IF YOU CANT GET IT RUNNING THATS YOUR FAULT NOT MINE. COMPLAIN TO THE CONTRIBUTOR OF THE NODES: https://github.com/smthemex/ComfyUI_LTX2_SM
NOT THE GUY WHO PROVIDES THE WORKFLOW TO YOU. I CANT FIX YOUR PROBLEM AND YOU LOOK LIKE A PRON ADDICT WHEN YOU RAGE IN MY COMMENTS. DONT BE THAT GUY...
LTX-2.3 Sulphur (NSFW MODEL) Distil Workflow — Installation Guide
This workflow runs on smthemex's ComfyUI_LTX2_SM custom node pack with the sulphur_distil distilled transformer GGUF. Below is everything you need to install before queuing the workflow.
1. Custom Nodes
Open a terminal in your ComfyUI custom_nodes directory and clone:
git clone https://github.com/smthemex/ComfyUI_LTX2_SM.git
Then install requirements with the embedded Python (portable users):
cd ComfyUI_LTX2_SM
..\..\..\python_embeded\python.exe -m pip install -r requirements.txt
Or with your venv Python if you're not on portable. Restart ComfyUI fully after install.
Make sure ComfyUI itself is updated to the latest stable — older builds won't have the Gemma / GGUF text encoder plumbing this pack relies on.
Repo: https://github.com/smthemex/ComfyUI_LTX2_SM
2. Models
All four core files come from smthem/LTX-2.3-test-gguf: https://huggingface.co/smthem/LTX-2.3-test-gguf/tree/main
⚠️ Important: Do NOT use a generic Gemma 3 GGUF from Google, Bartowski, Unsloth, etc. The smthemex loader expects HuggingFace-style tensor names. Standard llama.cpp-format GGUFs will throw an UnboundLocalError: embed_tokens_key on load. Only use the Gemma GGUF from the smthem repo above.
Transformer (Sulphur Distil)
File: sulphur_distil-Q6_K.gguf (18.1 GB) Folder: ComfyUI/models/gguf/
Goes in gguf/, not unet/ or diffusion_models/.
OPTIONAL LoRA (FOR VANILLA LTX-2.3)
do not use the lora with the model. the lora is for the regular vanilla LTX-2.3 model in case you have too low of vram to run this NSFW model by itself. the lora works, just not as well as the full model does. DO NOT USE THE LORA WITH THE MODEL! lol
https://huggingface.co/Seregil13th/Sulphur-2-base/blob/main/sulphur_lora_rank_768.safetensors
Text Encoder (Gemma 3)
File: gemma-3-12b-it-qat-Q4_0.gguf (8.7 GB) Folder: ComfyUI/models/gguf/
Connector
File: connector.safetensors (6.34 GB) Folder: ComfyUI/models/checkpoints/
film_net_fp16.safetensors goes in "frame_interpolation" folder
Video + Audio VAEs
Two options — either source works:
Option A — from the smthem repo (same page as everything else):
ltx-2.3-22b-distilled_video_vae.safetensors(1.45 GB) →ComfyUI/models/vae/ltx-2.3-22b-distilled_audio_vae.safetensors(365 MB) →ComfyUI/models/vae/
Option B — from Kijai's LTX2.3_comfy repo: https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/vae
Grab the matching video and audio VAE files from the vae/ subfolder and drop them in ComfyUI/models/vae/.
3. Final Folder Structure
ComfyUI/
└── models/
├── gguf/
│ ├── sulphur_distil-Q6_K.gguf
│ └── gemma-3-12b-it-qat-Q4_0.gguf
├── checkpoints/
│ └── connector.safetensors
└── vae/
├── ltx-2.3-22b-distilled_video_vae.safetensors
└── ltx-2.3-22b-distilled_audio_vae.safetensors
4. Hardware Notes
The smthemex repo lists 6 GB VRAM + 48 GB RAM (peak) as the spec, leaning on the streaming offload code. If you have less system RAM, enable a generous Windows pagefile or you'll hit OOM on the transformer load step. The Q6_K transformer alone is ~18 GB before the encoder and connector come in — there is no shortcut around the RAM requirement.
I HIGHLY RECOMMEND UPDATING YOUR BAT FILE WITH THESE FLAGS:
--lowvram --disable-xformers --use-pytorch-cross-attention --reserve-vram 2 --disable-smart-memory
(these flags will help with the text encoder tricking the push back onto cpu and will burn your vram as priority first)
5. Troubleshooting
UnboundLocalError: embed_tokens_key→ wrong Gemma GGUF. You needgemma-3-12b-it-qat-Q4_0.gguffrom the smthem repo specifically. See warning above.Connector not appearing in dropdown → it goes in
models/checkpoints/, notmodels/clip/ormodels/text_encoders/.Sulphur GGUF missing from dropdown → it goes in
models/gguf/, notmodels/diffusion_models/ormodels/unet/.Out of memory on load → check pagefile size; this workflow benchmarks at ~48 GB peak system memory.
If you hit issues outside this list, drop a comment with the full ComfyUI console traceback (not just the popup) and the file you put in each folder.
Description
FAQ
Comments (110)
i'm getting the following error: AttributeError: 'SiglipVisionModel' object has no attribute 'vision_model'
Disable enhance prompt. Do not change the workflow, its meant to run exactly as set up, its the node sets requirements. The options are just there for future updates, right now the gemma encoder doesnt house the architecture for the config to enhance prompting.
Same problem here.. Enhance prompt is disabled
@Guennibear ahhh so i guess its actually only for T2V at the moment. The gguf encoder doesnt allow for I2V yet. Not sure how i missed that but thats your issue. Disable enhance prompting and only do T2V. No image references.
@realrebelai t2v has same error for me.
@salstephen2451 really? Thats strange. Maybe your dependencies were blocked during node installation? If you have any custom nodes youve downloaded that have required dependencies that dont match the workflows requirements it will give errors like this.
But that being said, i ran your error through claude and it gave me that as answer. That its the image reference or the gguf not recognizing the enhancement for prompting. Im at a lost other than that
Run the workflow again an paste the full command prompt error code into the huggingface discussions on the contributors page for the model. Provide context. Theyre his nodes and hes pretty good about answering and correcting issues in a timely manner. Hes just in China so his time is flip fopped to ours. 12 or 14 hours.
