐鹿式識鴫竺軸宍雫七叱執失嫉室悉湿漆疾質実蔀篠偲柴芝屡蕊縞舎写射捨赦斜煮社紗者謝車遮蛇邪借勺尺杓灼爵酌釈錫若寂弱惹主取守手朱殊狩珠種腫趣酒首儒受呪寿授樹綬需囚収周宗就州修愁拾洲秀秋終繍習臭舟蒐衆襲讐蹴輯週酋酬集醜什住充十従戎柔汁渋獣縦重銃叔夙宿淑祝縮粛塾熟出術述俊峻春瞬竣舜駿准循旬楯殉淳準潤盾純巡遵醇順処初所暑曙渚庶緒署書薯藷諸助叙女序徐恕鋤除傷償勝匠升召哨商唱嘗奨妾娼宵将小少尚庄床廠彰承抄招掌捷昇昌昭晶松梢樟樵沼消渉湘焼焦照症省硝礁祥称章笑粧紹肖菖蒋蕉衝裳訟証詔詳象賞醤鉦鍾鐘障鞘上丈丞乗冗剰城場壌嬢常情擾条杖浄状畳穣蒸譲醸錠嘱埴飾拭植殖燭織職色触食蝕辱尻伸信侵唇娠寝審心慎振新晋森榛浸深申疹真神秦紳臣芯薪親診身辛進針震人仁刃塵壬尋甚尽腎訊迅陣靭笥諏須酢図厨逗吹垂帥推水炊睡粋翠衰遂酔錐錘随瑞髄崇嵩数枢趨雛据杉椙菅頗雀裾澄摺寸世瀬畝是凄制勢姓征性成政整星晴棲栖正清牲生盛精聖声製西誠誓請逝醒青静斉税脆隻席惜戚斥昔析石積籍績脊責赤跡蹟碩切拙接摂折設窃節説雪絶舌蝉仙先千占宣専尖川戦扇撰栓栴泉浅洗染潜煎煽旋穿箭線繊羨腺舛船薦詮賎践選遷銭銑閃鮮前善漸然全禅繕膳糎噌塑岨措曾曽楚狙疏疎礎祖租粗素組蘇訴阻遡鼠僧創双叢倉喪壮奏爽宋層匝惣想捜掃挿掻操早曹巣槍槽漕燥争痩相窓糟総綜聡草荘葬蒼藻装走送遭鎗霜騒像増憎臓蔵贈造促側則即息捉束測足速俗属賊族続卒袖其揃存孫尊損村遜他多太汰詑唾堕妥惰打柁舵楕陀駄騨体堆対耐岱帯待怠態戴替泰滞胎腿苔袋貸退逮隊黛鯛代台大第醍題鷹滝瀧卓啄宅托択拓沢濯琢託鐸濁諾茸凧蛸只叩但達辰奪脱巽竪辿棚谷狸鱈樽誰丹単嘆坦担探旦歎淡湛炭短端箪綻耽胆蛋誕鍛団壇弾断暖檀段男談値知地弛恥智池痴稚置致蜘遅馳築畜竹筑蓄逐秩窒茶嫡着中仲宙忠抽昼柱注虫衷註酎鋳駐樗瀦猪苧著貯丁兆凋喋寵帖帳庁弔張彫徴懲挑暢朝潮牒町眺聴脹腸蝶調諜超跳銚長頂鳥勅捗直朕沈珍賃鎮陳津墜椎槌追鎚痛通塚栂掴槻佃漬柘辻蔦綴鍔椿潰坪壷嬬紬爪吊釣鶴亭低停偵剃貞呈堤定帝底庭廷弟悌抵挺提梯汀碇禎程締艇訂諦蹄逓邸鄭釘鼎泥摘擢敵滴的笛適鏑溺哲徹撤轍迭鉄典填天展店添纏甜貼転顛点伝殿澱田電兎吐堵塗妬屠徒斗杜渡登菟賭途都鍍砥砺努度土奴怒倒党冬凍刀唐塔塘套宕島嶋悼投搭東桃梼棟盗淘湯涛灯燈当痘祷等答筒糖統到董蕩藤討謄豆踏逃透鐙陶頭騰闘働動同堂導憧撞洞瞳童胴萄道銅峠鴇匿得徳涜特督禿篤毒独読栃橡凸突椴届鳶苫寅酉瀞噸屯惇敦沌豚遁頓呑曇鈍奈那内乍凪薙謎灘捺鍋楢馴縄畷南楠軟難汝二尼弐迩匂賑肉虹廿日乳入如尿韮任妊忍認濡禰祢寧葱猫熱年念捻撚燃粘乃廼之埜嚢悩濃納能脳膿農覗蚤巴把播覇杷波派琶破婆罵芭馬俳廃拝排敗杯盃牌背肺輩配倍培媒梅楳煤狽買売賠陪這蝿秤矧萩伯剥博拍柏泊白箔粕舶薄迫曝漠爆縛莫駁麦函箱硲箸肇筈櫨幡肌畑畠八鉢溌発醗髪伐罰抜筏閥鳩噺塙蛤隼伴判半反叛帆搬斑板氾汎版犯班畔繁般藩販範釆煩頒飯挽晩番盤磐蕃蛮匪卑否妃庇彼悲扉批披斐比泌疲皮碑秘緋罷肥被誹費避非飛樋簸備尾微枇毘琵眉美鼻柊稗匹疋髭彦膝菱肘弼必畢筆逼桧姫媛紐百謬俵彪標氷漂瓢票表評豹廟描病秒苗錨鋲蒜蛭鰭品彬斌浜瀕貧賓頻敏瓶不付埠夫婦富冨布府怖扶敷斧普浮父符腐膚芙譜負賦赴阜附侮撫武舞葡蕪部封楓風葺蕗伏副復幅服福腹複覆淵弗払沸仏物鮒分吻噴墳憤扮焚奮粉糞紛雰文聞丙併兵塀幣平弊柄並蔽閉陛米頁僻壁癖碧別瞥蔑箆偏変片篇編辺返遍便勉娩弁鞭保舗鋪圃捕歩甫補輔穂募墓慕戊暮母簿菩倣俸包呆報奉宝峰峯崩庖抱捧放方朋法泡烹砲縫胞芳萌蓬蜂褒訪豊邦鋒飽鳳鵬乏亡傍剖坊妨帽忘忙房暴望某棒冒紡肪膨謀貌貿鉾防吠頬北僕卜墨撲朴牧睦穆釦勃没殆堀幌奔本翻凡盆摩磨魔麻埋妹昧枚毎哩槙幕膜枕鮪柾鱒桝亦俣又抹末沫迄侭繭麿万慢満漫蔓味未魅巳箕岬密蜜湊蓑稔脈妙粍民眠務夢無牟矛霧鵡椋婿娘冥名命明盟迷銘鳴姪牝滅免棉綿緬面麺摸模茂妄孟毛猛盲網耗蒙儲木黙目杢勿餅尤戻籾貰問悶紋門匁也冶夜爺耶野弥矢厄役約薬訳躍靖柳薮鑓愉愈油癒諭輸唯佑優勇友宥幽悠憂揖有柚湧涌猶猷由祐裕誘遊邑郵雄融夕予余与誉輿預傭幼妖容庸揚揺擁曜楊様洋溶熔用窯羊耀葉蓉要謡踊遥陽養慾抑欲沃浴翌翼淀羅螺裸来莱頼雷洛絡落酪乱卵嵐欄濫藍蘭覧利吏履李梨理璃痢裏裡里離陸律率立葎掠略劉流溜琉留硫粒隆竜龍侶慮旅虜了亮僚両凌寮料梁涼猟療瞭稜糧良諒遼量陵領力緑倫厘林淋燐琳臨輪隣鱗麟瑠塁涙累類令伶例冷励嶺怜玲礼苓鈴隷零霊麗齢暦歴列劣烈裂廉恋憐漣煉簾練聯蓮連錬呂魯櫓炉賂路露労婁廊弄朗楼榔浪漏牢狼篭老聾蝋郎六麓禄肋録論倭和話歪賄脇惑枠鷲亙亘鰐詫藁蕨椀湾碗腕

