Buy me coffee
https://unfazedanomaly964.gumroad.com/l/buymeacoffee
Pose Lora for XL
Best result with weight between : 0.6-0.8 and the CFG between 6-8
Reviews are really appreciated, i love to see the community use my work.
If you like my work, please tip me
Description
FAQ
Comments (28)
Right now crypto is the only way that people can get these early access loras. While i don't think hard work should be ignored, nor do i feel entitled to everything being free. This situation is also not ideal. Because crypto can truly be an obstacle for those not familiar with it, or not set up to easily work with it.
I now what you are saying. Alternatively, you can support me on gumroad and I can send the buzz to you for the early access
there is another way to technically get buzz, via the onsite civitai game where u basically help label images as sfw or nsfw. u could get enough with that approach / it might a bit
____NULL____ Does it give blue or yellow buzz?
You can buy a Buzz Card or Gift Card or whatever its called then you can convert it to buzz. That could be payed with a Credit Card. At least it did work some weeks ago.
@Hellborn the entire buzz gift card selling service has been not functional for a long time now. Which is what I was talking about here.
@____NULL____ it gives a separate currency of buzz, i have 2000 of that, but i can't spend it. Crypto is literally the only way.
You're not wrong, but it also demonstrates why people, especially who hang around these parts of the internet, should onboard to crypto already, at least for stablecoins. Ethereum is an instrument of freedom.
I'm looking to grab this today, but I only see one safetensors file. Shouldn't there be a high and low noise file and, if not, why not and which side does this attach to?
It uses high noise. I usually don’t do low-noise setups; I stick with WAN 2.1 for low noise.
@unfazedanomaly964 ah ok, thanks
@unfazedanomaly964 weird
@duAIguy why do you think it is weird?
@unfazedanomaly964 because 2.1 isn't as good a model and isn't tuned for the job. Might be why your examples look softer than what folks using your lora already are generating.
@unfazedanomaly964 Because Wan2.2 has been specifically made to use both high and low noise models. They did not release only the high noise model, and then say to use Wan2.1 for the low noise.
But to set that aside for a moment, I guess your approach would allow for the use of 2.1 loras to some degree?
@duAIguy @Supernormal_Stimulus Using Using Wan2.1 instead of Wan2.2 (Low) for low-noise stages won’t significantly affect quality in most workflows. The impact depends on the specific use case and how closely Wan2.1 matches the style of Wan2.2
As for the softer look in my examples. That’s more due to the image and prompt choices, not the LoRAs themselves. The models were used as intended; Wan2.2 High for creative generation, Wan2.1 for refinement.
So does your workflow combine 2.2 high and 2.1 for Low?
@NoNuck Just test with your Workflow and tell me if I am wrong
@Supernormal_Stimulus 2.2 is using 2.1 loras pretty darned well, because "2.2" is really more like "2.1.5" which is why it's still 16fps and still uses the 2.1 VAE whereas the 5B model is its own thing, is 24fps and has a new VAE (it's just not as good and begs the question "why?")
not sure why people are confused. high noise gens most of the movement. low noise just refines the image. for stuff like this, its all about the movement, so the high noise is considerably more important to the generation
There might be something a little funky with the metadata on the pony Lora:
It (at least looks like) there's invalid-JSON formatting in the metadata section.
I'm seeing it through both an online Safetensor metadata viewer (I don't think posting links is allowed so punching "Lora Metadata Viewer" into google and hitting the .io github result gets you there)
It also breaks the import when using "InvokeAI" to import it.
Looks like it still works with Comfy though, so it's not technically broken but I just wanted to give you the heads up
Could you please share the information on Discord? My Discord is UnfazedAnomaly
Sorry for being dumb, but there's no Low Noise Lora, do I use the High Noise for both?
i wonder the same thing...
@Westado you can use wan 2.1 for low noise
@Westado you can use wan 2.1 for low noise
@unfazedanomaly964 I thought the low noise was the 2.2 aspect so is this just 2.1?
@Yoorak_Hunt I can see how that’d be confusing, but just to clarify — 2.1 is the one intended for Low Noise. 2.2 has some updates, but for Low Noise scenarios, I usually stick with 2.1. That said, I’ve been experimenting with WAN 2.2 for Low Noise in some cases. If I already have a working WAN 2.1, I don’t retrain for 2.2. But if I don’t have 2.1 available, then I’ll train a WAN 2.2 Low Noise version.
Details
Files
POV_missionary_Wan2.2_I2V_v1.0.safetensors
Mirrors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
传教士POVH.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
Sex-Missionary-POV_missionary-v1.0_Wan Video 2.2 I2V-A14B_HIGH.safetensors
传教士POVH.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
POV_missionary_Wan2.2_I2V_v1.0.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.