Also, your using the model files in the description right? Your not using a seperate gguf encoder or anything? The workflow REQUIRES his files the nodes are coded specifically to his files.
@realrebelai Tried that and then I get missing token for the clip. My comfyui portable is new, but I upgraded to pytorch 2.11 so I could use mxfp8 models.
if I upgrade my transformer to 4.5; the new error is 'Gemma3ForConditionalGeneration' object has no attribute 'model'
Definitely no "vision_model" in "SiglipVisionModel"
SiglipVisionModel(
(embeddings): SiglipVisionEmbeddings(
(patch_embedding): Conv2d(3, 1152, kernel_size=(14, 14), stride=(14, 14), padding=valid)
(position_embedding): Embedding(4096, 1152)
)
(encoder): SiglipEncoder(
(layers): ModuleList(
(0-26): 27 x SiglipEncoderLayer(
(layer_norm1): LayerNorm((1152,), eps=1e-06, elementwise_affine=True)
(self_attn): SiglipAttention(
(k_proj): Linear(in_features=1152, out_features=1152, bias=True)
(v_proj): Linear(in_features=1152, out_features=1152, bias=True)
(q_proj): Linear(in_features=1152, out_features=1152, bias=True)
(out_proj): Linear(in_features=1152, out_features=1152, bias=True)
)
(layer_norm2): LayerNorm((1152,), eps=1e-06, elementwise_affine=True)
(mlp): SiglipMLP(
(activation_fn): GELUTanh()
(fc1): Linear(in_features=1152, out_features=4304, bias=True)
(fc2): Linear(in_features=4304, out_features=1152, bias=True)
))))
(post_layernorm): LayerNorm((1152,), eps=1e-06, elementwise_affine=True)
)
@delta45424155 I'm also on pytorch 2.11.0
@dlqpsxmm684 try
pip install transformers==4.49.0 --break-system-packages
it may be trasnformers
So I used Tavris1's aio auto installer that comes with older pytorch. And this workflow works with bypassing the external lora node. So it's either this workflow isn't compatible with newer version of comfyui portable or there is some missing requirement that isn't listed.
@dlqpsxmm684 imo, you're wasting your time on this. I got it working and used his own prompt and every time the girl is spinning on a detached dick while screaming. It's nothing like the video's he shared.
@delta45424155 did you use the wrong seed? im not sure what your doing wrong but i literally just posted MORE examples lol dont blame me for your dependency issues. im just a messenger of the model.
@delta45424155 its your requirements being blocked by other dependencies. its not the workflow. mine works just fine nad i just updated to the latest comfy today and retested. results are in the examples now.
@delta45424155 on mine it made a flying dildo with the woman spinning on her back 😯
@salstephen2451 it may be your transformers. something blocked it in the requirements txt.
run this command in your comfy root folder
python_embeded\python.exe -m pip install "transformers==4.53.3"
those are the required transformers. if it doesnt give you those, something in your custom nodes is blocking it. remove custom nodes to a folder run the install again then put your custom nodes bacdk in the folder. only way ive found to bypass dependecy update blocks. just make sure when it finishes running it, the transformers are the exact same numbers, 4.53.3
@salstephen2451 post that result to the gallery, i gotta see it lol
@realrebelai the requirement.txt for this custom node doesn't list a min transformer requirement. It needs to be updated.
@realrebelai so I downgraded to the transformer you listed and now I'm getting a new error metafloattype only dense cpu tensors can be pinned.
@realrebelai Did this and it started to work, then got LTX2_SM_ENCODER
RuntimeError: cannot pin 'MetaFloatType' only dense CPU tensors can be pinned
@tgale67523 you have to enable the swap or whatever option at the bottom of that node or you'll get that error. This node pack has options that never got finished. Very broken.
@delta45424155 thanks for sending me your workflow. It works but sometimes I get oof out of memory?
@salstephen2451 I don't see this anywhere.
@tgale67523 did you try to run the encoder on cpu? it can only run on CUDA.
@delta45424155its not mine, tell the contributor. he mentioned in the discussions about transformer mismatch. thats all i know
@realrebelai No it's set to cuda
@tgale67523 at this point, i recommend installing a fresh comfy and trying it that way, whatever the issue is, if it corrects on a new comfy its your dependencies or something from the requirements getting blocked. i cant help you any further tbh im sorry. i just dont know how else to help :(
try contacting the creator of the nodes on github, he may be able to help. he should be waking up in China as we speak lol
@realrebelai It's cool. I actually just reinstalled comfy today to use my 4 main workflows. I had a mess of custom nodes, but right now it's basically barebones. I barely have any custom nodes. Still doesn't work, but maybe someone will figure it out.
@tgale67523 man you are just having the worst luck with this lol that hurt me to see youu reinstalled today. usually thats the quick fix for me. ill do some digging with claude over the next few hours while im working to see if i cant troubleshoot this and fix your issue
@realrebelai I did bypass the Lora but that's basically what the guide said. Not sure if that makes any difference. Appreciate you trying to help.
@tgale67523 deltas idea of using an aio auto comfyui installer works for this workflow but it uses older versions of stuff and doesn't support the new mxfp8 for 5000 cards. I think he is right because look at the custom nodes date for this workflow. It is much older than the new pytorch and comfyui latest release.
@salstephen2451 I do have a 50 series card. Guess it just hates me. Not about to install another version of comfy just for this. Even though it looks great.
@tgale67523 the two things positive I can say about this workflow is the audio is, oddly, better than what I get from sulpher's workflow and the render is fast.
@delta45424155 Wish I could tell you. Waiting with 64GB of RAM and a 5080 to try but my odds are looking slim
@tgale67523 I also have this MetaFloatType error.. Just cloned latest comfy and it didn't help
@asd123qwerty Yep, just reinstalled today with my regular workflow nodes. Got rid of so much stuff and on the latest version.
@tgale67523 youd be suprised how common it is to install seperate comfys for various models. i have 3 comfys currently. all on my SSD
@asd123qwerty newest comfyui portable didn't work for this workflow for me either. But it did run on https://github.com/Tavris1/ComfyUI-Easy-Install
judging by the node's python code; i'm thinking this node pack was meant for ltx2.0. The LTX2_SM_Model's last boolen option is missing code for the false condition to work. At least that is what Qwen3-Coder (30B) coder is telling me. Just ran it by Claude and it is giving me the code to add that should fix it.
