千问人像 Lora：解决豆包脸 & 去油光，优化摄影质感 (Qwen Portrait Lora: Fixing "Doubao Face" & Oiliness, Optimizing Photographic Texture)
训练目的，千问有强大的底模能力，但原生模型生成人像不太好，用lora解决一下。
解决千问豆包脸的问题 (Fixing the issue of Qwen's "Always Same Face" [cute but less realistic facial features])
解决千问出图油的问题 (Resolving the oiliness in Qwen's generated images)
优化摄影构图 (Optimizing photographic composition)
优化摄影大光圈效果 (Enhancing photographic large aperture effect)
擅长半身，头部特写镜头 (Specializing in half-body and close-up shots of the head)
分享有礼'%3e%3cpath d='M22.941142499999998%2c18.4907265625C20.7973025%2c21.2337465625%2c18.653602499999998%2c23.9764965625%2c16.5096155%2c26.719196562500002C15.7464715%2c27.6958965625%2c16.4596735%2c29.1035265625%2c17.7179125%2c29.1035265625L30.5804125%2c29.1035265625C31.8385125%2c29.1035265625%2c32.5519125%2c27.6958965625%2c31.7883125%2c26.719196562500002C29.6448125%2c23.9764965625%2c27.500812500000002%2c21.2337965625%2c25.3570725%2c18.4907265625C24.7489725%2c17.7127265625%2c23.5494225%2c17.7127265625%2c22.941142499999998%2c18.4907265625ZM24.0441925%2c19.3528665625L24.0440425%2c19.3530465625L19.6246325%2c25.0074065625L17.6126225%2c27.5813865625Q17.5882425%2c27.6125865625%2c17.6045625%2c27.6447765625Q17.6343025%2c27.7034865625%2c17.7179125%2c27.7034865625L30.5804125%2c27.7034865625Q30.663912500000002%2c27.7034865625%2c30.6937125%2c27.6448265625Q30.7099125%2c27.6128265625%2c30.685412499999998%2c27.5814965625L24.254022499999998%2c19.3528565625Q24.2183525%2c19.3072265625%2c24.149182500000002%2c19.3072265625Q24.0800125%2c19.3072265625%2c24.0441925%2c19.3528665625Z' fill-rule='evenodd' fill='%23BEC4CA' fill-opacity='1'/%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
56'%3e%3cg%3e%3cpath d='M16.45%2c9L16.45%2c15.75Q16.45%2c15.81894%2c16.436500000000002%2c15.88656Q16.423099999999998%2c15.954180000000001%2c16.3967%2c16.017879999999998Q16.3703%2c16.08157%2c16.332%2c16.1389Q16.2937%2c16.19622%2c16.244999999999997%2c16.244979999999998Q16.196199999999997%2c16.29373%2c16.1389%2c16.33203Q16.0816%2c16.37033%2c16.017899999999997%2c16.396720000000002Q15.9542%2c16.423099999999998%2c15.8866%2c16.43655Q15.8189%2c16.45%2c15.75%2c16.45L2.25%2c16.45Q2.181056%2c16.45%2c2.1134370000000002%2c16.43655Q2.045818%2c16.423099999999998%2c1.982122%2c16.39671Q1.918426%2c16.37033%2c1.8611010000000001%2c16.33203Q1.803776%2c16.29373%2c1.755025%2c16.244970000000002Q1.706275%2c16.19622%2c1.667971%2c16.1389Q1.6296680000000001%2c16.08157%2c1.603284%2c16.017879999999998Q1.5769009999999999%2c15.954180000000001%2c1.56345%2c15.88656Q1.55%2c15.81894%2c1.55%2c15.75L1.55%2c9.00311222Q1.55%2c8.9341683%2c1.56345%2c8.866549Q1.5769009999999999%2c8.79893%2c1.603284%2c8.735234Q1.6296680000000001%2c8.671538%2c1.667971%2c8.614213Q1.706275%2c8.556888%2c1.755025%2c8.508138Q1.803776%2c8.459387%2c1.8611010000000001%2c8.421084Q1.918426%2c8.38278%2c1.982122%2c8.356397Q2.045818%2c8.330013%2c2.1134370000000002%2c8.316563Q2.181056%2c8.303112%2c2.25%2c8.303112Q2.318944%2c8.303112%2c2.3865629999999998%2c8.316563Q2.454182%2c8.330013%2c2.517878%2c8.356397Q2.581574%2c8.38278%2c2.638899%2c8.421084Q2.696224%2c8.459387%2c2.744975%2c8.508137Q2.7937250000000