LTX Photo Refocus — v1.0
The LTX image-to-video model can unblur a photo! This workflow has two versions - one that uses a VLM and LLM combination to craft a tailored prompt, and a simplified version that uses less VRAM.
Runtime profile
- Access
- External workflow
- Status
- Listed
- Source
- Civitai
- Updated
- Sep 1, 2026
Source excerpt
The LTX image-to-video model can unblur a photo! This workflow has two versions - one that uses a VLM and LLM combination to craft a tailored prompt, and a simplified version that uses less VRAM.
What you need
- Model files
- 5.3 GiB known subtotal
Remaining hardware or download-size estimates are not available yet.
Requirements
Models
6CLIPLoader
PossibleText encoder · Unknown
google_t5-v1_1-xxl_encoderonly-fp16.safetensors
Not resolvedText encoder · Unknown
Lexi (8B) - Llama3 - Q4_K_M.gguf
Not resolvedCheckpoint · Unknown
lmms-lab/llava-onevision-qwen2-7b-ov
File unverifiedVision model · Hugging Face · lmms-lab/llava-onevision-qwen2-7b-ov
Checkpoint · 5.32 GB · SAFETENSORS · fp16 · full · Unknown
VAEDecode
PossibleVAE · Unknown
Node packs
23BasicScheduler
PossibleNode pack · Unknown
CLIPTextEncode
PossibleNode pack · Unknown
DownloadAndLoadLLaVAOneVisionModel
FoundNode pack · Unknown
Fast Groups Muter (rgthree)
Not resolvedNode pack · Unknown
ImageResizeKJ
PossibleNode pack · Unknown
JjkConcat
FoundNode pack · Unknown
JjkText
FoundNode pack · Unknown
KSamplerSelect
PossibleNode pack · Unknown
LLaVA_OneVision_Run
FoundNode pack · Unknown
- Found
Node pack · Unknown
- Found
Node pack · Unknown
LoadImage
PossibleNode pack · Unknown
LTXVApplySTG
FoundNode pack · Unknown
LTXVLoader
Not resolvedNode pack · Unknown
LTXVModelConfigurator
Not resolvedNode pack · Unknown
LTXVShiftSigmas
Not resolvedNode pack · Unknown
PreviewImage
PossibleNode pack · Unknown
RandomNoise
Not resolvedNode pack · Unknown
SamplerCustomAdvanced
Not resolvedNode pack · Unknown
ShowText|pysssss
FoundNode pack · Unknown
STGGuider
Not resolvedNode pack · Unknown
String
PossibleNode pack · Unknown
VHS_VideoCombine
PossibleNode pack · Unknown
AI-generated commentary
AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.
Purpose
LTX Photo Refocus is for using a photo to create an image-to-video result described as able to unblur it. It offers a VLM-and-LLM version for tailored prompt crafting and a simplified version described as using less VRAM.
Flow
The graph includes LoadImage and PreviewImage nodes, plus VHS_VideoCombine.
Inputs and outputs
The stated input is a photo, and the described result is image-to-video.
Setup
Find the workflow on Civitai: https://civitai.com/models/1057138.
Dependencies
Named model files and repository: google_t5-v1_1-xxl_encoderonly-fp16.safetensors, Lexi (8B) - Llama3 - Q4_K_M.gguf, lmms-lab/llava-onevision-qwen2-7b-ov, and ltx-video-2b-v0.9.1.safetensors.
Named workflow components: BasicScheduler, CLIPLoader, CLIPTextEncode, DownloadAndLoadLLaVAOneVisionModel, Fast Groups Muter (rgthree), ImageResizeKJ, JjkConcat, JjkText, KSamplerSelect, LLaVA_OneVision_Run, LLMLoader, and LLMSampler.
Other named workflow components: LoadImage, LTXVApplySTG, LTXVLoader, LTXVModelConfigurator, LTXVShiftSigmas, PreviewImage, RandomNoise, SamplerCustomAdvanced, ShowText|pysssss, STGGuider, String, VAEDecode, and VHS_VideoCombine.
Configuration
Choose the VLM-and-LLM version for tailored prompt crafting; choose the simplified version if lower VRAM use matters.
Suggestion · not verified
Cautions
The listing does not specify a usage license, so check Civitai for applicable terms before use.
Suggestion
Does this need to be edited?
Sign in to send an edit request.
Sources
1 source
Sources
1 sourceSource excerpts
1 excerpt
Source excerpts
1 excerptSource context: 505 downloads · Type Workflows · Base model LTXV