IDK if the issue im having with this workflow is because im on a 5080 and have cu130 toolkit installed or what. OP's naked girl running prompt worked for me on this workflow but she had no nipples lol.
IMO, FP8 base, or eros/sulphur models is the lowest i'd go. Skip the gguf for now. Or go to a wan2 with ltx audio workflow. (I haven't tried any wan /w ltx options so idk if it's even a good option.)
@tgale67523 I'm having success with a 5080 and 64gigs of ram. But i'm using a modified workflow provided by sulphur and it's t2v only.
@realrebelai Not surprised, just not something I'm interested in doing. I have a solid 4 workflows that all work on the latest and I'm too lazy to switch to an old version for 1
ComfyUI/custom_nodes/ComfyUI_LTX2_SM/LTX2/ltx_core/text_encoders/gemma/encoders/encoder_configurator.py", line 157, in create_and_populate
v_model = model.model.vision_tower.vision_model
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
using transformers 4.57.3 can fix the problem
@john2zhang099266 Thank you, this method solved the problem.
@john2zhang099266 Interesting. This did indeed fix the issue. Thank you!
load lora node throws an error; AttributeError: 'OmniPipeline' object has no attribute 'model'
other than that it works on older comfyui, but not the current comfyui. And the results are nothing like the two advertised results. It's just bad...
specifically stated in the description NOT to use the lora with the workflow lol
i also literally just updated comfy and tested it. the results were posted here. currently genning more examples as well. the women running n0od on the beach were from right after i updated.
@realrebelai wasn;t trying to use that lora... but another lora.
AttributeError: 'SiglipVisionModel' object has no attribute 'vision_model'
I see the requirements use torch 2.11, i'm on 2.9.1, reinstall 2.9.1. not sure if this is the issue im seeing here?
same cant fix
you have dependencies in seperate custom nodes blocking the installation of the required dependencies for this node set. its common, pull all your custom nodes and re-run the requirements. then put your custom nodes back in the custom nodes folder. this usually fixes this issue for me.
same here.. if this error does not happen than the next one appears:
LTX2_SM_ENCODER
AttributeError: 'Gemma3TextConfig' object has no attribute 'rope_local_base_freq'
Me too..
@dertobi85331 Yep, had same errors, when was trying to "fix" errors manually by editing nodes code. Looks like gonna spend evening on fixing dependencies.
same
where is the film net fp16 frame interpolation supposed to go? your workflow doesn't work without it and you don't even mention it at all.
forgot to place that in the readme. it goes in your "frame_interpolation" folder.
You say not to use the LORA with the sulphur model but your workflow uses it anyway. The LORA is bugged on the download site and nobody can download it in either case.
ITS IN THE WORKFLOW FOR OTHER LORAS! i included the lora loader NODE so others can use OTHER LORAS to generate with the model. the sulphur lora is in the workflow itself because it tested it with the VANILLA LTX to make sure it worked. just bypass it and move on.
Im simply a messenger providing a model and nodes I DIDNT CREATE to you guys, not my fault if it doesnt work because your comfys arent clean. Do not throw shade at me.
Bro it literally says not to use it. Just bypass it
Hey, how long do think it would take to generate a 5-second clip on an RTX 3060 12GB?
mine with rtx 4050 taking 300sec
@RedditUser981 You replied so late that I answered my own question xD. For me, 5 seconds of footage takes 575-600 seconds, depending on whether I have Chrome open or not. Thanks for the info.
Hi, thanks for the workflow and instructions.
I followed the setup, but LTX2_SM_ENCODER crashes during prompt encoding.
Setup:
ComfyUI 0.21.0, Windows, Python 3.13, Torch 2.11 cu128, RTX 4080 Super 16 GB.
Gemma file: gemma-3-12b-it-qat-Q4_0.gguf
Error:
RuntimeError: Tensor.item() cannot be called on meta tensors
The log shows many “Uninitialized parameters or buffers” for Gemma vision_tower and language_model weights.
I checked the GGUF and it only exposes token_embd.weight, no lm_head.weight. I patched lm_head from embed_tokens, but the model still stays partly on meta tensors.
What exact Gemma file/format is expected for your workflow?
will this work with the new unlsloth Eros/Sulphur GGUF models?
The nodes? They SHOULD but dont quote me, these nodes arent mine sadly. I would put in a discussion on the nodes and ask him if it does
yes, unsloth released a gguf version of sulphur https://huggingface.co/vantagewithai/Sulphur-2-Base-GGUF/tree/main
i'm still trying to make it work with your workflow
@Kingp0dd529t it wont work with the workflow. youd need to code it into the nodes seperately and the contributor who makes them has a special coding format i cant get claude to edit properly. i leave it to him most times. id just ask him to code other gguf formats into his nodes for you. mention unsloth specifically
thanks for the clarification. i got stuck in
AttributeError: 'Gemma3TextConfig' object has no attribute 'rope_local_base_freq'
@Kingp0dd529t yea thats the model node or the sampler node not having the config that the gguf file from smthem has baked in.
Thats why you have to use his sadly. Im sure there are other workflows now for this with oher ggufs, i was just one of the first to put one out so i had to use buddys format
@Kingp0dd529t yea thats the model nide or the sampler node not having the config that the gguf file from smthem has baked in.
Thats why you have to use his sadly. Im sure there are other workflows now for this with oher ggufs, i was just one of the first to put one out so i had to use buddys format
Im kinda new at this but i just cant get this to work even with a fresh comfyui install. I updated everything, installed comfyui manager, added the custom node via git clone, did the dependencies step but when i load into the workflow, it still shows red X's over all of the LTX2 nodes (LTX_SM_Model, LTX_SM_VAE ETC are all red). when i go to "install missing cutom nodes" it just says "Failed to find the following ComfyRegistry list". I also changed the channel type with no success. quite a pickle indeed lol
you used the wrong requirements command. do you have desktop or portable
@realrebelai I have portable
When I try to load a custom lora from CivitAI with the GGUF model, I get the error message: "AttributeError 27 LoraModelLoaderOnly - AttributeError: 'OmniPipeline' object has no attribute 'model' "
Should I use a non-distilled GGUF model?
