ComfyUI Node
📏 Qwen Image Scale V2
A ComfyUI node in ArchAi3d/Qwen/Image with 23 inputs and 7 outputs.
📏 Qwen Image Scale V2
- image
- mask
- image_vl
- image_latent
- mask
- latent_width
- latent_height
- stitch_data
- debug_text
◄crop_modedisabled►
◄crop_padding64►
◄context_factor1.0►
◄blend_pixels16►
◄aspect_ratio_modeauto►
◄preferred_aspect_ratio16:9 (Panorama)►
◄vl_target_area147456►
◄vl_divisible_by32►
◄latent_target_area1763584►
◄latent_divisible_by32►
◄latent_area_tolerance0.30►
◄vl_use_latent_sourcetrue►
◄vl_ignore_latent_letterboxtrue►
◄vl_ignore_latent_croptrue►
◄vl_upscale_methodarea►
◄vl_cropdisabled►
◄vl_letterboxfalse►
◄latent_upscale_methodlanczos►
◄latent_cropdisabled►
◄latent_letterboxfalse►
◄debugtrue►
CategoryArchAi3d/Qwen/Image
Inputs (23)
| Name | Type | Default | Description |
|---|---|---|---|
| image | IMAGE | — | |
| crop_modeopt | COMBO | disabled | disabled: Use full image (V1 behavior). mask_crop: Crop to mask region with padding/context. |
| crop_paddingopt | INT | 640–512 | Fixed pixel padding around mask bounding box. |
| context_factoropt | FLOAT | 1.01–3 | Context expansion factor (1.0 = no expansion, 1.5 = 50% extra context). Applied after padding. |
| blend_pixelsopt | INT | 160–128 | Feather width for seamless blending in stitch node. |
| aspect_ratio_modeopt | COMBO | auto | Auto: Find closest preferred aspect ratio. Manual: Use selected ratio. |
| preferred_aspect_ratioopt | COMBO | 16:9 (Panorama) | Manually select preferred aspect ratio (only used in manual mode). |
| vl_target_areaopt | INT | 1474563000–500000 | Target pixel area for QwenVL output (~147K optimal). |
| vl_divisible_byopt | INT | 321–256 | VL dimensions divisible by this number. |
| latent_target_areaopt | INT | 17635843000–4500000 | Target pixel area for latent output. |
| latent_divisible_byopt | INT | 321–256 | Latent dimensions divisible by this number. |
| latent_area_toleranceopt | FLOAT | 0.300.1–0.8 | Flexibility for latent pixel count (±%). |
| vl_use_latent_sourceopt | BOOLEAN | true | True: Use Latent as VL source (pixel-perfect). False: Use original input. |
| vl_ignore_latent_letterboxopt | BOOLEAN | true | When using Latent source, ignore vl_letterbox setting. |
| vl_ignore_latent_cropopt | BOOLEAN | true | When using Latent source, ignore vl_crop setting. |
| vl_upscale_methodopt | COMBO | area | Resampling algorithm for VL. |
| vl_cropopt | COMBO | disabled | Crop mode for VL output. |
| vl_letterboxopt | BOOLEAN | false | Add black bars to preserve aspect ratio. |
| latent_upscale_methodopt | COMBO | lanczos | Resampling algorithm for latent. |
| latent_cropopt | COMBO | disabled | Crop mode for latent output. disabled: stretch to fit (no content lost). center: crop to fill frame. |
| latent_letterboxopt | BOOLEAN | false | Add black bars to latent. |
| maskopt | MASK | Mask for crop mode (white = region to crop) and output scaling. | |
| debugopt | BOOLEAN | true | Show detailed debug info. |
Outputs (7)
| Name | Type | Description |
|---|---|---|
| image_vl | IMAGE | — |
| image_latent | IMAGE | — |
| mask | MASK | — |
| latent_width | INT | — |
| latent_height | INT | — |
| stitch_data | STITCH_DATA | — |
| debug_text | STRING | — |