JIS X 0208 Level 2:

弌丐丕个丱丶丼丿乂乖乘亂亅豫亊舒弍于亞亟亠亢亰亳亶从仍仄仆仂仗仞仭仟价伉佚估佛佝佗佇佶侈侏侘佻佩佰侑佯來侖儘俔俟俎俘俛俑俚俐俤俥倚倨倔倪倥倅伜俶倡倩倬俾俯們倆偃假會偕偐偈做偖偬偸傀傚傅傴傲僉僊傳僂僖僞僥僭僣僮價僵儉儁儂儖儕儔儚儡儺儷儼儻儿兀兒兌兔兢竸兩兪兮冀冂囘册冉冏冑冓冕冖冤冦冢冩冪冫决冱冲冰况冽凅凉凛几處凩凭凰凵凾刄刋刔刎刧刪刮刳刹剏剄剋剌剞剔剪剴剩剳剿剽劍劔劒剱劈劑辨辧劬劭劼劵勁勍勗勞勣勦飭勠勳勵勸勹匆匈甸匍匐匏匕匚匣匯匱匳匸區卆卅丗卉卍凖卞卩卮夘卻卷厂厖厠厦厥厮厰厶參簒雙叟曼燮叮叨叭叺吁吽呀听吭吼吮吶吩吝呎咏呵咎呟呱呷呰咒呻咀呶咄咐咆哇咢咸咥咬哄哈咨咫哂咤咾咼哘哥哦唏唔哽哮哭哺哢唹啀啣啌售啜啅啖啗唸唳啝喙喀咯喊喟啻啾喘喞單啼喃喩喇喨嗚嗅嗟嗄嗜嗤嗔嘔嗷嘖嗾嗽嘛嗹噎噐營嘴嘶嘲嘸噫噤嘯噬噪嚆嚀嚊嚠嚔嚏嚥嚮嚶嚴囂嚼囁囃囀囈囎囑囓囗囮囹圀囿圄圉圈國圍圓團圖嗇圜圦圷圸坎圻址坏坩埀垈坡坿垉垓垠垳垤垪垰埃埆埔埒埓堊埖埣堋堙堝塲堡塢塋塰毀塒堽塹墅墹墟墫墺壞墻墸墮壅壓壑壗壙壘壥壜壤壟壯壺壹壻壼壽夂夊夐夛梦夥夬夭夲夸夾竒奕奐奎奚奘奢奠奧奬奩奸妁妝佞侫妣妲姆姨姜妍姙姚娥娟娑娜娉娚婀婬婉娵娶婢婪媚媼媾嫋嫂媽嫣嫗嫦嫩嫖嫺嫻嬌嬋嬖嬲嫐嬪嬶嬾孃孅孀孑孕孚孛孥孩孰孳孵學斈孺宀它宦宸寃寇寉寔寐寤實寢寞寥寫寰寶寳尅將專對尓尠尢尨尸尹屁屆屎屓屐屏孱屬屮乢屶屹岌岑岔妛岫岻岶岼岷峅岾峇峙峩峽峺峭嶌峪崋崕崗嵜崟崛崑崔崢崚崙崘嵌嵒嵎嵋嵬嵳嵶嶇嶄嶂嶢嶝嶬嶮嶽嶐嶷嶼巉巍巓巒巖巛巫已巵帋帚帙帑帛帶帷幄幃幀幎幗幔幟幢幤幇幵并幺麼广庠廁廂廈廐廏廖廣廝廚廛廢廡廨廩廬廱廳廰廴廸廾弃弉彝彜弋弑弖弩弭弸彁彈彌彎弯彑彖彗彙彡彭彳彷徃徂彿徊很徑徇從徙徘徠徨徭徼忖忻忤忸忱忝悳忿怡恠怙怐怩怎怱怛怕怫怦怏怺恚恁恪恷恟恊恆恍恣恃恤恂恬恫恙悁悍惧悃悚悄悛悖悗悒悧悋惡悸惠惓悴忰悽惆悵惘慍愕愆惶惷愀惴惺愃愡惻惱愍愎慇愾愨愧慊愿愼愬愴愽慂慄慳慷慘慙慚慫慴慯慥慱慟慝慓慵憙憖憇憬憔憚憊憑憫憮懌懊應懷懈懃懆憺懋罹懍懦懣懶懺懴懿懽懼懾戀戈戉戍戌戔戛戞戡截戮戰戲戳扁扎扞扣扛扠扨扼抂抉找抒抓抖拔抃抔拗拑抻拏拿拆擔拈拜拌拊拂拇抛拉挌拮拱挧挂挈拯拵捐挾捍搜捏掖掎掀掫捶掣掏掉掟掵捫捩掾揩揀揆揣揉插揶揄搖搴搆搓搦搶攝搗搨搏摧摯摶摎攪撕撓撥撩撈撼據擒擅擇撻擘擂擱擧舉擠擡抬擣擯攬擶擴擲擺攀擽攘攜攅攤攣攫攴攵攷收攸畋效敖敕敍敘敞敝敲數斂斃變斛斟斫斷旃旆旁旄旌旒旛旙无旡旱杲昊昃旻杳昵昶昴昜晏晄晉晁晞晝晤晧晨晟晢晰暃暈暎暉暄暘暝曁暹曉暾暼曄暸曖曚曠昿曦曩曰曵曷朏朖朞朦朧霸朮朿朶杁朸朷杆杞杠杙杣杤枉杰枩杼杪枌枋枦枡枅枷柯枴柬枳柩枸柤柞柝柢柮枹柎柆柧檜栞框栩桀桍栲桎梳栫桙档桷桿梟梏梭梔條梛梃檮梹桴梵梠梺椏