I dont think smthems nodes are structured for lora intake. Youll probably have to find a new gguf and workflow as his are coded specifically for his files and formatted nodes. Im not sure why he doesnt fork the unet loader and code his architecture into those nodes but it is what it is. I think he likes to make his own format to showcase new ways to load in models
Turns out I had to downgrade transformers and disable the disilled lora addon, then use LTS2_SM_MODEL for LORA loading. Now it works. I hope this helps for anyone, who also has the same problem.
@trashcanjunk365 how to disable the distilled lora addon and use LTS2_SM_MODEL for LORA loading? LTX2_SM_Model node is loading sulphur_distil_Q6_K.ggif
Total waste of time. You might be able to get this to work if you want to spend hours jumping through a million hoops, but i would save your energy and move on.
if you struggled cuz your inexperienced just say that. plenty of people ran it no problem.
hay bro you need perfect comfy ui version for this otherwise you cant run it
complain to the person who made it not the guy who brought it to you
https://github.com/smthemex/ComfyUI_LTX2_SM.git
cant get passed the stick motion of characters , no fulid motion or realistic, also cant set beyound 81 frames, any comments?
quality is like shit guys dont waste time here
I agree, I don't think it's his workflow either, but the small model. The videos I posted were using his own prompts. Who has time to play the seed lotto with ltx2.3?
@delta45424155 oh wow youve been here a whileeeeeeeeeee lmfao you realllyyyyy wanna see bouncing ai b o 0 bs huh buddy....
here lil boy complain to the right person
https://github.com/smthemex/ComfyUI_LTX2_SM.git
complain to the right person next time.
https://github.com/smthemex/ComfyUI_LTX2_SM.git
While the setup guide is not complete and likely leads to the errors detailed before, once this is set up, it actually works surprisingly well. Does what it says on the tin: It runs on low VRAM and produces results it possibly should not achieve when compared to other models on the same hardware, easily outputting 720p clips of 7-10 seconds in a mere 3-4 minutes on a 5060Ti/16GB and 32GB sytem RAM.
The quality is heavily prompt related - obviously - and it helped me tremendously bypassing the frame interpolation node.
Overall, this is good work. Thank you for this, @realrebelai :)
I'm not installing a transformer for one workflow that uses a old unfinished custom node that limits me to it's own nodes...
Installed lower version of transformers. Now I get the error you mentioned above, despite using the Gemma version in the HF-Rep: UnboundLocalError: embed_tokens_key ?
Wrong transformers. Need 4.57.6
@realrebelai I installed lower version of transformers. Now I get the error you mentioned above, despite using the Gemma version in the HF-Repe: UnboundLocalError: embed_tokens_key
Tried to hack in a workaround, but in the next step, the "Load Lora" fails, AttributeError: 'OmniPipeline' object has no attribute 'model'
I give up on this. Thanks for your support.
Getting this error: # ComfyUI Error Report ## Error Details - Node ID: 4 - Node Type: LTX2_SM_ENCODER - Exception Type: NotImplementedError - Exception Message: NotImplementedError: Cannot copy out of meta tensor; no data!
Very new at this, but with a little claude help, got it running on a 3060 12gb with 32gb ram pretty quickly. Biggest issue for me was making sure to use the venv folder when in cmd. Quality is very 2005, but that could be my terrible prompting.
custom audio not working
AttributeError: 'Gemma3TextConfig' object has no attribute 'rope_local_base_freq'
# ComfyUI Error Report
## Error Details
- Node ID: 4
- Node Type: LTX2_SM_ENCODER
- Exception Type: AttributeError
- Exception Message: AttributeError: 'Gemma3TextConfig' object has no attribute 'rope_local_base_freq'
## Stack Trace
```
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 535, in execute
output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 335, in get_output_data
return_values = await asyncmap_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 309, in asyncmap_node_over_list
await process_inputs(input_dict, i)
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 297, in process_inputs
result = f(**inputs)
^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy_api\internal\__init__.py", line 149, in wrapped_func
return method(locked_class, **inputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy_api\latest\_io.py", line 1833, in EXECUTE_NORMALIZED
to_return = cls.execute(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2_node.py", line 222, in execute
positive,negative=encoder_text(clip,prompt,negative_prompt,images,enhance_prompt,save_emb,streaming_prefetch_count)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\load_utils.py", line 80, in encoder_text
context = prompt_encoder(
^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_pipelines\utils\blocks.py", line 469, in call
with self._text_encoder_ctx(streaming_prefetch_count) as text_encoder:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_pipelines\utils\blocks.py", line 451, in textencoder_ctx
self._text_encoder_builder.build(device=torch.device("cpu"), dtype=self._dtype).eval(),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\loader\single_gpu_model_builder.py", line 145, in build
meta_model = self.meta_model_(config, self.module_ops)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\loader\single_gpu_model_builder.py", line 91, in meta_model_
model = module_op.mutator(model)
^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\text_encoders\gemma\encoders\encoder_configurator.py", line 162, in create_and_populate
base = config.rope_local_base_freq
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\.venv\Lib\site-packages\transformers\configuration_utils.py", line 164, in getattribute
return super().__getattribute__(key)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
```
## System Information
- ComfyUI Version: 0.22.3
