Since people asked me for LoRA for PIXHELL :) here it is
Use it WELL!
Few tips and tricks:
The lora will give you great results for a base and overall pixelart feel, but for optimal pixelart results use this extension for automatic1111:
https://github.com/AUTOMATIC1111/stable-diffusion-webui-pixelization
This way you're results will look like the examples I've uploaded!
Enjoy!
Official YouTube Channel: [CLICK HERE]
Patreon: [CLICK HERE]
Discord: [CLICK HERE]
Description
FAQ
Comments (30)
Getting
RuntimeError: mat1 and mat2 shapes cannot be multiplied (77x768 and 1024x128)
Then got
output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320]"
When used. What's wrong here, anyone know?
are you using controlnet?
@R6SPY Nope. Fresh 1111 webui, your lora is the only one which causes this issue for some reason. Tried multiple other models.
I'll try some other UIs and environments to see if the error persists.
This error comes up when you're trying to use it with SD1.5 or a model based off of that. (The lora is for SD2.1)
@ghzzzt Yeah I figured that was why, still getting an error though. Now it's: RuntimeError: output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320]
Not sure what's up. Fresh install, fresh env. Found some reported issues regarding this error by Googling, but no clear solution. Something about how the lora was made? Guess this one was just not made for me.
@sanasfs The same problem, miserable
@sanasfs @afd2046 to use this lora, you need to put it in the LORA folder with your models in stable diffusion/models/LORA, and load it from there. using the network extension :)
also you need to load on your top left a 2.1 SD model, like illuminati or any model you like. This lora is made to work with all 2.1 768 SD models :) , this is how you get it to work.
its because ur using 1.5 model not 2.1
you cant use 2.1 lora on 1.5 model , you gotta use different models
i have the same problem too ~~
@xzp check the comments above, to solve the issue :), basically don't use the lora on a 1.5 model, use it on a 2.1 SD model.
@R6SPY Already said I'm using a SD 2.1 model. The Lora extension is enabled (by default with A1111). Getting RuntimeError: output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320]. Do you mean something else by "network" extension? Could you be a bit more precise by what you mean exactly? Which extension specifically are you talking about here?
@sanasfs @R6SPY i'm having the RuntimeError: output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320] running SD 2.1 768
@R6SPY thx, i will try later ~
@xzp @SoundGuy @sanasfs in your command bat file add argument --xformers
another user had similar issue. try that, let me know
@R6SPY I've followed the conversation here and tried the suggested fixes - no luck. I have xformers enabled, a 2.1 model, proper resolution, no additional extensions, and files in the right spot.
Same error as others reported: "output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320]"
Everything seems to work fine when I don't enable the lora, oddly enough.
@BrownianMotion Welcome to purgatory. Tried everything myself, xformers, no xformers, half vae, no half, default everything, changing everything. Heck, I even installed a linux distro instead of using Windows, still the same. I'm convinced there's something fucky about this lora that is doing this. I tested 120 other loras, all worked fine.
Hi, I am having a similar problem. It also happens when using sd 2.1 too. It's too bad because this lora looks reeally cool
@bosebangho1 @BrownianMotion guys share me your pc specs, i want to double check something. And see if I can share a solution.
@R6SPY Admittedly I'm on the bottom end of required specs for 2.1 models, which could contribute perhaps? I should also note that I run with --lowvram, for reasons that should be clear below, haha.
GPU NVIDIA GTX 970M VRAM - 3GB
Processor Intel(R) Core(TM) i7-4710HQ CPU @ 2.50GHz 2.49 GHz
Installed RAM 16.0 GB
System type 64-bit operating system, x64-based processor
Thanks for following up so diligently!
@R6SPY WIN10+5900X+3090+64G
@BrownianMotion the ram may be the main issue in your scenario.
@afd2046 @bosebangho1 @BrownianMotion @sanasfs I've updated the lora try V 2.0 (in the list bellow you'll see it) let me know did it solve the issue for any of you.
@SPYBG The problem has been resolved. Thank you for your update
@afd2046 my pleasure my man :P
I still have this problem on my iMac M1:RuntimeError:
AttributeError: 'NoneType' object has no attribute 'startswith'
activating extra network lora with arguments [<modules.extra_networks.ExtraNetworkParams object at 0x2a2bd0520>]: RuntimeError
Traceback (most recent call last):
File "/Users/shelly/Documents/stable-diffusion-webui/modules/extra_networks.py", line 75, in activate
extra_network.activate(p, extra_network_args)
File "/Users/shelly/Documents/stable-diffusion-webui/extensions-builtin/Lora/extra_networks_lora.py", line 23, in activate
lora.load_loras(names, multipliers)
File "/Users/shelly/Documents/stable-diffusion-webui/extensions-builtin/Lora/lora.py", line 170, in load_loras
lora = load_lora(name, lora_on_disk.filename)
File "/Users/shelly/Documents/stable-diffusion-webui/extensions-builtin/Lora/lora.py", line 132, in load_lora
module.weight.copy_(weight)
RuntimeError: output with shape [126, 320] doesn't match the broadcast shape [126, 320, 126, 320]
@SPYBG Hey great work on what you have built. I'm still have a weight issue.
AI\stable-diffusion-webui\extensions-builtin\Lora\lora.py", line 176, in load_lora
module.weight.copy_(weight)
RuntimeError: output with shape [126, 320] doesn't match the broadcast shape [126, 320, 126, 320]
I've loaded the v2 model and i'm not sure what to try next.
I tried applying this lora in my automatic111 with SD-2.1-768 - i'm getting this error:
Activating extra network lora with arguments [<modules.extra_networks.ExtraNetworkParams object at 0x000002134BAFE5C0>]: RuntimeError
Traceback (most recent call last):
File "G:\stable-diffusion-webui\modules\extra_networks.py", line 75, in activate
extra_network.activate(p, extra_network_args)
File "G:\stable-diffusion-webui\extensions-builtin\Lora\extra_networks_lora.py", line 23, in activate
lora.load_loras(names, multipliers)
File "G:\stable-diffusion-webui\extensions-builtin\Lora\lora.py", line 151, in load_loras
lora = load_lora(name, lora_on_disk.filename)
File "G:\stable-diffusion-webui\extensions-builtin\Lora\lora.py", line 113, in load_lora
module.weight.copy_(weight)
RuntimeError: output with shape [128, 320] doesn't match the broadcast shape [128, 320, 128, 320]
what am I missing?
try adding --xformers in your arguments, also make sure you're model you've loaded in SD is 2.1 based model, then add the lora. and it should work, few people had similar issues.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.