tinytensor/4x Upscale — Real-ESRGAN, runs on 8GB
More workflows, including the packs: tinytensor.gumroad.com Takes a 1024x1536 render to 4096x6144 with Real-ESRGAN 4x+. One model load, one scale node, no manual tiling. Measured on an RTX 3070 Ti Laptop with 8GB of VRAM: 14-16 seconds per image at 25 MP. The upscale model is 64 MB. Install: drop the JSON into ComfyUI/user/default/workflows, put RealESRGAN_x4plus.safetensors into models/upscale_models, open the graph, pick your image, queue. Nodes: LoadImage - UpscaleModelLoader (Real-ESRGAN 4x+) - ImageUpscaleWithModel - ImageScaleBy (1.0 for true 4x, 0.5 for a 2x result that keeps the detail
| Author / organization | tinytensor |
|---|---|
| Downloads | 22 (as reported by Civitai) |
| Likes | 1 |
| Task | text-to-image |
| Created | 31 October 2026 |
| Last updated at source | — |
Recent changes
- 31 Oct 2026 Description: Takes a 1024x1536 render to 4096x6144 with Real-ESRGAN 4x+. One model load, one scale node, no manual tiling. Measured on an RTX 3070 Ti Laptop with 8GB of VRAM: 14-16 seconds per image at 25 MP. The upscale model is 64 MB. Install: drop th → More workflows, including the packs: tinytensor.gumroad.com Takes a 1024x1536 render to 4096x6144 with Real-ESRGAN 4x+. One model load, one scale node, no manual tiling. Measured on an RTX 3070 Ti Laptop with 8GB of VRAM: 14-16 seconds per
Changes detected by Silytics between refreshes of the source.
More from tinytensor
- tinytensor/Qwen-Image — Text Inside the ImageModel
- tinytensor/Qwen-Image Edit — Change One Thing, Keep the RestModel
- tinytensor/Qwen-Image 2.1 4-Step — 8GB VRAM (1024² in 20s)Model
- tinytensor/MOSS 2.5D Anime on 8GB VRAM — 1024x1536 in 62s (Apache 2.0)Model
- tinytensor/MOSS Image — Simple Text2Image (8GB VRAM)Model
- tinytensor/Flat Print Style — Taisho Woodblock on Qwen-ImageModel
Silytics shows public metadata from Civitai and links to the original. Information may be incomplete or out of date. Last refreshed 31 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.