- Arguments: C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\main.py --user-directory C:\user --input-directory C:\input --output-directory C:\output --front-end-root C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\web_custom_versions\desktop_app --base-directory C:\ --database-url sqlite:///C:/user/comfyui.db --extra-model-paths-config C:\Users\Wolve\AppData\Roaming\ComfyUI\extra_models_config.yaml --log-stdout --listen 127.0.0.1 --port 8005 --enable-manager
- OS: win32
- Python Version: 3.12.12 (main, Feb 12 2026, 00:40:26) [MSC v.1944 64 bit (AMD64)]
- Embedded Python: false
- PyTorch Version: 2.10.0+cu130
## Devices
- Name: cuda:0 NVIDIA GeForce RTX 5060 Ti : cudaMallocAsync
- Type: cuda
- VRAM Total: 17102864384
- VRAM Free: 15908995072
- Torch VRAM Total: 0
- Torch VRAM Free: 0
## Logs
```
2026-05-30T14:35:55.298987 - setup plugin alembic.autogenerate.schemas
2026-05-30T14:35:55.298987 - setup plugin alembic.autogenerate.tables
2026-05-30T14:35:55.298987 - setup plugin alembic.autogenerate.types
2026-05-30T14:35:55.298987 - setup plugin alembic.autogenerate.constraints
2026-05-30T14:35:55.298987 - setup plugin alembic.autogenerate.defaults
2026-05-30T14:35:55.298987 - setup plugin alembic.autogenerate.comments
2026-05-30T14:35:55.471270 - Adding extra search path custom_nodes C:\custom_nodes
2026-05-30T14:35:55.471270 - Adding extra search path download_model_base C:\models
2026-05-30T14:35:55.471270 - Adding extra search path custom_nodes C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes
2026-05-30T14:35:55.471270 - Setting output directory to: C:\output
2026-05-30T14:35:55.471270 - Setting input directory to: C:\input
2026-05-30T14:35:55.471270 - Setting user directory to: C:\user
2026-05-30T14:35:55.693248 - [START] Security scan2026-05-30T14:35:55.693248 -
2026-05-30T14:35:56.657805 - [DONE] Security scan2026-05-30T14:35:56.657805 -
2026-05-30T14:35:56.658797 - ** ComfyUI startup time:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - 2026-05-30 14:35:56.6582026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** Platform:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - Windows2026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** Python version:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - 3.12.12 (main, Feb 12 2026, 00:40:26) [MSC v.1944 64 bit (AMD64)]2026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** Python executable:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - C:\.venv\Scripts\python.exe2026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** ComfyUI Path:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI2026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** ComfyUI Base Folder Path:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI2026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** User directory:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - C:\user2026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** ComfyUI-Manager config path:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - C:\user\__manager\config.ini2026-05-30T14:35:56.659798 -
2026-05-30T14:35:56.659798 - ** Log path:2026-05-30T14:35:56.659798 - 2026-05-30T14:35:56.659798 - C:\user\comfyui.log2026-05-30T14:35:56.659798 -
2026-05-30T14:35:57.538773 - [ComfyUI-Manager] Skipped fixing the 'comfyui-frontend-package' dependency because the ComfyUI is outdated.
2026-05-30T14:35:57.538773 - [PRE] ComfyUI-Manager
2026-05-30T14:35:57.953062 - C:\.venv\Lib\site-packages\torch\cuda\__init__.py:65: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
import pynvml # type: ignore[import]
2026-05-30T14:35:59.661934 - Found comfy_kitchen backend cuda: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_nvfp4']}
2026-05-30T14:35:59.661934 - Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_mxfp8', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_mxfp8', 'scaled_mm_nvfp4']}
2026-05-30T14:35:59.661934 - Found comfy_kitchen backend triton: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}
2026-05-30T14:35:59.666936 - Checkpoint files will always be loaded safely.
2026-05-30T14:35:59.785323 - Total VRAM 16311 MB, total RAM 32389 MB
2026-05-30T14:35:59.785323 - pytorch version: 2.10.0+cu130
2026-05-30T14:35:59.786325 - Set vram state to: NORMAL_VRAM
2026-05-30T14:35:59.786325 - Device: cuda:0 NVIDIA GeForce RTX 5060 Ti : cudaMallocAsync
2026-05-30T14:35:59.798328 - Using async weight offloading with 2 streams
2026-05-30T14:35:59.798328 - Enabled pinned memory 12955.0
2026-05-30T14:35:59.801324 - working around nvidia conv3d memory bug.
2026-05-30T14:35:59.978804 - Using pytorch attention
2026-05-30T14:36:01.912636 - DynamicVRAM support detected and enabled
2026-05-30T14:36:01.913637 - Python version: 3.12.12 (main, Feb 12 2026, 00:40:26) [MSC v.1944 64 bit (AMD64)]
2026-05-30T14:36:01.913637 - ComfyUI version: 0.22.3
2026-05-30T14:36:01.913637 - comfy-aimdo version: 0.3.0
2026-05-30T14:36:01.914636 - comfy-kitchen version: 0.2.8
2026-05-30T14:36:01.916637 - [Prompt Server] web root: C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\web_custom_versions\desktop_app
2026-05-30T14:36:01.916637 - Asset seeder disabled
2026-05-30T14:36:01.917639 - [START] ComfyUI-Manager
2026-05-30T14:36:02.055851 - [ComfyUI-Manager] Using GitPython backend2026-05-30T14:36:02.055851 -
2026-05-30T14:36:02.086850 - [ComfyUI-Manager] network_mode: public
2026-05-30T14:36:02.096851 - [ComfyUI-Manager] The matrix sharing feature has been disabled because the matrix-nio dependency is not installed.
To use this feature, please run the following command:
C:\.venv\Scripts\python.exe -m pip install matrix-nio
2026-05-30T14:36:03.479208 - [Crystools [0;32mINFO[0m] Crystools version: 1.27.4
2026-05-30T14:36:03.503270 - [Crystools [0;32mINFO[0m] Platform release: 11
2026-05-30T14:36:03.503270 - [Crystools [0;32mINFO[0m] JETSON: Not detected.
2026-05-30T14:36:03.504270 - [Crystools [0;32mINFO[0m] CPU: Intel(R) Core(TM) Ultra 7 265F - Arch: AMD64 - OS: Windows 11
2026-05-30T14:36:03.513270 - [Crystools [0;32mINFO[0m] pynvml (NVIDIA) initialized.
