comfyui图像类模型及工作流分享（2026.03.07更新）

小雨点

老大上边说的应该是一个很重要的知识点，我得慢慢理解消化一下行，

darkerz

wulesoicq 发表于 2025-12-3 15:51
如果更新后遇到!!! Exception during processing !!! too many values to unpack  ...

没有魔法，看不到，老大能直接发下教程吗，感谢

darkerz

wulesoicq 发表于 2025-12-3 15:51
如果更新后遇到!!! Exception during processing !!! too many values to unpack  ...

打开了，替换了个文件，成功了，感谢老大！

darkerz

darkerz 发表于 2025-12-3 22:06
打开了，替换了个文件，成功了，感谢老大！

运行了下， 两张黑图，我回退下再试试

darkerz

wulesoicq 发表于 2025-12-3 15:51
如果更新后遇到!!! Exception during processing !!! too many values to unpack  ...

除了图像1人物，图2和图3都是黑图，请求大佬帮忙

aaaaqtqt

在运行z_image_turbo文生图工作流时，在加载CLIP这一步出现问题。

“
加载CLIP

Error(s) in loading state_dict for Llama2:size mismatch for model.embed_tokens.weight: copying a param with shape torch.Size([151936, 2560]) from checkpoint, the shape in current model is torch.Size([128256, 4096]).size mismatch for model.layers.0.self_attn.q_proj.weight: copying a param with shape torch.Size([4096, 2560]) from checkpoint, the shape in current model is torch.Size([4096, 4096]).size mismatch for model.layers.0.self_attn.k_proj.weight: copying a param with shape torch.Size([1024, 2560]) from checkpoint, the shape in current model is torch.Size([1024, 4096]).size mismatch for model.layers.0.self_attn.v_proj.weight: copying a param with shape torch.Size([1024, 2560]) from checkpoint, the shape in current model is torch.Size([1024, 4096]).size mismatch for model.layers.0.self_attn.o_proj.weight: copying a param with shape torch.Size([2560, 4096]) from checkpoint, the shape in current model is torch.Size([4096, 4096]).size mismatch for model.layers.0.mlp.gate_proj.weight: copying a param with shape torch.Size([9728, 2560]) from checkpoint, the shape in current model is torch.Size([14336, 4096]).size mismatch for model.layers.0.mlp.up_proj.weight: copying a param with shape torch.Size([9728, 2560]) from checkpoint, the shape in current model is torch.Size([14336, 4096]).size mismatch for model.layers.0.mlp.down_proj.weight: copying a param with shape torch.Size([2560, 9728]) from checkpoint, the shape in current model is torch.Size([4096, 14336]).size mismatch for model.layers.0.input_layernorm.weight: copying a param with shape torch.Size([2560]) from checkpoint, the shape in current model is torch.Size([4096]).size mismatch for model.layers.0.post_attention_layernorm.weight: copying a param with shape torch.Size([2560]) from checkpoint, the shape in current model is torch.Size([4096]).size mismatch for model.layers.1.self_attn.q_proj.weight: copying a param with shape torch.Size([4096, 2560]) from checkpoint, the shape in current model is torch.Size([4096, 4096]).size mismatch for model.layers.1.self_attn.k_proj.weight: copying a param with shape torch.Size([1024, 2560]) from checkpoint, the shape in current model is torch.Size([1024, 4096]).size mismatch for model.layers.1.self_attn.v_proj.weight: copying a param with shape torch.Size([1024, 2560]) from checkpoint, the shape in current model is torch.Size([1024, 4096]).size mismatch for model.layers.1.self_attn.o_proj.weight: copying a param with shape torch.Size([2560, 4096]) from checkpoint, the shape in current model is torch.Size([4096, 4096]).size mismatch for model.layers.1.mlp.gate_proj.weight: copying a param with shape torch.Size([9728, 2560]) from checkpoint, the shape in current model is torch.Size([14336, 4096]).size mismatch for model.layers.1.mlp.up_proj.weight: copying a param with shape torch.Size([9728, 2560]) from checkpoint, the shape in current model is torch.Size([14336, 4096]).size mismatch for model.layers.1.mlp.down_proj.weight: copying a param with shape torch.Size([2560, 9728]) from checkpoint, the shape in current model is torch.Size([4096, 14336]).size mismatch for model.layers.1.input_layernorm.weight: copying a param with shape torch.Size([2560]) from checkpoint, the shape in current model is torch.Size([4096]).size mismatch for model.layers.1.post_attention_layernorm.weight: copying a param with shape torch.Size([2560]) from checkpoint, the shape in current model is torch.Size([4096]).size mismatch for model.layers.2.self_attn.q_proj.weight: copying a param with shape torch.Size([4096, 2560]) from checkpoint, the shape in current model is torch.Size([4096, 4096]).size mismatch for model.layers.2.self_attn.k_proj.weight: copying a param with shape torch.Size([1024, 2560]) from checkpoint, the shape in current model is torch.Size([1024, 4096]).size mismatch for model.layers.2.self_attn.v_proj.weight: copying a param with shape torch.Size([1024, 2560]) from checkpoint, the shape in current model is torch.Size([1024, 4096]).size mismatch for model.layers.2.self_attn.o_proj.weight: copying a param with shape torch.Size([2560, 4096]) from checkpoint, the shape in current model is torch.Size([4096, 4096]).size mismatch for model.layers.2.mlp.gate_proj.weight: copying a param with shape torch.Size([9728, 2560]) from checkpoint, the shape in current model is torch.Size([14336, 4096]).size mismatch for model.layers.2.mlp.up_proj.weight: copying a param with shape torch.Size([9728, 2560]) from checkpoint, the shape in current model is torch.Size([14336, 4096]).size mismatch for model.layers.2.mlp.down_proj.weight: copying a param with shape torch.Size([2560, 9728]) from checkpoint, the shape in current model is torch.Size([4096, 14336]).

