same thing, can't choose the loras i saved!

Same. It is a fixed value and not showing the same list as the standard LoadLoRA node.

Temp fix: Save your config.json file, open it in notepad++. Search and replace sd_xl_offset_example-lora_1.0.safetensors with SDXL\\sd_xl_offset_example-lora_1.0.safetensors (or whatever you called it) - note double backslash, then reload in ComfyUI.

@Gibbo @Nexustar thanks for confirming

@searge Which LoRA path is it looking for? The standard LoadLoRA node is finding them in the paths defined in extra_model_paths.yaml without issue.

@searge Yes I have SDXL installed along with other LoRAs. The field only recognises the offset example; If I click to get the drop down, or the arrows to browser through others, it cannot see any of my other LoRAs installed in the same (correct) directory.

fresh install of ComfyUI fixed it for me

@searge update, actually not fixed. got an error in the old version. Throwing Loras it in the wrong folder (checkpoints) on purpose got rid red of error message, but others kept coming. Old install might have been broken anyway. So now same problem as everybody else. l'll run with the tempfix Nexustar suggested. Thx for the awesome work anyway.

I use the portable ComfyUI and deleted the entire folder and put a fresh portable in. Still will only allow the sd_xl_offset_example-lora_1.0.safetensors to be used and shows no other Loras. I am able to select all Loras via the LoraLoader though. So ComfyUI is seeing them but for some reason your LoRA Model node will not show any of them, other than the sd_xl LoRA.

Mine was a fresh install as of 31 July. Did a git pull after I noticed the bug on the LoRAs which didn't resolve anything. If I hand edit the JSON, I can manually insert the names of LoRAs I have and they work fine. Indicates that ComfyUI can load and use them without issue and that they are installed in the correct location.

I had the same problem, I solved it by right clicking on the node and selecting convert_loraname to imput, something like that, and from there I can then select the lora

@ideeker Thanks,. I'll play around and see if I can replicate.

@ideeker I passed this on in more detail on the git page under an issue that someone had already started.

I also have the same problem. I restart the ui. Only shows the example lora

I also have this problem, and I even tested it with having the loras both in Comfy's own directory and the path-linked A1111 directory. I can change the control_after value, but not the lora itself.

I had the same problem, I solved it by right clicking on the node and selecting convert_loraname to input, something like this, and from there I can then select the lora

@ideeker It was Convert to widget, that that seems to have helped indeed! Thanks!

@searge Setting up MemoryEfficientCrossAttention. Query dim is 768, context_dim is 1280 and using 12 heads.

Setting up MemoryEfficientCrossAttention. Query dim is 768, context_dim is None and using 12 heads.

Setting up MemoryEfficientCrossAttention. Query dim is 768, context_dim is 1280 and using 12 heads.

model_type EPS

adm 2560

making attention of type 'vanilla-xformers' with 512 in_channels

building MemoryEfficientAttnBlock with 512 in_channels...

Working with z of shape (1, 4, 32, 32) = 4096 dimensions.

making attention of type 'vanilla-xformers' with 512 in_channels

building MemoryEfficientAttnBlock with 512 in_channels...

missing {'cond_stage_model.clip_g.transformer.text_model.embeddings.position_ids'}

DDIM Sampler: 100%|████████████████████████████████████████████████████████████████████| 20/20 [00:38<00:00, 1.95s/it]

DDIM Sampler: 100%|██████████████████████████████████████████████████████████████████████| 5/5 [00:05<00:00, 1.14s/it]

making attention of type 'vanilla-xformers' with 512 in_channels

building MemoryEfficientAttnBlock with 512 in_channels...

Working with z of shape (1, 4, 32, 32) = 4096 dimensions.

making attention of type 'vanilla-xformers' with 512 in_channels

building MemoryEfficientAttnBlock with 512 in_channels...

C:\AI\ComfyUI_windows_portable\ComfyUI\nodes.py:1228: RuntimeWarning: invalid value encountered in cast

img = Image.fromarray(np.clip(i, 0, 255).astype(np.uint8))

C:\AI\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\_utils.py:776: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()

return self.fget.__get__(instance, owner)()

Prompt executed in 80.46 seconds

Yep having this same issue. first I loaded in the JSON file but it showed error related to Upscale. But then used the cat image( text to image) and loaded in a workflow. It atleast started the work flow but gave me black pictures.

@DarkCover still no idea to fix it?

@k1ngcurses I was able to fix the black images but still can't run the nodes without the upscale(which takes like 30min).

So what I initially I did was update the nodes from the comfymanager.( Which is what I think is the cause of the problem.)

So what I did is download the nodes from here and then I copy pasted it in the custom node folder.

Then replaced existing files. And that worked for me.

But still the issue that persists for me is that I can't use this without the upscale( By disconnecting the upscale nodes). Which still gives me error. other than that text to image still works.

@DarkCover ahh okay thanks bro