Imglarger vs Real-ESRGAN

Imglarger is a collection of AI and browser-based tools for working with images. Its tools enlarge, upscale, enhance, sharpen, uncrop, remove backgrounds, restore, edit, and convert images. Enlargement options include 2x, 4x, and 8x, depending on the image and plan; the upscaler offers 2x or 4x resolution increases. An image extender can expand beyond borders and change aspect ratios. Browser editing and conversion include batch conversion, optional resizing, and ZIP downloads, and do not use credits. Supported formats are JPG, JPEG, PNG, and WebP. Some tools and plans support batch processing, while focused generative workflows process one image at a time. Imglarger is available on web, Android, iOS, and macOS. The free plan includes 100 credits per month, one image at a time, a 1024-pixel maximum input side, and 2x scaling. Premium costs $9.99/month; Business costs $19.99/month. The maker says cloud originals and results are generally deleted within two hours, with an exception for some 8x images, while local features do not upload files.

Real-ESRGAN is a free, open-source project developing algorithms for image and video restoration. It offers pretrained models for general images, anime images and animation video, including AnimeVideo-v3 and smaller anime models. Image inputs include JPG, PNG and WEBP; inference can handle alpha-channel, grayscale and 16-bit images, and outputs can be JPG, PNG or WebP. Portable inference supports scale ratios of 2, 3 or 4, with 4 as the default; the Python implementation also offers arbitrary output scaling. Portable NCNN executables are available for Windows, Linux and macOS, with required binaries and models included and no CUDA or PyTorch environment needed. These executables do not include every Python feature, including outscale. Python use requires Python 3.7 or newer and PyTorch 1.7 or newer. Online inference is available through an ARC demo and Colab demos. GFPGAN is integrated for face enhancement, and the training code supports fine-tuning on user data or paired data. The project uses synthetic training data and is released under the BSD-3-Clause license.