{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiazc6sg2h2s3blbo2ylocozlagyrnx6kccaaiwxscfj6daswnka4e",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mniska6dxl52"
  },
  "path": "/t/trying-to-get-omnimattezero-installed-but-cant-find-the-models/176547#post_2",
  "publishedAt": "2026-06-04T23:16:00.000Z",
  "site": "https://discuss.huggingface.co",
  "tags": [
    "wsbagnsv1/ltxv-13b-0.9.7-dev-GGUF",
    "ltxv-13b-0.9.7-dev-Q8_0.gguf",
    "smthemex/ComfyUI_OmnimatteZero README",
    "smthem/LTX-Video-0.9.7-diffusers-merge",
    "OmnimatteZero_node.py",
    "object_removal.py",
    "wsbagnsv1/ltxv-13b-0.9.7-dev-GGUF files",
    "calcuis/ltxv0.9.7-gguf",
    "calcuis/ltxv0.9.7-gguf files",
    "a-r-r-o-w/LTX-Video-0.9.7-diffusers"
  ],
  "textContent": "Hmm… I found something a little hard to find, but there is a likely substitute. I think it should work if you rename it:\n\n* * *\n\nThe exact filename from the `ComfyUI_OmnimatteZero` README:\n\n\n    LTX-Video-0.9.7-diffusers-Q8_0.gguf\n\n\ndoes not seem to exist publicly on Hugging Face under that exact name, at least not in the obvious linked repo. However, I think the likely intended GGUF is this one:\n\n\n    ltxv-13b-0.9.7-dev-Q8_0.gguf\n\n\nfrom:\n\nwsbagnsv1/ltxv-13b-0.9.7-dev-GGUF\n\nThe reason I think this is the right substitute is that the OmnimatteZero README appears to use a simplified / renamed filename, while the actual available GGUF repos use the LTXV 13B naming convention.\n\n## Short answer\n\nTry this file:\n\nltxv-13b-0.9.7-dev-Q8_0.gguf\n\nPut it here:\n\n\n    ComfyUI/models/gguf/ltxv-13b-0.9.7-dev-Q8_0.gguf\n\n\nThen, if the OmnimatteZero node expects the README-style name or you want to avoid confusion, rename it to:\n\n\n    ComfyUI/models/gguf/LTX-Video-0.9.7-diffusers-Q8_0.gguf\n\n\nIn the node, select the GGUF file from the `gguf` dropdown and set the `dit` / diffusion model field to `none`.\n\n## Why this looks like the right file\n\nThe OmnimatteZero README lists the model locations like this:\n\nComponent | README location | README filename\n---|---|---\nVAE | `ComfyUI/models/vae` | `LTX-Video-0.9.7-vae-diffusers.safetensors`\nDiT / safetensors | `ComfyUI/models/diffusion_models` | `LTX-Video-0.9.7-diffusers.safetensors`\nGGUF | `ComfyUI/models/gguf` | `LTX-Video-0.9.7-diffusers-Q8_0.gguf`\n\nSource:\n\nsmthemex/ComfyUI_OmnimatteZero README\n\nThe README links the DiT/GGUF model line to a Chinese cloud source and to this Hugging Face repo:\n\nsmthem/LTX-Video-0.9.7-diffusers-merge\n\nBut that `smthem` repo only appears to contain the safetensors file:\n\n\n    LTX-Video-0.9.7-diffusers.safetensors\n\n\nI do not see the GGUF file there.\n\nThe likely matching GGUF is instead in:\n\nwsbagnsv1/ltxv-13b-0.9.7-dev-GGUF\n\nThat model card says it is a direct GGUF conversion of the `13b-0.9.7-dev` variant from Lightricks/LTX-Video and that it can be used in ComfyUI with the ComfyUI-GGUF custom node.\n\nRelevant file:\n\n\n    ltxv-13b-0.9.7-dev-Q8_0.gguf\n\n\nFile page:\n\nltxv-13b-0.9.7-dev-Q8_0.gguf\n\n## Suggested download command\n\nFrom your ComfyUI directory:\n\n\n    huggingface-cli download \\\n      wsbagnsv1/ltxv-13b-0.9.7-dev-GGUF \\\n      ltxv-13b-0.9.7-dev-Q8_0.gguf \\\n      --local-dir ./models/gguf\n\n\nOr with an explicit ComfyUI path:\n\n\n    huggingface-cli download \\\n      wsbagnsv1/ltxv-13b-0.9.7-dev-GGUF \\\n      ltxv-13b-0.9.7-dev-Q8_0.gguf \\\n      --local-dir <path-to-your-ComfyUI>/models/gguf\n\n\nOptional rename:\n\n\n    mv \\\n      <path-to-your-ComfyUI>/models/gguf/ltxv-13b-0.9.7-dev-Q8_0.gguf \\\n      <path-to-your-ComfyUI>/models/gguf/LTX-Video-0.9.7-diffusers-Q8_0.gguf\n\n\nOn Windows PowerShell:\n\n\n    Rename-Item `\n      \"<path-to-your-ComfyUI>\\models\\gguf\\ltxv-13b-0.9.7-dev-Q8_0.gguf\" `\n      \"LTX-Video-0.9.7-diffusers-Q8_0.gguf\"\n\n\n## Why renaming should be okay\n\nFrom the OmnimatteZero code, the filename does not appear to be hardcoded. The node registers a `gguf` folder and then exposes files from that folder in the `gguf` dropdown.