2026-05-30T14:36:03.513270 - [Crystools [0;32mINFO[0m] GPU/s:
2026-05-30T14:36:03.521269 - [Crystools [0;32mINFO[0m] 0) NVIDIA GeForce RTX 5060 Ti
2026-05-30T14:36:03.521269 - [Crystools [0;32mINFO[0m] NVIDIA Driver: 595.97
2026-05-30T14:36:04.598827 - [ComfyUI-Easy-Use] server: 2026-05-30T14:36:04.598827 - v1.3.6 2026-05-30T14:36:04.598827 - Loaded2026-05-30T14:36:04.598827 -
2026-05-30T14:36:04.599827 - [ComfyUI-Easy-Use] web root: 2026-05-30T14:36:04.599827 - C:\custom_nodes\comfyui-easy-use\web_version/v2 2026-05-30T14:36:04.599827 - Loaded2026-05-30T14:36:04.599827 -
2026-05-30T14:36:04.615016 - ComfyUI-GGUF: Allowing full torch compile
2026-05-30T14:36:04.618016 - ### Loading: ComfyUI-Impact-Subpack (V1.3.5)
2026-05-30T14:36:04.619015 - [Impact Pack/Subpack] Using folder_paths to determine whitelist path: C:\user\default\ComfyUI-Impact-Subpack\model-whitelist.txt
2026-05-30T14:36:04.619015 - [Impact Pack/Subpack] Ensured whitelist directory exists: C:\user\default\ComfyUI-Impact-Subpack
2026-05-30T14:36:04.620016 - [Impact Pack/Subpack] Loaded 0 model(s) from whitelist: C:\user\default\ComfyUI-Impact-Subpack\model-whitelist.txt
2026-05-30T14:36:04.883643 - [Impact Subpack] ultralytics_bbox: C:\models\ultralytics\bbox
2026-05-30T14:36:04.883643 - [Impact Subpack] ultralytics_segm: C:\models\ultralytics\segm
2026-05-30T14:36:04.883643 - [Impact Subpack] ultralytics_bbox: C:\models\ultralytics\bbox
2026-05-30T14:36:04.883643 - [Impact Subpack] ultralytics_segm: C:\models\ultralytics\segm
2026-05-30T14:36:04.907709 - Blocked by policy: C:\custom_nodes\comfyui-manager
2026-05-30T14:36:04.934701 - [2026-05-30T14:36:04.934701 - UmeAiRT-Toolkit2026-05-30T14:36:04.935701 - ] 🧩 Loaded 62 nodes.2026-05-30T14:36:04.935701 -
2026-05-30T14:36:04.935701 - [2026-05-30T14:36:04.935701 - UmeAiRT-Toolkit2026-05-30T14:36:04.935701 - ] 2026-05-30T14:36:04.935701 - ⚡ Optimisation check:2026-05-30T14:36:04.935701 - SageAttention 2026-05-30T14:36:04.935701 - ❌2026-05-30T14:36:04.935701 - | Triton 2026-05-30T14:36:04.935701 - ✅2026-05-30T14:36:04.935701 -
2026-05-30T14:36:04.935701 - [2026-05-30T14:36:04.935701 - UmeAiRT-Toolkit2026-05-30T14:36:04.935701 - ] 2026-05-30T14:36:04.935701 - 📊 Initial CUDA memory:2026-05-30T14:36:04.935701 - 14.82GB free / 15.93GB total2026-05-30T14:36:04.935701 -
2026-05-30T14:36:04.935701 - [2026-05-30T14:36:04.935701 - UmeAiRT-Toolkit2026-05-30T14:36:04.935701 - ] 2026-05-30T14:36:04.935701 - ✅ Initialization Complete.2026-05-30T14:36:04.935701 -
2026-05-30T14:36:05.122896 - Warning: Could not load sageattention: No module named 'sageattention'
2026-05-30T14:36:05.122896 - sageattention package is not installed, sageattention will not be available
2026-05-30T14:36:05.350347 - WanVideoWrapper WARNING: FantasyPortrait nodes not available: No module named 'onnx'
2026-05-30T14:36:05.371554 - [C:\custom_nodes\comfyui_controlnet_aux] | INFO -> Using ckpts path: C:\custom_nodes\comfyui_controlnet_aux\ckpts2026-05-30T14:36:05.371554 -
2026-05-30T14:36:05.374065 - [C:\custom_nodes\comfyui_controlnet_aux] | INFO -> Using symlinks: False2026-05-30T14:36:05.374065 -
2026-05-30T14:36:05.375064 - [C:\custom_nodes\comfyui_controlnet_aux] | INFO -> Using ort providers: ['CUDAExecutionProvider', 'DirectMLExecutionProvider', 'OpenVINOExecutionProvider', 'ROCMExecutionProvider', 'CPUExecutionProvider', 'CoreMLExecutionProvider']2026-05-30T14:36:05.375064 -
2026-05-30T14:36:05.394743 - C:\custom_nodes\comfyui_controlnet_aux\node_wrappers\dwpose.py:26: UserWarning: DWPose: Onnxruntime not found or doesn't come with acceleration providers, switch to OpenCV with CPU device. DWPose might run very slowly
warnings.warn("DWPose: Onnxruntime not found or doesn't come with acceleration providers, switch to OpenCV with CPU device. DWPose might run very slowly")
2026-05-30T14:36:05.495935 - USDU batch patches applied successfully.
2026-05-30T14:36:05.495935 - USDU batch patches applied successfully.
2026-05-30T14:36:05.516943 - Traceback (most recent call last):
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\nodes.py", line 2204, in load_custom_node
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 999, in exec_module
File "<frozen importlib._bootstrap>", line 488, in callwith_frames_removed
File "C:\custom_nodes\nodes_utility.py", line 6, in <module>
from .utils import log
ImportError: attempted relative import with no known parent package
2026-05-30T14:36:05.516943 - Cannot import C:\custom_nodes\nodes_utility.py module for custom nodes: attempted relative import with no known parent package
2026-05-30T14:36:05.517944 - [PromptModels Studio] Loading custom nodes for ComfyUI v2.0.1...2026-05-30T14:36:05.517944 -
2026-05-30T14:36:05.518935 - [ComfyDeploy] ✅ get_last_frame importado correctamente2026-05-30T14:36:05.518935 -
2026-05-30T14:36:05.518935 - [GetLastFrame] Loaded2026-05-30T14:36:05.518935 -
2026-05-30T14:36:05.519936 - [TextPromptBlocker] Loaded2026-05-30T14:36:05.519936 -
2026-05-30T14:36:05.519936 - [DivisorDePrompts] Loaded2026-05-30T14:36:05.519936 -
2026-05-30T14:36:05.521935 - ✓ GETSETNODE_PRO v1.0.0 loaded2026-05-30T14:36:05.521935 -
2026-05-30T14:36:05.521935 - [GETSETNODE_PRO] Loaded2026-05-30T14:36:05.521935 -
2026-05-30T14:36:05.524935 - ✅ ComfyUI Selectores Pro cargado (4 nodos)2026-05-30T14:36:05.524935 -
2026-05-30T14:36:05.524935 - [Selectores Pro] Loaded2026-05-30T14:36:05.524935 -
2026-05-30T14:36:05.524935 - [BatchEscenas] Loaded2026-05-30T14:36:05.524935 -
2026-05-30T14:36:05.528935 - [GoogleAI] Ruta registrada: /google-ai/health
2026-05-30T14:36:05.528935 -
=================================================================
ComfyUI_GoogleAI v2.0.1 -- 15 nodos
Legacy (12): Texto, Vision, NanoBanana, Imagen 4, Veo 3.1, Diag
PMS_* (3): GeminiChat, NanaBanana, GeminiTTS
Workflows v1.x y v2.0 ambos funcionan.