002%2c8.556888%2c2.832029%2c8.614213Q2.870332%2c8.671538%2c2.896716%2c8.735234Q2.923099%2c8.79893%2c2.93655%2c8.866549Q2.95%2c8.9341682%2c2.95%2c9.0031122L2.95%2c15.05L15.05%2c15.05L15.05%2c9Q15.05%2c8.931056%2c15.0634%2c8.863437Q15.0769%2c8.795818%2c15.1033%2c8.732122Q15.1297%2c8.668426%2c15.168%2c8.611101Q15.2063%2c8.553776%2c15.255%2c8.505025Q15.3038%2c8.456275%2c15.3611%2c8.417971Q15.4184%2c8.379668%2c15.4821%2c8.353284Q15.5458%2c8.326901%2c15.6134%2c8.31345Q15.6811%2c8.3%2c15.75%2c8.3Q15.8189%2c8.3%2c15.8866%2c8.31345Q15.9542%2c8.326901%2c16.017899999999997%2c8.353284Q16.0816%2c8.379668%2c16.1389%2c8.417971Q16.196199999999997%2c8.456275%2c16.244999999999997%2c8.505025Q16.2937%2c8.553776%2c16.332%2c8.611101Q16.3703%2c8.668426%2c16.3967%2c8.732122Q16.423099999999998%2c8.795818%2c16.436500000000002%2c8.863437Q16.45%2c8.931056%2c16.45%2c9Z' fill-rule='evenodd' fill='%23BEC4CA' fill-opacity='1'/%3e%3c/g%3e%3cg%3e%3cpath d='M6.120042%2c8.13058028125L9%2c11.01053828125L11.87999%2c8.13054428125L11.88003%2c8.13051328125Q11.978480000000001%2c8.03205728125%2c12.10712%2c7.97877228125Q12.235759999999999%2c7.92548828125%2c12.375%2c7.92548828125Q12.44394%2c7.92548828125%2c12.51156%2c7.93893828125Q12.579180000000001%2c7.95238928125%2c12.64288%2c7.97877228125Q12.70657%2c8.00515628125%2c12.7639%2c8.04345928125Q12.82122%2c8.08176328125%2c12.86997%2c8.13051328125Q12.91872%2c8.17926428125%2c12.95703%2c8.23658928125Q12.99533%2c8.29391428125%2c13.021709999999999%2c8.35761028125Q13.0481%2c8.42130628125%2c13.06155%2c8.48892528125Q13.075%2c8.55654428125%2c13.075%2c8.62548828125Q13.075%2c8.76472728125%2c13.02172%2c8.89336628125Q12.96843%2c9.02200628125%2c12.86997%2c9.12046328125L12.86994%2c9.12049328125L9.49497%2c12.49545828125Q9.39652%2c12.59391828125%2c9.26788%2c12.64720828125Q9.139240000000001%2c12.70048828125%2c9%2c12.70048828125Q8.860759999999999%2c12.70048828125%2c8.73212%2c12.64720828125Q8.60348%2c12.59391828125%2c8.50503%2c12.49545828125L5.130025%2c9.12046328125Q5.031569%2c9.02200628125%2c4.978284%2c8.89336628125Q4.925%2c8.76472728125%2c4.925%2c8.62548828125Q4.925%2c8.55654428125%2c4.93845%2c8.48892528125Q4.951901%2c8.42130628125%2c4.978284%2c8.35761028125Q5.004668%2c8.29391428125%2c5.042971%2c8.23658928125Q5.081275%2c8.17926428125%2c5.130025%2c8.13051328125Q5.178776%2c8.08176328125%2c5.236101%2c8.04345928125Q5.293426%2c8.00515628125%2c5.357122%2c7.97877228125Q5.420818%2c7.95238928125%2c5.488437%2c7.93893828125Q5.556056%2c7.92548828125%2c5.625%2c7.92548828125Q5.764239%2c7.92548828125%2c5.892878%2c7.97877228125Q6.021518%2c8.03205728125%2c6.119975%2c8.13051328125L6.120042%2c8.13058028125Z' fill-rule='evenodd' fill='%23BEC4CA' fill-opacity='1'/%3e%3c/g%3e%3cg%3e%3cpath