梍桾椁棊椈棘椢椦棡椌棍棔棧棕椶椒椄棗棣椥棹棠棯椨椪椚椣椡棆楹楷楜楸楫楔楾楮椹楴椽楙椰楡楞楝榁楪榲榮槐榿槁槓榾槎寨槊槝榻槃榧樮榑榠榜榕榴槞槨樂樛槿權槹槲槧樅榱樞槭樔槫樊樒櫁樣樓橄樌橲樶橸橇橢橙橦橈樸樢檐檍檠檄檢檣檗蘗檻櫃櫂檸檳檬櫞櫑櫟檪櫚櫪櫻欅蘖櫺欒欖鬱欟欸欷盜欹飮歇歃歉歐歙歔歛歟歡歸歹歿殀殄殃殍殘殕殞殤殪殫殯殲殱殳殷殼毆毋毓毟毬毫毳毯麾氈氓气氛氤氣汞汕汢汪沂沍沚沁沛汾汨汳沒沐泄泱泓沽泗泅泝沮沱沾沺泛泯泙泪洟衍洶洫洽洸洙洵洳洒洌浣涓浤浚浹浙涎涕濤涅淹渕渊涵淇淦涸淆淬淞淌淨淒淅淺淙淤淕淪淮渭湮渮渙湲湟渾渣湫渫湶湍渟湃渺湎渤滿渝游溂溪溘滉溷滓溽溯滄溲滔滕溏溥滂溟潁漑灌滬滸滾漿滲漱滯漲滌漾漓滷澆潺潸澁澀潯潛濳潭澂潼潘澎澑濂潦澳澣澡澤澹濆澪濟濕濬濔濘濱濮濛瀉瀋濺瀑瀁瀏濾瀛瀚潴瀝瀘瀟瀰瀾瀲灑灣炙炒炯烱炬炸炳炮烟烋烝烙焉烽焜焙煥煕熈煦煢煌煖煬熏燻熄熕熨熬燗熹熾燒燉燔燎燠燬燧燵燼燹燿爍爐爛爨爭爬爰爲爻爼爿牀牆牋牘牴牾犂犁犇犒犖犢犧犹犲狃狆狄狎狒狢狠狡狹狷倏猗猊猜猖猝猴猯猩猥猾獎獏默獗獪獨獰獸獵獻獺珈玳珎玻珀珥珮珞璢琅瑯琥珸琲琺瑕琿瑟瑙瑁瑜瑩瑰瑣瑪瑶瑾璋璞璧瓊瓏瓔珱瓠瓣瓧瓩瓮瓲瓰瓱瓸瓷甄甃甅甌甎甍甕甓甞甦甬甼畄畍畊畉畛畆畚畩畤畧畫畭畸當疆疇畴疊疉疂疔疚疝疥疣痂疳痃疵疽疸疼疱痍痊痒痙痣痞痾痿痼瘁痰痺痲痳瘋瘍瘉瘟瘧瘠瘡瘢瘤瘴瘰瘻癇癈癆癜癘癡癢癨癩癪癧癬癰癲癶癸發皀皃皈皋皎皖皓皙皚皰皴皸皹皺盂盍盖盒盞盡盥盧盪蘯盻眈眇眄眩眤眞眥眦眛眷眸睇睚睨睫睛睥睿睾睹瞎瞋瞑瞠瞞瞰瞶瞹瞿瞼瞽瞻矇矍矗矚矜矣矮矼砌砒礦砠礪硅碎硴碆硼碚碌碣碵碪碯磑磆磋磔碾碼磅磊磬磧磚磽磴礇礒礑礙礬礫祀祠祗祟祚祕祓祺祿禊禝禧齋禪禮禳禹禺秉秕秧秬秡秣稈稍稘稙稠稟禀稱稻稾稷穃穗穉穡穢穩龝穰穹穽窈窗窕窘窖窩竈窰窶竅竄窿邃竇竊竍竏竕竓站竚竝竡竢竦竭竰笂笏笊笆笳笘笙笞笵笨笶筐筺笄筍笋筌筅筵筥筴筧筰筱筬筮箝箘箟箍箜箚箋箒箏筝箙篋篁篌篏箴篆篝篩簑簔篦篥籠簀簇簓篳篷簗簍篶簣簧簪簟簷簫簽籌籃籔籏籀籐籘籟籤籖籥籬籵粃粐粤粭粢粫粡粨粳粲粱粮粹粽糀糅糂糘糒糜糢鬻糯糲糴糶糺紆紂紜紕紊絅絋紮紲紿紵絆絳絖絎絲絨絮絏絣經綉絛綏絽綛綺綮綣綵緇綽綫總綢綯緜綸綟綰緘緝緤緞緻緲緡縅縊縣縡縒縱縟縉縋縢繆繦縻縵縹繃縷縲縺繧繝繖繞繙繚繹繪繩繼繻纃緕繽辮繿纈纉續纒纐纓纔纖纎纛纜缸缺罅罌罍罎罐网罕罔罘罟罠罨罩罧罸羂羆羃羈羇羌羔羞羝羚羣羯羲羹羮羶羸譱翅翆翊翕翔翡翦翩翳翹飜耆耄耋耒耘耙耜耡耨耿耻聊聆聒聘聚聟聢聨聳聲聰聶聹聽聿肄肆肅肛肓肚肭冐肬胛胥胙胝胄胚胖脉胯胱脛脩脣脯腋隋腆脾腓腑胼腱腮腥腦腴膃膈膊膀膂膠膕膤膣腟膓膩膰膵膾膸膽臀臂膺臉臍臑臙臘臈臚臟臠臧臺臻臾舁舂舅與舊舍舐舖舩舫舸舳艀艙艘艝艚艟艤艢艨艪艫舮艱艷艸艾芍芒芫芟芻芬苡苣苟苒苴苳苺莓范苻苹苞茆苜