Wan 2.2 v1 (T2V)
This is incredible. I did make a few changes to the dataset, added a few variety pieces, and I trained on 720p this time, but... I think the improvement is all Wan 2.2. Can't be certain, but... I'll let the samples speak for themselves.
Also, thanks to K3NK for the ridiculous workflow!
v1.1 (T2V)
This one is absolutely a one-trick pony, but that's kinda the point, right?
All of my videos always save the ComfyUI metadata, so right-click / drag-and-drop for a starting point!
v1.1 (I2V)
Same dataset as the T2V model, just trained on the 480P 14B model.
I2V seems to like higher shift, 12-15
it also seems to like higher CFG, like 6
oh, and a lower model weight on the LoRA, like 0.8
Experiment, and let me know / share your results!
a.k.a "The Adriano Angle"... IYKYK.
Post what you've got with this, I'd love to see some other successful prompts!
My first LoRA upload... This is WIP as I've been learning, but I figured I'd share what I got for feedback.
For images, I use the absolutely fantastic TAME Pony, which I don't think gets anywhere near the credit it deserves.
I followed dtwr434's very helpful Training Data guides for training WAN loras with diffusion-pipe, in fact I don't think I changed any settings at all for this one.
Description
FAQ
Comments (14)
Looks cool but for some reason this LORA throws up errors Tested with 14B model, others on the site work OK
Error while loading Loras: contains unexpected module keys, it is likely that this Lora is for a different model : '['blocks.0.cross_attn.k_img.lora_A.weight', 'blocks.0.cross_attn.k_img.lora_B.weight', 'blocks.0.cross_attn.v_img.lora_A.weight', 'blocks.0.cross_attn.v_img.lora_B.weight', 'blocks.1.cross_attn.k_img....'
Hmmm, I've never had that issue. I've used it both in SwarmUI and with the example workflow in Kijai's WanVideoWrapper - https://github.com/kijai/ComfyUI-WanVideoWrapper/blob/main/example_workflows/wanvideo_480p_I2V_example_02.json
@dngstn32 Hmm OK thanks
@tajimatenshinai837 I haven't tested this lora yet but I get this same error on 2 or 3 other WAN loras, different creators, , I commented about it here https://civitai.com/models/1547711 , Im using wan2.1-i2v-14b-720p-Q6_K.gguf in comfyui
@tajimatenshinai837 i had the same error but i discovered I was using the wrong lora loader , power lora loader by rgthree works fine no errors
Sorry but this Lora is not a T2V. Does not work on T2V.
I don't disagree with that statement at all! I've never gotten a good T2V generation. I did, however, train it on the T2V model, perhaps it's undertrained or something like that. As you can see with my examples, though, it works great with the I2V model.
yea they should label this an i2v lora, I downloaded this for t2v, didn't work but got good i2v gens
@MystAI It's probably undertrained. I've got a few more things in the queue, but I'll revisit this in the near future with a v2 focusing on T2V.
@MystAI Please do, the only other front facing anal is also broken for T2V. You make good loras so an improvement would be so welcome
@jj43797771 Yeah, I want to revisit this one after what I've learned with my other LoRAs. I'l start with a T2V model. Stay tuned!
@dngstn32 You're one of the only people doing anal so thank you. From what I've heard the T2V training makes it fully compatible with I2V. unless I'm wrong? Either way cant wait :D
@jj43797771 / @MystAI check out v1.1 T2V!
@dngstn32 I love you :)
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.