Optimized for rtx 5090/4090
User friendly. Workflow use special nodes:
- DWposeDeluxe Estimator
- LTX Gemma API Text Encode
- LTX Multimodal Guider
- AIO Aux Preprocessor
1) To use this workflow, you need to register on LTX oficial page (full instruction inside workflow) to receive a free key. This process is easier than you think.
2) To install missing nodes use (Manager -> Install missing Custom Nodes)
3) Uses models: ltx-2-19b-distilled.safetensors or ltx-2-19b-distilled-fp8.safetensors / IC LoRA - Union-Control / Z-image turbo models
4) other instructions inside workflow
Description
Optimized for rtx 5090/4090
User friendly. Workflow use special nodes:
- DWposeDeluxe Estimator
- LTX Gemma API Text Encode
- LTX Multimodal Guider
- AIO Aux Preprocessor
1) To use this workflow, you need to register on LTX oficial page (full instruction inside workflow) to receive a free key. This process is easier than you think.
2) To install missing nodes use (Manager -> Install missing Custom Nodes)
3) Uses models: ltx-2-19b-distilled.safetensors or ltx-2-19b-distilled-fp8.safetensors / IC LoRA - Union-Control / Z-image turbo models
4) other instructions inside workflow
FAQ
Comments (4)
Curious, do you mean this workflow integrates a remote API into your local workflow, or do you just need to register to download the model?
Check this out: (End-of-January LTX-2 Drop: Better Control for Real Workflows | LTX Blog) - "A new node that performs Gemma encoding through our API. This endpoint is completely free and exists specifically to help the community run LTX locally." So Lightricks has made life much easier. Previously, you had to use a text encoder in the workflow, now it’s available from their server and for free. This has now reduced the time it takes to load new text into vram when changing text. The main idea behind the new Gemma API Text Encoding (LTX) node is to offload the work of the text encoder (Gemma model) to a remote LTX server.
horrible outputs with literally all same
Try wan2.2-fun-control, quality is better, but you'll have to wait 15 minutes for 10 seconds of video, instead of 3 minutes on LTX-2.