Model
Trends
.ai
Model
Trends
.ai
los mejores modelos de IA de código abierto

Nodo Upscale Image and Upscale Image By en ComfyUI

Upscale Image (ImageScale) and Upscale Image By resize an IMAGE with classic interpolation (nearest, bilinear, bicubic, lanczos, area) to an exact size or by a factor, with no AI model involved.
Paquete de nodos: ComfyUI core
Categoría: Escalado
Esta página aún no está traducida, por eso se muestra en inglés.

Qué hace Upscale Image and Upscale Image By

These two nodes are plain resizers. Upscale Image takes a target width and height plus a crop option (disabled or center); Upscale Image By takes a scale factor. The upscale_method choices are the usual filters: nearest-exact for pixel art, bilinear and bicubic for smooth results, lanczos for the sharpest classic scaling, area for downscaling without aliasing.
Despite the name they are just as useful for shrinking: downscale a 4K photo before VAE Encode, or bring the output of a 4x AI upscaler back to 2x. For real detail gain, pair Upscale Image Using Model with these nodes, or use Upscale Latent for a hires pass.

Entradas

Nombre
Tipo
Qué es
image
IMAGE
The picture to resize.
upscale_method
COMBO
nearest-exact, bilinear, area, bicubic or lanczos.
width / height (Upscale Image)
INT
Target size in pixels; 0 keeps the aspect ratio from the other side.
crop (Upscale Image)
COMBO
disabled stretches, center keeps the aspect ratio and crops.
scale_by (Upscale Image By)
FLOAT
Factor, 0.5 halves, 2.0 doubles.

Salidas

Nombre
Tipo
Qué es
IMAGE
IMAGE
The resized picture.
AD
Olvida la instalación: workflows de ComfyUI preinstalados en la nube
BitVector ejecuta workflows de ComfyUI listos para usar en sus propias GPU. Sin instalar nada, sin nodos faltantes, sin cuadros rojos. Ábrelo en el móvil o el portátil y genera en un minuto.

Cómo usar Upscale Image and Upscale Image By

  1. Add Upscale Image By after the picture you want to resize.
  2. Pick lanczos for photos, nearest-exact for pixel art.
  3. Set scale_by (0.5 to halve a 4x upscaler result, 2.0 to double).
  4. Connect the result to the next node (VAE Encode for a second pass, Save Image to keep it).
  5. Use Upscale Image with crop center to make an input exactly 1024x1024.

Ajustes y consejos

  • Setting width or height to 0 in Upscale Image keeps the aspect ratio.
  • A 4x ESRGAN model followed by Upscale Image By 0.5 gives a clean 2x with fewer artefacts than a 2x model on some pictures.
  • Sizes must be multiples of 8 before VAE Encode; 1024, 1152, 1344 are safe.
  • area is the best method for downscaling; lanczos for upscaling.

Solución de problemas de Upscale Image and Upscale Image By

The result is blurry

Por qué ocurre
Interpolation cannot invent detail; a 2x bicubic upscale is a smooth enlargement.

Cómo arreglarlo
Use Upscale Image Using Model with an ESRGAN model, or a hires pass through the sampler, then resize.

The picture is squashed

Por qué ocurre
Upscale Image with crop disabled and a width/height that do not match the source ratio.

Cómo arreglarlo
Set one side to 0, or set crop to center.

VAE Encode fails after resizing

Por qué ocurre
The new size is not a multiple of 8.

Cómo arreglarlo
Choose multiples of 8 (or 64 for SDXL best results).

AD
Ejecuta este workflow desde tu móvil
Todos los workflows de esta página están preinstalados en BitVector con los modelos y nodos personalizados ya colocados. Elige uno, escribe un prompt y listo. Cero configuración, nada que descargar.

Preguntas sobre Upscale Image and Upscale Image By

Which method is best?

lanczos for upscaling photos and renders, area for downscaling, nearest-exact for pixel art.

Does this use the GPU?

Yes, torch interpolation runs on the GPU and is nearly instant.

Nodos relacionados

Upscale Image (using Model)
ComfyUI core
Upscale Image (using Model) runs an ESRGAN-style upscaler (4x-UltraSharp, RealESRGAN, 4x_NMKD-Siax, 4x_foolhardy_Remacri) over an IMAGE and returns it enlarged by the model's fixed factor with added detail.
Upscale Latent and Upscale Latent By
ComfyUI core
Upscale Latent resizes a latent to a new size (or by a factor) so a second KSampler can add detail at higher resolution: the classic hires fix.
Ultimate SD Upscale
ComfyUI_UltimateSDUpscale
Ultimate SD Upscale enlarges a picture with an ESRGAN model, then re-samples it tile by tile through the diffusion model with your prompt at low denoise, adding real detail at 2x-4x without running out of VRAM.
VAE Encode and VAE Encode (for Inpainting)
ComfyUI core
VAE Encode converts a pixel IMAGE into a LATENT for img2img; VAE Encode (for Inpainting) does the same while blanking the masked area so an inpainting model can repaint it.
AD
¿Cansado de arreglar nodos? Deja que lo haga la nube
BitVector mantiene cientos de workflows de ComfyUI instalados, actualizados y probados en GPU rápidas en la nube. Sin Python, sin errores de CUDA, sin límites de VRAM. Funciona desde cualquier navegador.

Más nodos de ComfyUI

CLIP Text Encode (Prompt)
ComfyUI core
CLIP Text Encode turns a text prompt into CONDITIONING using the model text encoder. One node holds the positive prompt, a second one the negative prompt.
ComfyUI Manager
ComfyUI-Manager
ComfyUI Manager is the extension that installs, updates and fixes custom node packs and models from inside the interface, resolves missing nodes in imported workflows and snapshots your setup.
KSampler
ComfyUI core
KSampler runs the denoising loop: it takes the model, prompts and a latent and produces the finished latent image, controlled by seed, steps, cfg, sampler, scheduler and denoise.
Load Checkpoint
ComfyUI core
Load Checkpoint (CheckpointLoaderSimple) opens a .safetensors or .ckpt model file and hands out the three parts every workflow needs: the diffusion MODEL, the CLIP text encoder and the VAE.
Save Image
ComfyUI core
Save Image writes the IMAGE tensor to ComfyUI/output as a PNG, with the whole workflow embedded in the file metadata so the picture can be dragged back into ComfyUI to restore the graph.
VAE Decode
ComfyUI core
VAE Decode converts the sampled LATENT into a pixel IMAGE with the VAE; VAE Decode (Tiled) does the same in tiles for very large images.