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

Eric Qwen-Edit Load Model

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

By EricRollei·Created 5 months ago·Updated 3 months ago· 19
Eric Qwen-Edit Load Model
    • pipeline
    model_pathH:\Training\Qwen-Image-Edit-2511
    precisionbf16
    devicecuda
    keep_in_vramtrue
    offload_vaefalse
    attention_slicingfalse
    sequential_offloadfalse
    vae_tile_blendcosine
    CategoryEric Qwen-Edit

    Inputs (8)

    NameTypeDefaultDescription
    model_pathSTRINGH:\Training\Qwen-Image-Edit-2511Path to Qwen-Image-Edit model (2509 or 2511)
    precisionoptCOMBObf16Model precision (bf16 recommended for RTX 40/50 series)
    deviceoptCOMBOcudaDevice to load model on
    keep_in_vramoptBOOLEANtrueKeep model in VRAM between runs (faster but uses memory)
    offload_vaeoptBOOLEANfalseMove VAE to CPU during transformer inference (saves ~2GB VRAM)
    attention_slicingoptBOOLEANfalseEnable attention slicing (reduces VRAM, slightly slower)
    sequential_offloadoptBOOLEANfalseAggressive CPU offloading - very slow but handles huge images
    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_EDIT_PIPELINE