size mismatch for model.layers.2.input_layernorm.weight: copying a param with shape torch.Size([2560]) from checkpoint, the shape in current model is torch.Size([4096]).

”

aaaaqtqt

试了下，问题很多，comfyui无法重启更新，降低安全等级也不行，电脑里装的东西太多了，干脆重装系统彻底清洁一遍，顺便把win10升到win11.

小雨点

aaaaqtqt 发表于 2025-12-5 00:26
在运行z_image_turbo文生图工作流时，在加载CLIP这一步出现问题。

“

你这个问题我遇到过，升级到新版本就好了。

小雨点

支持老大更新！

fxdm123

小雨点 发表于 2025-12-5 08:19
你这个问题我遇到过，升级到新版本就好了。

升级之后原来的qwen工作流生成出一堆乱码要么就无法生成，到采样器那步就显示什么什么值4的请问你们有这情况吗？如何解决？

吊那星

比B站那些卖教程的所谓的老师说得详细多了，牛逼。有些地方本来不明白，现在明了

superXY

wulesoicq 发表于 2025-12-1 10:05
你加载模型时虚拟显存填的多少？

128，我12g的显存

darkerz

darkerz 发表于 2025-12-4 23:57
除了图像1人物，图2和图3都是黑图，请求大佬帮忙

终于找到解决办法了，绘世中高级选项，Cross-Attention 优化方案，选择xFormers

wulesoicq

个人觉的FLUX2模型其实是现在开源模型中最强的，在泛化性，美学特征，提示词遵循程度等多个方面都要明显优于z_image模型，只是参数太大，采样时间长，所以网上才颇多吐槽，无脑吹z_image模型

1. **姿态、动作、身材**：  
   
2. 跪姿，双膝跪地、臀部上翘，身体挺直，手臂弯曲收于身侧；被麻绳捆绑的胸部因绳结勒起，腹部与大腿间绳索收紧，整体呈现饱满丰腴的身材曲线，肌肤因光线与汗液显现出油亮光泽。  
   
3. 
   