茉苙茵茴茖茲茱荀茹荐荅茯茫茗茘莅莚莪莟莢莖茣莎莇莊荼莵荳荵莠莉莨菴萓菫菎菽萃菘萋菁菷萇菠菲萍萢萠莽萸蔆菻葭萪萼蕚蒄葷葫蒭葮蒂葩葆萬葯葹萵蓊葢蒹蒿蒟蓙蓍蒻蓚蓐蓁蓆蓖蒡蔡蓿蓴蔗蔘蔬蔟蔕蔔蓼蕀蕣蕘蕈蕁蘂蕋蕕薀薤薈薑薊薨蕭薔薛藪薇薜蕷蕾薐藉薺藏薹藐藕藝藥藜藹蘊蘓蘋藾藺蘆蘢蘚蘰蘿虍乕虔號虧虱蚓蚣蚩蚪蚋蚌蚶蚯蛄蛆蚰蛉蠣蚫蛔蛞蛩蛬蛟蛛蛯蜒蜆蜈蜀蜃蛻蜑蜉蜍蛹蜊蜴蜿蜷蜻蜥蜩蜚蝠蝟蝸蝌蝎蝴蝗蝨蝮蝙蝓蝣蝪蠅螢螟螂螯蟋螽蟀蟐雖螫蟄螳蟇蟆螻蟯蟲蟠蠏蠍蟾蟶蟷蠎蟒蠑蠖蠕蠢蠡蠱蠶蠹蠧蠻衄衂衒衙衞衢衫袁衾袞衵衽袵衲袂袗袒袮袙袢袍袤袰袿袱裃裄裔裘裙裝裹褂裼裴裨裲褄褌褊褓襃褞褥褪褫襁襄褻褶褸襌褝襠襞襦襤襭襪襯襴襷襾覃覈覊覓覘覡覩覦覬覯覲覺覽覿觀觚觜觝觧觴觸訃訖訐訌訛訝訥訶詁詛詒詆詈詼詭詬詢誅誂誄誨誡誑誥誦誚誣諄諍諂諚諫諳諧諤諱謔諠諢諷諞諛謌謇謚諡謖謐謗謠謳鞫謦謫謾謨譁譌譏譎證譖譛譚譫譟譬譯譴譽讀讌讎讒讓讖讙讚谺豁谿豈豌豎豐豕豢豬豸豺貂貉貅貊貍貎貔豼貘戝貭貪貽貲貳貮貶賈賁賤賣賚賽賺賻贄贅贊贇贏贍贐齎贓賍贔贖赧赭赱赳趁趙跂趾趺跏跚跖跌跛跋跪跫跟跣跼踈踉跿踝踞踐踟蹂踵踰踴蹊蹇蹉蹌蹐蹈蹙蹤蹠踪蹣蹕蹶蹲蹼躁躇躅躄躋躊躓躑躔躙躪躡躬躰軆躱躾軅軈軋軛軣軼軻軫軾輊輅輕輒輙輓輜輟輛輌輦輳輻輹轅轂輾轌轉轆轎轗轜轢轣轤辜辟辣辭辯辷迚迥迢迪迯邇迴逅迹迺逑逕逡逍逞逖逋逧逶逵逹迸遏遐遑遒逎遉逾遖遘遞遨遯遶隨遲邂遽邁邀邊邉邏邨邯邱邵郢郤扈郛鄂鄒鄙鄲鄰酊酖酘酣酥酩酳酲醋醉醂醢醫醯醪醵醴醺釀釁釉釋釐釖釟釡釛釼釵釶鈞釿鈔鈬鈕鈑鉞鉗鉅鉉鉤鉈銕鈿鉋鉐銜銖銓銛鉚鋏銹銷鋩錏鋺鍄錮錙錢錚錣錺錵錻鍜鍠鍼鍮鍖鎰鎬鎭鎔鎹鏖鏗鏨鏥鏘鏃鏝鏐鏈鏤鐚鐔鐓鐃鐇鐐鐶鐫鐵鐡鐺鑁鑒鑄鑛鑠鑢鑞鑪鈩鑰鑵鑷鑽鑚鑼鑾钁鑿閂閇閊閔閖閘閙閠閨閧閭閼閻閹閾闊濶闃闍闌闕闔闖關闡闥闢阡阨阮阯陂陌陏陋陷陜陞陝陟陦陲陬隍隘隕隗險隧隱隲隰隴隶隸隹雎雋雉雍襍雜霍雕雹霄霆霈霓霎霑霏霖霙霤霪霰霹霽霾靄靆靈靂靉靜靠靤靦靨勒靫靱靹鞅靼鞁靺鞆鞋鞏鞐鞜鞨鞦鞣鞳鞴韃韆韈韋韜韭齏韲竟韶韵頏頌頸頤頡頷頽顆顏顋顫顯顰顱顴顳颪颯颱颶飄飃飆飩飫餃餉餒餔餘餡餝餞餤餠餬餮餽餾饂饉饅饐饋饑饒饌饕馗馘馥馭馮馼駟駛駝駘駑駭駮駱駲駻駸騁騏騅駢騙騫騷驅驂驀驃騾驕驍驛驗驟驢驥驤驩驫驪骭骰骼髀髏髑髓體髞髟髢髣髦髯髫髮髴髱髷髻鬆鬘鬚鬟鬢鬣鬥鬧鬨鬩鬪鬮鬯鬲魄魃魏魍魎魑魘魴鮓鮃鮑鮖鮗鮟鮠鮨鮴鯀鯊鮹鯆鯏鯑鯒鯣鯢鯤鯔鯡鰺鯲鯱鯰鰕鰔鰉鰓鰌鰆鰈鰒鰊鰄鰮鰛鰥鰤鰡鰰鱇鰲鱆鰾鱚鱠鱧鱶鱸鳧鳬鳰鴉鴈鳫鴃鴆鴪鴦鶯鴣鴟鵄鴕鴒鵁鴿鴾鵆鵈鵝鵞鵤鵑鵐鵙鵲鶉鶇鶫鵯鵺鶚鶤鶩鶲鷄鷁鶻鶸鶺鷆鷏鷂鷙鷓鷸鷦鷭鷯鷽鸚鸛鸞鹵鹹鹽麁麈麋麌麒麕麑麝麥麩麸麪麭靡黌黎黏黐黔黜點黝黠黥黨黯黴黶黷黹黻黼黽鼇鼈皷鼕鼡鼬鼾齊齒齔齣齟齠齡齦齧齬齪齷齲齶龕龜龠堯槇遙瑤凜熙

