How to upload a WAN 2.2 Lora

Step 1: Uploading the LoRAs We’ll start by uploading the two versions of this LoRA from CivitAI: Wan 2.2 80s Fantasy Movie LoRA](https://civitai.com/models/789313?modelVersionId=2221806) (for test purpose, you can use your own Lora)

There are two versions of this LoRA, both need to be uploaded.

High Noise Transformer

Low Noise Transformer

Below are the payloads used:

High Noise LoRA Payload

{ "taskType": "modelUpload", "taskUUID": "{{$guid}}", "category": "lora", "air": "runware:800@1", "name": "Wan 2.2 80s Fantasy Movie", "heroImageURL": "https://image.civitai.com/.../1050 (2).jpeg", "downloadURL": "https://civitai-delivery-worker-prod...high_noise.safetensors/", "uniqueIdentifier": "WAN22A14B80FANTASYLORAHIGH", "version": "1.0", "tags": ["Wan", "Video", "Fantasy", "80s", "T2V", "Wan 2.2"], "defaultTransformer": "high", "architecture": "wan_2_2_a14b_t2v", "type": "base", "format": "safetensors", "positiveTriggerWords": "ArsMovieStill, 80s Fantasy Movie Still", "shortDescription": "This LoRA is designed to capture the distinct look and feel of early 80s fantasy movies." }

Low Noise LoRA Payload

{ "taskType": "modelUpload", "taskUUID": "{{$guid}}", "category": "lora", "air": "runware:800@2", "name": "Wan 2.2 80s Fantasy Movie", "heroImageURL": "https://image.civitai.com/.../1050 (2).jpeg", "downloadURL": "https://civitai-delivery-worker-prod...low_noise.safetensors/", "uniqueIdentifier": "WAN22A14B80FANTASYLORALOW", "version": "1.0", "tags": ["Wan", "Video", "Fantasy", "80s", "T2V", "Wan 2.2"], "defaultTransformer": "low", "architecture": "wan_2_2_a14b_t2v", "type": "base", "format": "safetensors", "positiveTriggerWords": "ArsMovieStill, 80s Fantasy Movie Still", "shortDescription": "This LoRA is designed to capture the distinct look and feel of early 80s fantasy movies." }

Important: Allow up to 1 minute after uploading for the LoRAs to fully load. Ensure the correct transformer is set (High or Low). Replace the air ids with your own

Step 2: Running Video Inference with the Uploaded LoRAs Once both LoRAs are uploaded, you can run video inference with them:

Example

Video Inference Payload

{ "taskType": "videoInference", "duration": 5, "model": "runware:200@6", "outputFormat": "mp4", "height": 720, "width": 1280, "numberResults": 1, "seed": 234126, "includeCost": true, "outputQuality": 85, "steps": 20, "lora": [ { "model": "runware:800@1", "transformer": "high", "weight": 1 }, { "model": "runware:800@2", "transformer": "low", "weight": 1 } ], "CFGScale": 2, "acceleration": "high", "positivePrompt": "A cat running on road while wearing red glasses, ArsMovieStill, 80s Fantasy Movie Still", "taskUUID": "{{$guid}}", "fps": 32 }

Note: You don’t need to re-specify the transformer in the inference payload (the system uses the default from upload). It’s included here just for clarity.