d='M8.29560546875%2c2.25Q8.29560546875%2c2.181056%2c8.30905546875%2c2.1134370000000002Q8.32250646875%2c2.045818%2c8.34888946875%2c1.982122Q8.37527346875%2c1.918426%2c8.41357646875%2c1.8611010000000001Q8.45188046875%2c1.803776%2c8.50063046875%2c1.755025Q8.54938146875%2c1.706275%2c8.60670646875%2c1.667971Q8.66403146875%2c1.6296680000000001%2c8.72772746875%2c1.603284Q8.79142346875%2c1.5769009999999999%2c8.85904246875%2c1.56345Q8.92666146875%2c1.55%2c8.99560546875%2c1.55Q9.06454946875%2c1.55%2c9.13216846875%2c1.56345Q9.19978746875%2c1.5769009999999999%2c9.26348346875%2c1.603284Q9.32717946875%2c1.6296680000000001%2c9.38450446875%2c1.667971Q9.44182946875%2c1.706275%2c9.49058046875%2c1.755025Q9.53933046875%2c1.803776%2c9.57763446875%2c1.8611010000000001Q9.61593746875%2c1.918426%2c9.64232146875%2c1.982122Q9.66870446875%2c2.045818%2c9.68215546875%2c2.1134370000000002Q9.69560546875%2c2.181056%2c9.69560546875%2c2.25L9.69560546875%2c12Q9.69560546875%2c12.06894%2c9.68215546875%2c12.13656Q9.66870446875%2c12.20418%2c9.64232146875%2c12.2679Q9.61593746875%2c12.3316%2c9.57763446875%2c12.3889Q9.53933046875%2c12.4462%2c9.49058046875%2c12.495Q9.44182946875%2c12.5437%2c9.38450446875%2c12.582Q9.32717946875%2c12.6203%2c9.26348346875%2c12.6467Q9.19978746875%2c12.6731%2c9.13216846875%2c12.6865Q9.06454946875%2c12.7%2c8.99560546875%2c12.7Q8.92666146875%2c12.7%2c8.85904246875%2c12.6866Q8.79142346875%2c12.6731%2c8.72772746875%2c12.6467Q8.66403146875%2c12.6203%2c8.60670646875%2c12.582Q8.54938146875%2c12.5437%2c8.50063046875%2c12.495Q8.45188046875%2c12.4462%2c8.41357646875%2c12.3889Q8.37527346875%2c12.3316%2c8.34888946875%2c12.2679Q8.32250646875%2c12.20418%2c8.30905546875%2c12.13656Q8.29560546875%2c12.06894%2c8.29560546875%2c12L8.29560546875%2c2.25Z' fill-rule='evenodd' fill='%23BEC4CA' fill-opacity='1'/%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
下载
1
3神都猫的丹炉
2025-11-06 更新
关注
神都猫的丹炉
2025-11-06 更新
关注
工作流介绍
'%3e%3cg%3e%3cpath d='M2.00048828125%2c13.15673828125C2.00048828125%2c13.43287828125%2c2.22434628125%2c13.65673828125%2c2.50048828125%2c13.65673828125L3.00048828125%2c13.65673828125L3.00048828125%2c6.15673828125C3.00048828125%2c5.88059628125%2c2.77663028125%2c5.65673828125%2c2.50048828125%2c5.65673828125C2.22434628125%2c5.65673828125%2c2.00048828125%2c5.88059628125%2c2.00048828125%2c6.15673828125L2.00048828125%2c13.15673828125Z' fill='%23D8D8D8' fill-opacity='1'/%3e%3c/g%3e%3cg transform='matrix(0%2c1%2c-1%2c0%2c22.656814575195312%2c2.6558380126953125)'%3e%3cpath d='M10.00048828125%2c13.156326293945312L10.00048828125%2c20.656326293945312L10.50048828125%2c20.656326293945312C10.77663028125%2c20.656326293945312%2c11.00048828125%2c20.43246629394531%2c11.00048828125%2c20.156326293945312L11.00048828125%2c13.156326293945312C11.00048828125%2c12.880184293945312%2c10.77663028125%2c12.656326293945312%2c10.50048828125%2c12.656326293945312C10.22434628125%2c12.656326293945312%2c10.00048828125%2c12.880184293945312%2c10.00048828125%2c13.156326293945312Z' fill='%23D8D8D8' fill-opacity='1'/%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
千问人像 Lora：解决豆包脸 & 去油光，优化摄影质感 (Qwen Portrait Lora: Fixing "Doubao Face" & Oiliness, Optimizing Photographic Texture)
训练目的，千问有强大的底模能力，但原生模型生成人像不太好，用lora解决一下。
解决千问豆包脸的问题 (Fixing the issue of Qwen's "Always Same Face" [cute but less realistic facial features])
解决千问出图油的问题 (Resolving the oiliness in Qwen's generated images)
优化摄影构图 (Optimizing photographic composition)
优化摄影大光圈效果 (Enhancing photographic large aperture effect)
擅长半身，头部特写镜头 (Specializing in half-body and close-up shots of the head)