Special characters, NEC (PC-98), and IBM (DOS/V) extensions:

仝纊褜鍈銈蓜俉炻昱棈鋹曻彅丨仡仼伀伃伹佖侒侊侚侔俍偀倢俿倞偆偰偂傔僴僘兊兤冝冾凬刕劜劦勀勛匀匇匤卲厓厲叝﨎咜咊咩哿喆坙坥垬埈埇﨏塚增墲夋奓奛奝奣妤妺孖寀甯寘寬尞岦岺峵崧嵓﨑嵂嵭嶸嶹巐弡弴彧德忞恝悅悊惞惕愠惲愑愷愰憘戓抦揵摠撝擎敎昀昕昻昉昮昞昤晥晗晙晴晳暙暠暲暿曺朎朗杦枻桒柀栁桄棏﨓楨﨔榘槢樰橫橆橳橾櫢櫤毖氿汜沆汯泚洄涇浯涖涬淏淸淲淼渹湜渧渼溿澈澵濵瀅瀇瀨炅炫焏焄煜煆煇凞燁燾犱犾猤猪獷玽珉珖珣珒琇珵琦琪琩琮瑢璉璟甁畯皂皜皞皛皦益睆劯砡硎硤硺礰礼神祥禔福禛竑竧靖竫箞精絈絜綷綠緖繒罇羡羽茁荢荿菇菶葈蒴蕓蕙蕫﨟薰蘒﨡蠇裵訒訷詹誧誾諟諸諶譓譿賰賴贒赶﨣軏﨤逸遧郞都鄕鄧釚釗釞釭釮釤釥鈆鈐鈊鈺鉀鈼鉎鉙鉑鈹鉧銧鉷鉸鋧鋗鋙鋐﨧鋕鋠鋓錥錡鋻﨨錞鋿錝錂鍰鍗鎤鏆鏞鏸鐱鑅鑈閒隆﨩隝隯霳霻靃靍靏靑靕顗顥飯飼餧館馞驎髙髜魵魲鮏鮱鮻鰀鵰鵫鶴鸙黑

If you include all of these, you have all the kanji available in IBM/Microsoft codepage 932, which is the DOS/V and legacy Windows (3.x/9x) character set. I think we can consider that the exhaustive kanji set for legacy computers, as that was the dominent system from 1990 to 2000, and does include all the NEC PC-98 kanji as well.

Of couse, that's a whole 6716 kanji to add…

Edit: Forgot 1 kanji (仝) hiding in the special characters 0x81xx and NEC extension ranges, they are now properly included.

[bluetoad07]

It'll be a long time from now but I'll try to support as many as I can eventually! Thanks for the lists!

[bluetoad07]

Made some small touch-ups to existing characters while this is still open. Mostly subtle changes, but う, ぅ, ゔ, ウ, ゥ, and ヴ all have new designs for consistency and clarity (the U sounds were bothering me today apparently.)

[bluetoad07]

Okay now I'm done.

[theridane]

This also fixes カ/ガ being swapped in current releases, thanks :)

[bluetoad07]

A couple more consistency changes, I also redid a lot of the half-width katakana with better designs. I'm going to update the root comment shortly to show the new things.

[PhMajerus]

You forgot to include in your test grid the more recent foreign sounds available in the full-width Katakana block: ヴヵヶヷヸヹヺ (U+30F4 - U+30FA).
I checked your Cozette.sfd file and you included them, but you should add them to your display test if you use it to make sure everything is consistent.

[bluetoad07]

Fixed! Thanks for pointing that out :)

[PhMajerus]

I went through all CP932 and realized I forgot 5 kanjis (仝㍻㍼㍽㍾) hiding in the special characters and NEC extension blocks 0x81xx and 0x87xx. These have been added to the NEC & IBM extensions above.

Also, here is the list of all the other characters that are currently missing from Cozette to complete the CP932 support:

℃℡ÅⅦⅧ∮∽≒≦≧≪≫⊿⌒◯★☆　〒〓〝〟㈱㈲㈹㊤㊥㊦㊧㊨㌃㌍㌔㌘㌢㌣㌦㌧㌫㌶㌻㍉㍊㍍㍑㍗㎎㎏㎜㎝㎞㎡㏄㏍｡｢｣､･ｦｧｨｩｪｫｬｭｮｯｰｱｲｳｴｵｶｷｸｹｺｻｼｽｾｿﾀﾁﾂﾃﾄﾅﾆﾇﾈﾉﾊﾋﾌﾍﾎﾏﾐﾑﾒﾓﾔﾕﾖﾗﾘﾙﾚﾛﾜﾝﾞﾟ￠￡￢￣￤￥

This means these characters and the 6720 kanjis from the three groups above would make all of CP932 supported.

I already included ℃ⅦⅧ◯★☆①②③④⑤⑥⑦⑧⑨⑩⑪⑫⑬⑭⑮⑯⑰⑱⑲⑳㊤㊥㊦㊧㊨ in my PR #144, and I believe 〒〓〝〟｡｢｣､･ｦｧｨｩｪｫｬｭｮｯｰｱｲｳｴｵｶｷｸｹｺｻｼｽｾｿﾀﾁﾂﾃﾄﾅﾆﾇﾈﾉﾊﾋﾌﾍﾎﾏﾐﾑﾒﾓﾔﾕﾖﾗﾘﾙﾚﾛﾜﾝﾞﾟ￠￡￢￣￤￥ and the fullwidth space 　 are included in this PR, but I'm providing the full list here to make sure we don't forget any during testing.

