Nodes/Eric_Qwen_Edit_Experiments/Eric Qwen-Image Load Model
ComfyUI Node

Eric Qwen-Image Load Model

A ComfyUI node in Eric Qwen-Image with 8 inputs and 1 output.

By EricRollei·Created 5 months ago·Updated 3 months ago· 19
Eric Qwen-Image Load Model
    • pipeline
    model_path
    precisionbf16
    devicecuda
    keep_in_vramtrue
    offload_vaefalse
    attention_slicingfalse
    sequential_offloadfalse
    vae_tile_blendcosine
    CategoryEric Qwen-Image

    Inputs (8)

    NameTypeDefaultDescription
    model_pathSTRINGPath to Qwen-Image or Qwen-Image-2512 model directory
    precisionoptCOMBObf16Model precision (bf16 recommended for RTX 40/50 series)
    deviceoptCOMBOcudaDevice to load model on
    keep_in_vramoptBOOLEANtrueCache model between runs (faster but uses VRAM)
    offload_vaeoptBOOLEANfalseMove VAE to CPU during transformer inference (saves ~1 GB)
    attention_slicingoptBOOLEANfalseTrade speed for lower peak VRAM
    sequential_offloadoptBOOLEANfalseExtreme VRAM savings via sequential CPU offload
    vae_tile_blendoptCOMBOcosineTile-seam blending for tiled VAE decode/encode. • cosine — C¹-smooth blend (zero slope at tile edges), eliminates faint grid lines on smooth gradients. • linear — original diffusers behaviour. Slightly sharper at tile edges; may show faint seams at high resolutions.

    Outputs (1)

    NameTypeDescription
    pipelineQWEN_IMAGE_PIPELINE