=================================================================
2026-05-30T14:36:05.528935 -
2026-05-30T14:36:05.528935 - [ComfyUI_GoogleAI] Loaded (v2.0.0 — GeminiChat + NanaBanana + GeminiTTS)2026-05-30T14:36:05.528935 -
2026-05-30T14:36:05.532935 - [ComfyUI_GrokAI] Loaded (v2.0.0 — Chat + Image + Video + TTS + STT)2026-05-30T14:36:05.532935 -
2026-05-30T14:36:05.532935 - [PromptModels Studio] Total nodes: 382026-05-30T14:36:05.532935 -
2026-05-30T14:36:05.532935 - [PromptModels Studio] Ready! v2.0.12026-05-30T14:36:05.532935 -
2026-05-30T14:36:05.534935 - Traceback (most recent call last):
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\nodes.py", line 2204, in load_custom_node
module_spec.loader.exec_module(module)
File "<frozen importlib._bootstrap_external>", line 995, in exec_module
File "<frozen importlib._bootstrap_external>", line 1132, in get_code
File "<frozen importlib._bootstrap_external>", line 1190, in get_data
FileNotFoundError: [Errno 2] No such file or directory: 'C:\\custom_nodes\\WAN2.2-14B-Rapid-AllInOne\\__init__.py'
2026-05-30T14:36:05.535936 - Cannot import C:\custom_nodes\WAN2.2-14B-Rapid-AllInOne module for custom nodes: [Errno 2] No such file or directory: 'C:\\custom_nodes\\WAN2.2-14B-Rapid-AllInOne\\__init__.py'
2026-05-30T14:36:05.536935 -
Import times for custom nodes:
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\websocket_image_save.py
2026-05-30T14:36:05.536935 - 0.0 seconds (IMPORT FAILED): C:\custom_nodes\WAN2.2-14B-Rapid-AllInOne
2026-05-30T14:36:05.536935 - 0.0 seconds (IMPORT FAILED): C:\custom_nodes\nodes_utility.py
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\ComfyUI-GGUF
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\comfyui-custom-scripts
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\comfyui_ultimatesdupscale
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\comfyui-frame-interpolation
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\flashvsr_ultra_fast
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\promptmodels
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\ComfyUI-KJNodes
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\comfyui-umeairt-toolkit
2026-05-30T14:36:05.536935 - 0.0 seconds: C:\custom_nodes\comfyui_controlnet_aux
2026-05-30T14:36:05.536935 - 0.1 seconds: C:\custom_nodes\ComfyUI_LTX2_SM
2026-05-30T14:36:05.536935 - 0.1 seconds: C:\custom_nodes\comfyui-videohelpersuite
2026-05-30T14:36:05.536935 - 0.1 seconds: C:\custom_nodes\cg-use-everywhere
2026-05-30T14:36:05.536935 - 0.2 seconds: C:\custom_nodes\ComfyUI-Crystools
2026-05-30T14:36:05.536935 - 0.3 seconds: C:\custom_nodes\comfyui-impact-subpack
2026-05-30T14:36:05.536935 - 0.3 seconds: C:\custom_nodes\ComfyUI-WanVideoWrapper
2026-05-30T14:36:05.536935 - 1.1 seconds: C:\custom_nodes\comfyui-easy-use
2026-05-30T14:36:05.536935 -
2026-05-30T14:36:05.541944 - Context impl SQLiteImpl.
2026-05-30T14:36:05.541944 - Will assume non-transactional DDL.
2026-05-30T14:36:05.545936 - Failed to initialize database. Please ensure you have installed the latest requirements. If the error persists, please report this as in future the database will be required: Could not acquire lock on database 'C:/user/comfyui.db'. Another ComfyUI process may already be using it. Use --database-url to specify a separate database file.
2026-05-30T14:36:05.610166 - Starting server
2026-05-30T14:36:05.610166 - To see the GUI go to: http://127.0.0.1:8005
2026-05-30T14:36:06.518950 - [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/buttonGroup.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
2026-05-30T14:36:06.518950 - [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
2026-05-30T14:36:06.521950 - [DEPRECATION WARNING] Detected import of deprecated legacy API: /extensions/core/widgetInputs.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
2026-05-30T14:36:06.647305 - [2026-05-30T14:36:06.647305 - UmeAiRT-Toolkit2026-05-30T14:36:06.647305 - ] 2026-05-30T14:36:06.647305 - Bundle Loader:2026-05-30T14:36:06.648305 - 📦 Using local model manifest (cache is fresh).2026-05-30T14:36:06.648305 -
2026-05-30T14:36:06.651306 - [2026-05-30T14:36:06.651306 - UmeAiRT-Toolkit2026-05-30T14:36:06.651306 - ] 2026-05-30T14:36:06.651306 - SeedVR2:2026-05-30T14:36:06.651306 - Using vendored copy2026-05-30T14:36:06.651306 -
2026-05-30T14:36:06.895801 - [ComfyUI-Manager] The ComfyRegistry cache update is still in progress, so an outdated cache is being used.