So that should leave us with ℡Å∮∽≒≦≧≪≫⊿⌒㈱㈲㈹㌃㌍㌔㌘㌢㌣㌦㌧㌫㌶㌻㍉㍊㍍㍑㍗㎎㎏㎜㎝㎞㎡㏄㏍ to add besides the kanjis.

I also built a complete charmap of CP932 so it's easier to find out all characters: curl "https://raw.githubusercontent.com/PhMajerus/Documents/refs/heads/main/References/CP932%20Charmap.txt".

Edit: I included the large circle and circled digits and kanjis in my PR #144.

[bluetoad07]

Thank you!! The map makes it much easier to visualize what's left. In my next PR, which will be for the JLPT N5 kanji set, I will hopefully include all the remaining non-kanji you listed, as well as ◯〓〝〟￠￡￢￣￤￥ because they actually aren't in this one.

Also, is the ideographic space not already in the font? It doesn't show up in the changelog but it's functioned properly and appeared correctly in fontforge since my characters were first added.

[PhMajerus]

Yeah, sorry, I based my list on the characters.txt file, and it didn't include the ideographic space U+3000, probably because it's not very interesting visually 😉. It is indeed already included.

Also, if you want more tools to help you with your enormous task, I just finished building a FIGlet font that contains all the CP932 characters from the original DOS/V using octants, this makes it very quick and easy to display any of the Japanese character within your terminal to check out how they were designed back then as 16×16 glyphs:

You can then compare between the input echo and the FIGlet result within the same terminal.

You can get my phm-dosv FIGfont from my FIGfonts collection.
You need a font that contains the octants, but you could build Cozette from my PR to get those.

[PhMajerus]

Actually, I was wrong, there was only 1 extra kanji 仝 hiding in the special characters 0x81xx range.
The four others ㍻, ㍼, ㍽, and ㍾ are still required, but these are not kanji, they are square era names.

I finished examining the whole set of glyphs from DOS/V and can confirm this list contains them all. Everything matches perfectly!
The only extra characters I found are hiding in C0 (0x00 to 0x1F), probably to be used by poking them directly to text-screen memory to avoid them being interpreted as control characters, similarly to CP437 ☺☻♥♦♣♠•◘○◙♂♀♪♫☼▶◀↕‼¶§▬↨↑↓→←∟↔▲▼ characters. Most are shading blocks and half-width double lines box drawings, but I'm still trying to identify with certainty 2 glyphs at 0x0B and 0x0E that look like 〿 and ▆, I'm pretty sure 0x0B indeed is 〿, but 0x0E looks more like the exact same rectangle but filled, it's not a complete ¾ block. But those are not officially part of CP932.

I suspect 〿 is used to note an incomplete double-byte sequence, marking an invalid encoding encountered by the text-mode display driver in DOS/V, so it would probably be good to include it as part of the symbols added to handle Japanese.

[PhMajerus]

I already added ①②③④⑤⑥⑦⑧⑨⑩⑪⑫⑬⑭⑮⑯⑰⑱⑲⑳㊤㊥㊦㊧㊨ and a bunch of other circled digits to my PR, so these are all consistent with each other:

[PhMajerus]

I'll add ◯ soon because it's supposed to be consistent with the circled digits and kanji, so I might as well just provide it as well.

Edit: Now in my PR, I forgot to mention it in the commit, but it was bundled in Lucida Console / Windows 2000 console compatibility.

[PhMajerus]

I've been thinking about this CJK thing…

Considering:

• The JIS X 0208 standard defines 6,716 kanji, and Unicode 16.0 contains 97,680 ideographs.
• Some of these have different designs between Japan, China, and Taiwan for the same code point.
• The day you add Hangul, you'll have 11,172 Hangul syllables as well.

I think it could be smart to create a separate .sfd file for Japanese kanjis.
The build script could then build a "base/lightweight" version of Cozette that doesn't contain the CJK ideographs, and a "Japanese" version that contains them for system that can handle a font that's probably a few megabytes. Later, a "full" version could contain Japanese+Korean for users who want to have as many characters supported as possible.

It would probably also make merging pull requests easier since someone working on kanjis would not be impacted by other PRs touching the main Cozette.sfd file.

[bluetoad07]

I like this idea! I don't exactly intend on supporting the full 6716 in JIS X 0208, partly because some kanji are definitely too dense to represent at this size and partly because I don't want to. But I am planning to do at least ~300, so this would help a lot with organization and minimizing merging issues. I do think some of the more common kanji should remain in the base font though, maybe just what I've added so far, what do you think?

[PhMajerus]

I'm pretty sure there are rules for strokes reductions that makes it possible to include almost all of them, and DOS/V included all of them as 16×16 glyphs with 1px on the left and 1px at the bottom that always seems to be clear for spacing, so they're 15×15, not that far off 11×12 if you make them as large as possible. It is common for kana and kanji to be larger than latin letters, so that could be acceptable to maximize the chances of being able to include them all someday.

As for the number you're working on, 300 is already a nice number of glyphs to design. I'm more thinking about long term, for example if Japanese users figure they want them all, and 22 users contribute 300 glyphs each, you suddenly have the whole JIS X 0208 and need an adequate build system if you want to build lightweight versions of the font. It's easier to merge files than to prune out sets of characters that are in different Unicode blocks to reduce the set of characters.

@slavfox will probably look at our discussions next week and have the final say in both design decisions and build system.

[PhMajerus]

BTW, I noticed the existing ㏑ is designed as half-width.
All the Japanese SI units ㎎, ㎏, ㎜, ㎝, ㎞, ㎡, ㏄, ㏍ are considered full-width by most terminals.

[PhMajerus]

Microsoft "MS Gothic" at 12pt can display the whole set of JIS X 0208 kanjis as 11×11 pixels.
That font uses embedded bitmaps for low resolutions, so I suspect those are precisely designed by using strokes reduction and pixel-level adjustments. But the whole set is quite readable at a resolution that would fit Cosette's 12×13 with a 1px of spacing left and 2px of spacing bottom.
This makes me think Cozette should standardize on that size of kanji as well, as it would make it possible to someday support the whole set.

