[One Click] Generation On-site Demo
Wan2_2_T2V*文生图_14B-PAseer-480P RunningHub Supported the Generation.
Experience link: https://www.runninghub.ai/ai-detail/1954014983799787522/?inviteCode=rh-v1005
--- 500 RH For Free
24G 18G-Usage 81 frames 480P (480*832)
V2: Average 150 seconds per Video.
Kamado Tanjiro (炭治郎) --LORA is here
V1: Average 170 seconds per Video.
Based on defult workflow and KJnodes
RunningHub Online ComfyUI and AI generator:
Invitation Code: mxkwmeaf --- 500 RH For Free
Description
V2
FAQ
Comments (13)
What affect will this have on the generated vid besides acceleration?
I don't 100% know, the quality is similar. welcome to discuss your results.
Where do I get those 2 Loras you used in this workflow ? LowWan22-炭治郎.safetensors and Wan22-炭治郎.safetensors ?
Tanjiro(炭治郎) is the character from demon slayer.
https://civitai.com/models/1837242/kamadotanjiro-wan22-t2v-paseer
I'm Getting this error;
Prompt execution failed
Prompt outputs failed validation: LoraLoaderModelOnly: - Value not in list: lora_name: '001Wan2.2/Wan22_low_ACClora_PAseer.safetensors' not in ['001Wan2.2\\001Wan2.2Wan22_high_ACClora_PAseer.safetensors', '001Wan2.2\\001Wan2.2Wan22_low_ACClora_PAseer.safetensors', '001Wan2.2\\LowWan22-炭治郎.safetensors', '001Wan2.2\\Wan22-炭治郎.safetensors'] LoraLoaderModelOnly: - Value not in list: lora_name: '003Wan2.2-LORA/Wan22-炭治郎.safetensors' not in ['001Wan2.2\\001Wan2.2Wan22_high_ACClora_PAseer.safetensors', '001Wan2.2\\001Wan2.2Wan22_low_ACClora_PAseer.safetensors', '001Wan2.2\\LowWan22-炭治郎.safetensors', '001Wan2.2\\Wan22-炭治郎.safetensors'] LoraLoaderModelOnly: - Value not in list: lora_name: '004Wan2.2-Low-LORA/LowWan22-炭治郎.safetensors' not in ['001Wan2.2\\001Wan2.2Wan22_high_ACClora_PAseer.safetensors', '001Wan2.2\\001Wan2.2Wan22_low_ACClora_PAseer.safetensors', '001Wan2.2\\LowWan22-炭治郎.safetensors', '001Wan2.2\\Wan22-炭治郎.safetensors'] LoraLoaderModelOnly: - Value not in list: lora_name: '001Wan2.2/Wan22_high_ACClora_PAseer.safetensors' not in ['001Wan2.2\\001Wan2.2Wan22_high_ACClora_PAseer.safetensors', '001Wan2.2\\001Wan2.2Wan22_low_ACClora_PAseer.safetensors', '001Wan2.2\\LowWan22-炭治郎.safetensors', '001Wan2.2\\Wan22-炭治郎.safetensors']
Whats the correct file structure? I have all the files but I can't figure out the dir layout required...
我成功了！太棒了！I got it working! Its Spectacular!
炭治郎的那个LORA没下载，忽略就可以，或者换成其他LORA。--解决了就好。
[One Click] Generation On-site Demo
Wan2_2_T2V*文生图_14B-PAseer-480P RunningHub Supported the Generation.
Experience link: https://www.runninghub.ai/ai-detail/1954014983799787522/?inviteCode=mxkwmeaf
--- 500 RH For Free
Iwent and wrote a 500 word prompt only to be asked to sign in after hitting the "RUN" button after signing in with google my prompt was gone. awesome.
这有点坑了，汗颜中。
Aseer别担心——很高兴你钓到鱼了。Wan 22-I2V 有什么计划吗？我找到了一些 FP4 量化 I2V 扩散器模型，想试试。我把你的模型推到 540P（用 Rife 2x 之后是 1080P），结果好坏参半。不知怎么的，那个女人开始长出三条腿了！
don't worry about it - I'm glad you reeled in a fish. any plans for the Wan 22-I2V? I found FP4 quantized I2V diffuser models I want to try. I pushed your model to 540P (1080P after Rife 2x) and results were mixed. for some reason the woman started growing three legs!
KSampler
CompilationError: at 1:0: def triton_poi_fused_add_0(in_ptr0, in_ptr1, out_ptr0, xnumel, XBLOCK : tl.constexpr): ^ ValueError("type fp8e4nv not supported in this architecture. The supported fp8 dtypes are ('fp8e4b15', 'fp8e5')") Set TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to PyTorch). For even more developer context, set TORCH_LOGS="+dynamo"
I downloaded required loras.
this debuge seems you download/use a wrong version of CKPT of Wan2.2, Wan2.2 fp8 VS. Wan2.2 fp8 scaled VS. Wan2.2 fp8 dilled [These three are different], and some can be used in defult diffusion loader, some can be used in WrapperWan-Kijai node.
So, please check you Base Model of Wan in the Unet Loader/ WrapperKijai Unet Loader or some other nodes.
Yeah I was using wrong models. 3000 cards models can use fp8_e5m2 wan 2.2 models. I don't know fp8_e5m2 results must be worst. by the way great stuff I was impressed how people got outstanding results with your workflow and models. I have 3060ti.