2026-05-30T14:36:06.958934 - FETCH DATA from: C:\.venv\Lib\site-packages\comfyui_manager\custom-node-list.json2026-05-30T14:36:06.972567 - [DONE]2026-05-30T14:36:06.972567 -
2026-05-30T14:36:06.994181 - [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/button.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
2026-05-30T14:36:10.731037 - got prompt
2026-05-30T14:36:10.740552 - After Max GPU memory allocated: 0.00 GB2026-05-30T14:36:10.741552 -
2026-05-30T14:36:10.742553 - After Max GPU memory allocated: 0.00 GB2026-05-30T14:36:10.742553 -
2026-05-30T14:36:10.815348 - !!! Exception during processing !!! 'Gemma3TextConfig' object has no attribute 'rope_local_base_freq'
2026-05-30T14:36:10.850215 - Traceback (most recent call last):
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 535, in execute
output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 335, in get_output_data
return_values = await asyncmap_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 309, in asyncmap_node_over_list
await process_inputs(input_dict, i)
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py", line 297, in process_inputs
result = f(**inputs)
^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy_api\internal\__init__.py", line 149, in wrapped_func
return method(locked_class, **inputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Wolve\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy_api\latest\_io.py", line 1833, in EXECUTE_NORMALIZED
to_return = cls.execute(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2_node.py", line 222, in execute
positive,negative=encoder_text(clip,prompt,negative_prompt,images,enhance_prompt,save_emb,streaming_prefetch_count)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\load_utils.py", line 80, in encoder_text
context = prompt_encoder(
^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_pipelines\utils\blocks.py", line 469, in call
with self._text_encoder_ctx(streaming_prefetch_count) as text_encoder:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_pipelines\utils\blocks.py", line 451, in textencoder_ctx
self._text_encoder_builder.build(device=torch.device("cpu"), dtype=self._dtype).eval(),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\loader\single_gpu_model_builder.py", line 145, in build
meta_model = self.meta_model_(config, self.module_ops)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\loader\single_gpu_model_builder.py", line 91, in meta_model_
model = module_op.mutator(model)
^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\text_encoders\gemma\encoders\encoder_configurator.py", line 162, in create_and_populate
base = config.rope_local_base_freq
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\.venv\Lib\site-packages\transformers\configuration_utils.py", line 164, in getattribute
return super().__getattribute__(key)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'Gemma3TextConfig' object has no attribute 'rope_local_base_freq'
2026-05-30T14:36:10.851248 - Prompt executed in 0.12 seconds
```
## Attached Workflow
Please make sure that workflow does not contain any sensitive information such as API keys or passwords.
```
Workflow too large. Please manually upload the workflow from local file system.
```
## Additional Context
(Please add any additional context or steps to reproduce the error here)
The nodes are broken, dont use this.
here complain to the guy you actually should be complaining to:
https://github.com/smthemex/ComfyUI_LTX2_SM.git
@realrebelai Wow! You're the creator of this page/article (or whatever it's called) and you're supposed to do your homework and assist users. If you can't assist or (at least) politely reply, you shouldn't be on a site where mature people talk to each other, not like angst Taylor Swift teenage fans.
Or you're simple a nutter (Wow!) > IF YOU CANT GET IT RUNNING THATS YOUR FAULT NOT MINE. COMPLAIN TO THE CONTRIBUTOR OF THE NODES: https://github.com/smthemex/ComfyUI_LTX2_SM
@TribalDiffusion why would i troubleshoot nodes that dont belong to me when theres clearly a github repo designed to take issues? you sound crazy bro. all im doing is providing the car, if you cant drive thats not my problem. not my job to uber you there. especially for free. yall are lucky i even gave it to you. be humble gang. not angry you cant get it done.
MY nodes get troubleshot. accordingly. not my fault yall are mentally inept and cant handle transformers degregation.
keep pullin on my shiii, it doesnt do anything but make yall look thirsty.
I ran this workflow successfully, but the prompt seems to have no effect.Any thoughts?
Did you adjust anything in the workflow by chance?
Getting this error, would appreciate the help if you can :) -
After Max GPU memory allocated: 0.01 GB
[ERROR] !!! Exception during processing !!! 'SiglipVisionModel' object has no attribute 'vision_model'
[ERROR] Traceback (most recent call last):
File "D:\ComfyUI - New 26\ComfyUI\execution.py", line 542, in execute
output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\execution.py", line 341, in get_output_data
return_values = await asyncmap_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\comfyui-lora-manager\py\metadata_collector\metadata_hook.py", line 171, in async_map_node_over_list_with_metadata
results = await original_map_node_over_list(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\execution.py", line 315, in asyncmap_node_over_list
await process_inputs(input_dict, i)
File "D:\ComfyUI - New 26\ComfyUI\execution.py", line 303, in process_inputs
result = f(**inputs)
^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\comfy_api\internal\__init__.py", line 149, in wrapped_func
return method(locked_class, **inputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\comfy_api\latest\_io.py", line 1935, in EXECUTE_NORMALIZED
to_return = cls.execute(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\ComfyUI_LTX2_SM\LTX2_node.py", line 222, in execute
positive,negative=encoder_text(clip,prompt,negative_prompt,images,enhance_prompt,save_emb,streaming_prefetch_count)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\ComfyUI_LTX2_SM\load_utils.py", line 80, in encoder_text
context = prompt_encoder(
^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_pipelines\utils\blocks.py", line 469, in call
with self._text_encoder_ctx(streaming_prefetch_count) as text_encoder:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_pipelines\utils\blocks.py", line 451, in textencoder_ctx
self._text_encoder_builder.build(device=torch.device("cpu"), dtype=self._dtype).eval(),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\loader\single_gpu_model_builder.py", line 145, in build
meta_model = self.meta_model_(config, self.module_ops)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\loader\single_gpu_model_builder.py", line 91, in meta_model_
model = module_op.mutator(model)
^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI - New 26\ComfyUI\custom_nodes\ComfyUI_LTX2_SM\LTX2\ltx_core\text_encoders\gemma\encoders\encoder_configurator.py", line 157, in create_and_populate
v_model = model.model.vision_tower.vision_model
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ComfyUI\ComfyUI\venv\Lib\site-packages\torch\nn\modules\module.py", line 1962, in getattr
raise AttributeError(
AttributeError: 'SiglipVisionModel' object has no attribute 'vision_model'
Paste it in the github repos issues. He made them and i cant help troubleshoot without his code base.
https://github.com/smthemex/ComfyUI_LTX2_SM
@realrebelai he doesnt reply to any of them lmao ive seen like 20 people post this exact issue with not one response
@mrkeeganator805 thats weird, hes usually pretty active everytimr i had contacted him, i just got tired of waiting for quants so i started doing it myself and havent talked to him in a month or so
@mrkeeganator805 im going to contact him myself as he usually responds to me. We have a good rapport between us as ive brought alot of eyes to his HF and github. Let me see what i can do. Youve been respectful and i appreciate that
I downgraded transformers dependency to 4.57.6 (freshest 4 currently), Seems v5 broke some compatibility (not surprised as it's major update)
@FluffyEars this was one of the first implementations of gguf variation to this model. nodes are outdated. find a new one boss