You can cat the following file in a terminal that supports octants and large type pieces (such as my PR of Cozette, or Microsoft Cascadia), and ideally all the kanjis as well, to check it out:
MS Gothic 12pt Kanji.txt

You can follow the cat command with | sed 'y/ 𜺣𜴉▖𜴀𜴹🯦𜵈𜺨𜴶𜴊𜵅▘𜴺𜴍▌𜺠▂𜶀𜶻𜵴𜶯𜶄𜶿𜵱𜶬𜶁𜶼𜵵𜶰𜶅𜷀𜴘𜵑𜴧𜵡𜴜𜵕𜴫𜵥𜴙𜵒𜴨𜵢𜴝𜵖𜴬𜵦▗𜷋𜶜▄𜶓𜷏𜶠𜷞𜶐𜷌𜶝𜷛▚𜷐𜶡▙𜴃𜴽𜴐𜵋𜴆𜵁𜴔𜵎𜴄𜴾𜴑𜵌𜴇𜵂𜴕𜵏𜵸𜶳𜶈𜷃𜵼𜶷𜶌𜷇𜵹𜶴𜶉𜷄𜵽𜶸𜶍𜷈🯧𜵙𜴯𜵩𜴣𜵝𜴳𜵭𜴠𜵚𜴰𜵪𜴤𜵞𜴴𜵮𜶖𜷓𜶤𜷡𜶙𜷗𜶨▆𜶗𜷔𜶥𜷢𜶚𜷘𜶩𜷤𜺫𜴷𜴋𜵆𜴁𜴻𜴎𜵉🮂𜴸𜴌𜵇𜴂𜴼𜴏𜵊𜵲𜶭𜶂𜶽𜵶𜶱𜶆𜷁𜵳𜶮𜶃𜶾𜵷𜶲𜶇𜷂𜴚𜵓𜴩𜵣𜴞𜵗𜴭𜵧𜴛𜵔𜴪𜵤𜴟𜵘𜴮𜵨𜶑𜷍𜶞𜷜𜶔𜷑𜶢𜷟𜶒𜷎𜶟𜷝𜶕𜷒𜶣𜷠▝𜴿𜴒▞𜴈𜵃𜴖𜵐𜴅𜵀𜴓𜵍▀𜵄𜴗▛𜵺𜶵𜶊𜷅𜵾𜶹𜶎𜷉𜵻𜶶𜶋𜷆𜵿𜶺𜶏𜷊𜴡𜵛𜴱𜵫𜴥𜵟𜴵𜵯𜴢𜵜𜴲𜵬𜴦𜵠🮅𜵰▐𜷕𜶦▟𜶛𜷙𜶪𜷥𜶘𜷖𜶧𜷣▜𜷚𜶫█/⠀⡀⠄⡄⠂⡂⠆⡆⠁⡁⠅⡅⠃⡃⠇⡇⢀⣀⢄⣄⢂⣂⢆⣆⢁⣁⢅⣅⢃⣃⢇⣇⠠⡠⠤⡤⠢⡢⠦⡦⠡⡡⠥⡥⠣⡣⠧⡧⢠⣠⢤⣤⢢⣢⢦⣦⢡⣡⢥⣥⢣⣣⢧⣧⠐⡐⠔⡔⠒⡒⠖⡖⠑⡑⠕⡕⠓⡓⠗⡗⢐⣐⢔⣔⢒⣒⢖⣖⢑⣑⢕⣕⢓⣓⢗⣗⠰⡰⠴⡴⠲⡲⠶⡶⠱⡱⠵⡵⠳⡳⠷⡷⢰⣰⢴⣴⢲⣲⢶⣶⢱⣱⢵⣵⢳⣳⢷⣷⠈⡈⠌⡌⠊⡊⠎⡎⠉⡉⠍⡍⠋⡋⠏⡏⢈⣈⢌⣌⢊⣊⢎⣎⢉⣉⢍⣍⢋⣋⢏⣏⠨⡨⠬⡬⠪⡪⠮⡮⠩⡩⠭⡭⠫⡫⠯⡯⢨⣨⢬⣬⢪⣪⢮⣮⢩⣩⢭⣭⢫⣫⢯⣯⠘⡘⠜⡜⠚⡚⠞⡞⠙⡙⠝⡝⠛⡛⠟⡟⢘⣘⢜⣜⢚⣚⢞⣞⢙⣙⢝⣝⢛⣛⢟⣟⠸⡸⠼⡼⠺⡺⠾⡾⠹⡹⠽⡽⠻⡻⠿⡿⢸⣸⢼⣼⢺⣺⢾⣾⢹⣹⢽⣽⢻⣻⢿⣿/' if you want to use Braille instead of octants to make the pixels separations more visible.

Note that font is copyrighted, this is just to show that if we pick that size, it's possible to achieve the complete JIS X 0208 with some efforts.

[bluetoad07]

Yeah, I noticed that with the units too. I will redesign that when I do those.

By the way, thank you very much for your help with this!! I am still early on in learning Japanese and your knowledge on both the language and its codecs is incredibly useful. Do you know where I can find more info on kanji stroke reduction? I can't seem to find much on it.

Edit: You answered my question as I typed it, thanks!

[PhMajerus]

Unfortunately, I don't have any document on strokes reduction, I only saw it mentioned on the Wikipedia page about Meiryo.

MS Gothic and MS Mincho were the original Windows 3.1 Japanese fonts, and because of the low resolution of screens when they were designed, they used embedded bitmaps and apparently applied strokes reduction as part of their TrueType hinting (guides to displace pixels to where they look better instead of where they should be according to the original character shape) to stay legible with a limited number of pixels.

Meiryo on the other hand is a newer font that takes advantage of higher resolution and ClearType (Microsoft sub-pixel rendering), so it doesn't contain embedded bitmaps and doesn't adjust the strokes for lower resolutions.
Because of this, when you want good low-resolution Japanese on Windows, you should select MS Gothic or MS Mincho over Meiryo. (or their "UI" versions that condenses the kana a bit more to take less horizontal space in UI elements).

I guess one of the ways to find out about which strokes are ok to simplify would be to experiment with MS Gothic at several resolutions and see how each kanji behaves when it reaches the embedded bitmaps.

I wonder if there is a public domain font that contains similarly optimized 11×11 embedded bitmaps. Maybe it is worth investigating Kochi Gothic/Mincho.
After all, I don't think there is a lot of artistic choice available to fit all those Kanji in that resolution, so maybe just merging the whole JIS X 0208 set from an existing licence-compatible project could make sense.

[PhMajerus]

I did some more tests and managed to merge the Shinonome12/東雲12 public domain font into Cozette.
It contains all the JIS X 0208 kanji but not the NEC and IBM extensions. The vast majority of the glyphs are 11px wide so it leaves a 1px gap between kanjis, except for a few that are 12px.

