ltx-2.3-spatial-upscaler-x2-1.1.safetensors
Spatial Upscaler x2 v1.1
Official v1.1 latent spatial upscaler — doubles the spatial resolution of LTX 2.3 output inside a two-stage pipeline, working in latent space rather than on decoded pixels. Place in models/latent_upscale_models/.
Released 2026-05-27 · Source: Lightricks/LTX-2.3 (HuggingFace) — Official v1.1 spatial upscaler. Improved quality over the v1.0 release.
Download ltx-2.3-spatial-upscaler-x2-1.1.safetensors
Direct HuggingFace download. ~1 GB · Free.
No 4GB GPU? Try ltx-2.3-spatial-upscaler-x2-1.1.safetensors online — free generation included
Skip the ~1 GB download and ComfyUI setup. Generate a 6-second video using this exact model in your browser, ~30 seconds.
Technical details
ltx-2.3-spatial-upscaler-x2-1.1.safetensors is the official v1.1 latent spatial upscaler for LTX 2.3. It operates on latents, doubling spatial resolution (e.g. 768→1536 on the long edge) before the VAE decode, so detail is added by the model rather than interpolated by a pixel resizer. This is the spatial counterpart to the temporal upscaler (which adds frames); they can be combined in a two-stage graph.
It goes in ComfyUI/models/latent_upscale_models/ — not models/vae/ or models/loras/ — and is loaded by the LTX upscaler node, not a checkpoint or VAE loader. At ~1 GB and 4 GB VRAM it is cheap to add to any workflow.
The typical pattern: generate at a fast base resolution, apply this x2 upscaler as a second pass, then decode. This is faster and sharper than generating at full resolution directly.
When to choose ltx-2.3-spatial-upscaler-x2-1.1.safetensors
Use the x2 v1.1 spatial upscaler when you want higher final resolution without paying the full cost of generating at that resolution from the start — generate small, upscale x2, decode. Prefer it over the v1.0 upscaler for better quality.
Use the x1.5 upscaler instead if x2 overshoots your target resolution or pushes you into OOM at the decode stage — 1.5x is a gentler bump.
Use the temporal upscaler (ltx-2.3-temporal-upscaler-x2-1.0.safetensors) instead — or in addition — when your problem is frame rate / motion smoothness rather than spatial sharpness. The two solve different axes and stack in a two-stage pipeline.
Skip all upscalers entirely for quick drafts or if your base resolution is already where you want it.
Will this run on my GPU?
Minimum: 4GB VRAM.
Recommendation: Optional. Generate at a lower base resolution for speed, then run this x2 latent upscaler as a second stage for a sharper final result. Prefer v1.1 over v1.0. Tiny (~1 GB) and light on VRAM.
How to use ltx-2.3-spatial-upscaler-x2-1.1.safetensors
- Download the file from HuggingFace.
- Place it in ComfyUI/models/latent_upscale_models/ inside your ComfyUI directory.
- Restart ComfyUI (or refresh the model list from the menu).
- Load a compatible workflow — see below.
Compatible official workflows:
- LTX-2.3_T2V_I2V_Two_Stage_Distilled.json— T2V / I2V Two Stage Distilled
Don't want to run this locally? Try ltx-2.3-spatial-upscaler-x2-1.1.safetensors online with a free generation — no GPU, no install, ~30 seconds per clip.
ComfyUI says it can't find ltx-2.3-spatial-upscaler-x2-1.1.safetensors?
Some published workflow JSONs reference this file under a custom subdirectory. If ComfyUI shows a "cannot find model" error and your workflow references one of these path-prefixed variants:
- latent_upscale_models/ltx-2.3-spatial-upscaler-x2-1.1.safetensors
- ltx23\ltx-2.3-spatial-upscaler-x2-1.1.safetensors
The prefix before the slash or backslash is a subdirectory the workflow author used. The actual file is the same ltx-2.3-spatial-upscaler-x2-1.1.safetensors — you have two fixes:
- Create the matching subdirectory inside ComfyUI/models/latent_upscale_models/ and place the file there. Example: if the workflow references latent_upscale_models/ltx-2.3-spatial-upscaler-x2-1.1.safetensors, create the corresponding subfolder under ComfyUI/models/latent_upscale_models/ and put ltx-2.3-spatial-upscaler-x2-1.1.safetensors inside it.
- Or open the workflow JSON in a text editor and replace the prefixed string with just ltx-2.3-spatial-upscaler-x2-1.1.safetensors. ComfyUI then resolves it directly from ComfyUI/models/latent_upscale_models/.
On Windows the separator is \, on macOS/Linux it is / — they refer to the same nested folder regardless of platform.
Common issues
Upscaler node: 'ltx-2.3-spatial-upscaler-x2-1.1.safetensors not in list'▼
File placed in models/vae/, models/loras/, or the root instead of latent_upscale_models/. Fix: Move it to ComfyUI/models/latent_upscale_models/ and click refresh on the upscaler node so ComfyUI re-scans.
OOM at the upscale / decode stage even though base generation was fine▼
Doubling spatial resolution quadruples the latent area; the VAE decode at 2x can spike VRAM well above the base pass. Fix: Switch to the x1.5 spatial upscaler, lower the base resolution, or enable tiled VAE decode in ComfyUI.
Upscaled result looks soft or introduces artifacts▼
Using the v1.0 upscaler, or feeding it latents from a mismatched base resolution. Fix: Use this v1.1 file (improved over v1.0) and keep the base resolution within the range your workflow's upscaler node expects.
ComfyUI doesn't see the file after I downloaded it▼
Make sure the file is in ComfyUI/models/latent_upscale_models/ (not a subfolder). Restart ComfyUI fully — the menu refresh sometimes misses new files. Filename must match exactly: ltx-2.3-spatial-upscaler-x2-1.1.safetensors.
CUDA out of memory error when loading the model▼
ltx-2.3-spatial-upscaler-x2-1.1.safetensors needs ~4GB VRAM minimum. If you're hitting OOM: • Enable Sequential Offloading in ComfyUI settings • Lower the resolution (768×512 instead of 1280×704) — both dimensions must be divisible by 32 • Reduce frame count (65 frames instead of 161) — must be 8n+1 • Use a smaller variant — see Related models below.
How do I apply this LoRA in ComfyUI?▼
Load it in a 'LoraLoader' node and connect it after your model loader. Pair this LoRA with the dev base model (not the distilled one) for the right behavior. LoRA strength 1.0 is the trained value — start there.
Get notified when Spatial Upscaler x2 v1.1 updates
Occasional updates on what's new in LTX 2.3 — new FP8 quants, LoRAs, IC-LoRA releases — with our hands-on verdict on whether they're worth re-downloading. No fixed cadence.
No spam. Sent occasionally when there's real news. Unsubscribe in one click.
Related models
ltx-2.3-22b-distilled-1.1_lora-dynamic_fro09_avg_rank_111_bf16.safetensors
ltx-2.3-22b-distilled-lora-1.1_fro90_ceil72_condsafe.safetensors
LTX-2.3-OmniNFT-RL-Lora_bf16.safetensors
ltx-2.3-22b-distilled-lora-384-1.1.safetensors
ltx-2.3-22b-distilled-lora-384.safetensors
ltx-2.3-22b-distilled-lora-dynamic_fro09_avg_rank_105_bf16.safetensors