4. **发型、妆容**：  
   
5. 齐刘海短发，发丝整齐服帖，略带蓬松感；妆容以淡妆为主，眼妆突出、唇色自然，眉形清晰，凸显日系清新感。  
   
6. 
   
7. **服装、饰品**：  
   
8. 全身被棕色麻绳捆绑，绳结分布在颈部、胸部、腰部及大腿部位，构成紧密绳缚造型；无其他饰品，麻绳质感粗犷，与肌肤形成视觉对比。  
   
9. 
   
10. **场景环境**：  
    
11. 室内日式房间，背景有深色木质门框与立柱，地面铺有榻榻米或草席，整体环境昏暗，仅人物主体被光线聚焦，营造出传统日式空间氛围。  
    
12. 
    
13. **风格，光影**：  
    
14. 日系绳缚艺术风格，光影采用侧光照射，突出肌肤油亮光泽与身体立体曲线；阴影柔和过渡，强化绳缚的束缚感与肌肤质感，整体氛围温润且充满张力。

复制代码

1. **姿态、动作、身材**：  
   
2. 跪姿，双膝跪地、臀部上翘，身体挺直，手臂弯曲收于身侧；被麻绳捆绑的胸部因绳结勒起，腹部与大腿间绳索收紧，身材极端瘦弱，肌肤苍白。  
   
3. 
   
4. **发型、妆容**：  
   
5. 齐刘海短发，发丝整齐服帖，略带蓬松感；妆容以淡妆为主，眼妆突出、唇色自然，眉形清晰，凸显日系清新感。  
   
6. 
   
7. **服装、饰品**：  
   
8. 全身被棕色麻绳捆绑，绳结分布在颈部、胸部、腰部及大腿部位，构成紧密绳缚造型；无其他饰品，麻绳质感粗犷，与肌肤形成视觉对比。  
   
9. 
   
10. **场景环境**：  
    
11. 室内日式房间，背景有深色木质门框与立柱，地面铺有榻榻米或草席，整体环境昏暗，仅人物主体被光线聚焦，营造出传统日式空间氛围。  
    
12. 
    
13. **风格，光影**：  
    
14. 日系绳缚艺术风格，光影采用侧光照射；阴影柔和过渡，强化绳缚的束缚感与肌肤质感，整体氛围温润且充满张力。

复制代码

可以看得出FLUX2在人物生成的皮肤身材等方面的泛化性很强，小伙伴们还是可以尝试玩下FLUX2模型的

小雨点

wulesoicq 发表于 2025-12-7 03:06
个人觉的FLUX2模型其实是现在开源模型中最强的，在泛化性，美学特征，提示词遵循程度等多个方 ...

老大言之有理

pally2024

superXY 发表于 2025-12-6 13:58
128，我12g的显存

朋友，N卡，3060-【12G】可以成功吗？我的总是断开连接，

ZeroW.

牛逼，太牛逼了，这一套教程在外头怎么着也值几千块

2510397440

一些技术问题（在各种途径找了一整天了，没找到解决方法），首先我是下载的git仓库中的amd版本comfui（显卡为RX6750GRE10G），然后前面环境部署和ui本体都没问题，comfyui-manager也搞好了，但是打开ui后拖入楼主分享的工作流后会显示缺失节点，我尝试用manager自动补齐但是并没有起到作用，刷新后依旧缺失，且在manager（安装缺失节点）的界面多个节点会显示（进口失败；import failed）等字样。

wulesoicq

2510397440 发表于 2025-12-9 15:05
一些技术问题（在各种途径找了一整天了，没找到解决方法），首先我是下载的git仓库中的amd版本comfui（显卡 ...

你要发个截图而其 AMD的显卡不支持cuda 所以 有些节点会报错 用官方给的工作流模版试试 尽量不要装插件

yihua99580

小雨点 发表于 2025-12-5 08:19
你这个问题我遇到过，升级到新版本就好了。

这个升级到什么版本呢，我升级到目前0.3.7.7最新的版本，就显示报错，缺少节点213，然后就了好久，排查插件，显示导入失败，一直用不了