No restoration path in the LTX-2.5 family: Deblur and Decompression IC-LoRAs are a different trade from re-rendering

[dkackman]

tool/endpoint: templates/ltx2/* — no restoration pass exists; two-stage is the only sharpening route

repro: the family's only answer to a soft or noisy render is templates/ltx2/two-stage
(eight sigmas at 768x448, 2x latent upsample, three stage-two sigmas at 1536x896) or
templates/ltx2/generative-upscale. Both re-render. Neither cleans up an existing result.

what's now available: two LTX-2.5 IC-LoRAs from 2026-09-10:

• LTX-2.5-22b-IC-LoRA-Deblur — ltx-2.5-22b-ic-lora-deblur-0.9.safetensors
• LTX-2.5-22b-IC-LoRA-Decompression — ltx-2.5-22b-ic-lora-decompression-0.9.safetensors

why it is worth measuring rather than assuming. These are a genuinely different trade from
what the family has: restoration over a cheap low-resolution render, versus re-rendering at
higher cost. That is the exact decision that came up in practice today on the H3 side — a
grainy, mushy background at 544p, where the only routes on offer were "render bigger" (which hit
the 24 GB ceiling) or "accept it". Whether a restoration pass beats a refine pass on quality per
minute is an open question, and answering it needs a comparison, not an adoption.

proposed: one template for each, plus a measured comparison against two-stage on the same
source clip — quality and wall clock.

same cheap-plumbing point as #151: generative-upscale already runs an IC-LoRA step through
LTX2ReferenceCondition / reference_conditions=, so these are templates and prompts, not
engine work.

caveats: both weights are 0.9 previews (only Clean-Plate shipped at 1.0), and both repos
are gated: auto — lem has access, but the cards could not be read unauthenticated here, so
the input format each expects is unverified.

related: #151 (Ingredients, same IC-LoRA batch); full write-up on #151.

[dkackman]

Blocked on the same thing as #151, and parked with Don for the same reason. Worked as model opus via provider anthropic.

Neither weight is reachable. The gates on these repos have not been accepted by the account lem's Hugging Face token belongs to. Checked from lem with that token: Deblur and Decompression return 403 on both the card and a file range request, while the upscaler repo dw already uses returns 200 on both. So gated: auto is per repo, and this 2026-09-10 batch was never accepted — the inference on #151 that access came along with the upscaler does not hold.

That blocks the measurement this issue is actually asking for, not just the download. You were right that the interesting question is whether a restoration pass beats a refine pass on quality per minute, and that it needs a comparison rather than an adoption. I cannot set up a comparison against two-stage on weights I cannot fetch, and neither can the tester.

Also worth stating before anyone builds it: the input format is unconfirmed for exactly the same reason (the cards are 403). For a restoration IC-LoRA the open questions are whether the degraded clip goes in as a LTX2ReferenceCondition at reference_downscale_factor: 1 (unlike the upscaler's 2), what conditioning_attention_strength it wants, and whether the prompt should describe the clean target or the degraded source. Guessing those produces a template that validates and restores nothing in particular.

To unblock: accept the licence on
https://huggingface.co/Lightricks/LTX-2.5-22b-IC-LoRA-Deblur and
https://huggingface.co/Lightricks/LTX-2.5-22b-IC-LoRA-Decompression
as the account whose token is on lem, then hand back with owner:implementer and no status label. Both are 0.9 previews, as you noted — I would still build them as two templates and measure rather than assume, which is what you proposed.

When that happens, the comparison worth running is the one you framed: a cheap low-resolution render restored, against two-stage's refine, on the same source clip, scored on quality and wall clock. I would add the same run through templates/ltx2/diffusion-decode (shipped for #153), since that is a third answer to "the picture is soft" that costs only a decode.

[dkackman]

licenses accepted