\n\nRelevant code:\n\nOmnimatteZero_node.py\n\nThe important part is that it adds:\n\n\n    ComfyUI/models/gguf\n\n\nas a model folder, and then the node has separate dropdowns for:\n\n\n    dit\n    gguf\n    vae\n\n\nSo the practical setup should be:\n\n\n    dit  = none\n    gguf = ltxv-13b-0.9.7-dev-Q8_0.gguf\n    vae  = LTX-Video-0.9.7-vae-diffusers.safetensors\n\n\nor, if renamed:\n\n\n    dit  = none\n    gguf = LTX-Video-0.9.7-diffusers-Q8_0.gguf\n    vae  = LTX-Video-0.9.7-vae-diffusers.safetensors\n\n\nThe loader code also has a separate GGUF path. If `gguf_path` is not `None`, it loads the transformer from the GGUF file with `LTXVideoTransformer3DModel.from_single_file(...)` and `GGUFQuantizationConfig(...)`.\n\nRelevant code:\n\nobject_removal.py\n\n## If Q8_0 is too heavy\n\nQ8_0 is large. The `wsbagnsv1` repo also has smaller quantized files:\n\nQuant | File | Approx. size\n---|---|---\nQ8_0 | `ltxv-13b-0.9.7-dev-Q8_0.gguf` | 14 GB\nQ6_K | `ltxv-13b-0.9.7-dev-Q6_K.gguf` | 10.9 GB\nQ5_K_M | `ltxv-13b-0.9.7-dev-Q5_K_M.gguf` | 9.82 GB\nQ4_K_M | `ltxv-13b-0.9.7-dev-Q4_K_M.gguf` | 8.82 GB\nQ3_K_S | `ltxv-13b-0.9.7-dev-Q3_K_S.gguf` | 5.86 GB\n\nRepo file list:\n\nwsbagnsv1/ltxv-13b-0.9.7-dev-GGUF files\n\nFor a first test, `Q4_K_M` may be easier:\n\n\n    huggingface-cli download \\\n      wsbagnsv1/ltxv-13b-0.9.7-dev-GGUF \\\n      ltxv-13b-0.9.7-dev-Q4_K_M.gguf \\\n      --local-dir ./models/gguf\n\n\n## Alternative GGUF repo\n\nThere is also another GGUF repo here:\n\ncalcuis/ltxv0.9.7-gguf\n\nIt contains a similarly named Q8 file:\n\n\n    ltxv-13b-0.9.7-dev-q8_0.gguf\n\n\nFile list:\n\ncalcuis/ltxv0.9.7-gguf files\n\nI would try the `wsbagnsv1` repo first, because its model card explicitly says it is a direct GGUF conversion of the LTXV `13b-0.9.7-dev` variant and mentions ComfyUI / ComfyUI-GGUF usage.\n\n## VAE note\n\nFor the VAE, the OmnimatteZero README points to:\n\na-r-r-o-w/LTX-Video-0.9.7-diffusers\n\nThe README says to place it as:\n\n\n    ComfyUI/models/vae/LTX-Video-0.9.7-vae-diffusers.safetensors\n\n\nSo if your downloaded VAE has a generic name like:\n\n\n    diffusion_pytorch_model.safetensors\n\n\nrename it to:\n\n\n    LTX-Video-0.9.7-vae-diffusers.safetensors\n\n\n## My guess\n\nI think the README filename:\n\n\n    LTX-Video-0.9.7-diffusers-Q8_0.gguf\n\n\nis probably one of these:\n\n  1. a local / renamed copy of `ltxv-13b-0.9.7-dev-Q8_0.gguf`;\n  2. a filename from the Chinese cloud mirror mentioned in the README;\n  3. a simplified name written by the OmnimatteZero wrapper author to match the safetensors naming style.\n\n\n\nBut the actual public HF file that seems to correspond to it is most likely:\n\n\n    ltxv-13b-0.9.7-dev-Q8_0.gguf\n\n\nfrom:\n\nwsbagnsv1/ltxv-13b-0.9.7-dev-GGUF\n\n## Practical checklist\n\n\n    ComfyUI/models/gguf/\n      ltxv-13b-0.9.7-dev-Q8_0.gguf\n      # or renamed:\n      LTX-Video-0.9.7-diffusers-Q8_0.gguf\n\n    ComfyUI/models/vae/\n      LTX-Video-0.9.7-vae-diffusers.safetensors\n\n    ComfyUI/models/diffusion_models/\n      # leave empty / do not use if using GGUF\n\n\nThen in `OmnimatteZero_SM_Model`:\n\n\n    dit  = none\n    gguf = ltxv-13b-0.9.7-dev-Q8_0.gguf\n    vae  = LTX-Video-0.9.7-vae-diffusers.safetensors\n\n\nIf it runs out of memory, try `Q6_K`, `Q5_K_M`, or `Q4_K_M` instead of `Q8_0`.",
  "title": "Trying to get omnimattezero installed but cant find the models"
}