It definitely looks better than whatever I could work on to try to bring 6355 kanji to Cozette, and adding the 361 missing NEC+IBM extensions in the same style to achieve full CP932 compatibility seems like a reachable goal.
Also, check out the 雲, it is strokes-reduced.

Another candidate would be Naga10 as 10×10 might better fit Cozette latin characters size, but that's yet another format, I'll have a look when I have some free time.

Also, looking at other fonts, Kanji often take almost the whole square. I agree with your observation about Kanji vs Kana sizes, but Kana are also often larger than latin alphanumeric. So maybe the sweet spot to keep the best legibility would be to use the 11×11 Kanji and make the Kana slightly larger to be between latin and Kanji.

[bluetoad07]

I considered making kanji bigger, as most fonts I've seen have them around 10% bigger than kana glyphs. But at 11px, they would be 22% bigger, and I find the size difference somewhat jarring. I've tried 10px too, but that puts them off center and looks odd as well, so I ultimately decided against it.

Luckily, today in my own quests for a 9px font to use as reference, I ran across this stack exchange post which mentions a 9px font used in Etrian Odyssey on the DS, which is apparently easily understood, especially in comparison to slightly smaller sizes. I unfortunately cannot track this font down, but it's still useful in reassuring that it's actually possible to make a full readable set at this size.

^{The font in question, from the post.}

The other fonts you've mentioned will be very helpful in learning stroke reduction patterns and techniques for making efficient use of the space I have, but I definitely want to stick with 9x9 glyphs. Thank you!

[PhMajerus]

I completely agree.
While Shinonome12 shows that around 11×11 is very readable, and would work, it looks a bit too large for Cozette which has small letters in general and use the extra height more to support proper descenders and all kind of accents than to make glyphs tall.
That's why I had hopes of tracking down and testing with Naga10/ナガ10 by Sadakazu NAGAO (制一 永尾) as well, as it is designed for kanji to fit 9×9. And I have some good and some bad news.

Naga10's original web site at http://www.vector.co.jp/authors/VA013391/ is not online anymore.
The good news is I managed to get a copy of Naga10 v1.1 archive. It includes the 10×10 Kanjis from JIS Level 1 and 2, and understand its structure. With the work I already did to adjust Shinonome12, I believe I can quite easily convert it and merge it into Cozette as well.

I believe that Naga10 was probably used as the base or inspiration for the font in your screenshot, but isn't the same font. Some kanji are identical (海 in your screenshot is exactly the same), but some are completely different, the one you found uses less spacing between parts of each Kanji and is in my opinion a great improvement.
They would be size-compatible, so if Naga10's license is compatible with Cozette, we could start with that and improve each glyph based on ideas taken from other pixel-art fonts, but I wouldn't use Naga10 as it is.

We could even support both Naga10 or an improved version of it and Shinonome12 as options, as I believe Shinonome12 will stay more legible for people less used to some characters. With my idea of having kanjis in separate files we can merge, this could be easily achieved, and the work to be able to import Shinonome12 is already done, I have a file ready.

Keep in mind old font might be public domains or free software because back in the early days of the Internet, users wanted information to be freely available and would contribute to such project to make life better for everyone, but a font used in a DS game might be from a commercial company and be unusable because of licensing terms.

I believe Naga10 license could work for Cozette, here's a translation of its terms:

Copyright and Usage Conditions:
This font is free software. The copyright for all data included in this archive is owned by the author (Seiichi Nagao). Please adhere to the following conditions when using it:
1. The author shall not be liable for any problems or issues that arise from the use of this font.
2. There are no restrictions on using the font data contained in this archive for non-commercial software or distribution (we would appreciate it if you could mention somewhere, "Used Naga10" ;-)
3. This version allows the use of the font data for commercial purposes, but please consult the author in that case. If you have any questions or concerns, please inquire with the author. 

Distribution and Redistribution Conditions:
1. You may distribute freely as long as you do not alter the structure of the archive. Distribution of only font files is prohibited.
2. If you wish to convert and distribute in a format other than BDF, you may do so freely as long as you adhere to the following conditions:
  (1) Include the original README (this file).
  (2) The usage, distribution, and redistribution conditions of the new archive must follow the conditions of the original archive.
  (3) Make it clear to the distribution recipients in some way that (2) applies.
3. For distribution through publication in magazines or inclusion in supplemental CD-ROMs, it is acceptable as long as the above conditions are followed, without needing to particularly notify us.
If you have any questions or uncertainties regarding this, please ask the author.

I'm not well versed in open source licensing details though, so it would be good to have someone more knowledgeable having a look at it.

Anyway, this is all interesting and good news in general, I didn't even think we'd be able to support all the 6355 JIS Lvl1&2 Kanjis and now I have one design working and probably soon another size option. At least we'll be able to experiment and pick the best size from the start and get on a good track if designing specific glyphs.

[PhMajerus]

I managed to convert and center the Naga10 Kanji and import them into Cozette as well, here's how it looks:

For some reason, Naga10 seems to be missing the two last Kanji from the JIS X 0208 standard: 凜 and 熙, so I'll have to design those as well as the 4 era names.

Once that is done, I can provide the .bdf files to import either Shinonome12 (~11×11) or Naga10 (9×9) adjusted for Cozette.

I would be curious to see all the glyphs from the Etrian Odyssey 9×9 font, because theirs seems much more readable than Naga10. But remember game developers know which Kanji are effectively in use in all their dialogs and UI, so they might have only designed the ones they used and not the 6355 we need. Even worse, they might have simply swapped Kanjis that they couldn't make look good at 9×9 with Hiraganas and avoid them altogether.
This means we're still not sure how good the complete set can look at that size when designed with similar principles.

Basically, now that I have both versions built and easy to test, I prefer the size of 9×9, but with the current design, I find 11×11 much more readable. This is however a font design decision, Naga10 has been used on low resolution X11 displays for years, so I guess users were ok with it to get more content on screen, it's just that Cozette actually has the space to fit the 11×11 ones without taking more space.

Edit: I found another promising one that is 9×9, M⁺ Bitmap Fonts by Coji Morishita (森下 浩司). I'll be experimenting with that next.

[the-moonwitch]

Hi! Is this ready for me to start merging?

[bluetoad07]

Yep! Thank you!

[the-moonwitch]

Merged manually and releasing now, thank you!