Recently we shared a tutorial on using ComfyUI for AI image generation, and its most important feature is the adoption of “nodes” that, from a certain point of view, allow us to “program” its environment. The ChaiNNer processor works the same way, and if you're thinking about batch upscaling or downscaling, generating edges, converting text to images, cropping, adjusting brightness and contrast, or injecting filters, you should give it a look.
Any image editor that wants to carry that title needs to incorporate some basic functions. However, when it comes to processing entire batches of images, resizing them, and automatically correcting brightness and contrast details, it's better to look for a more specialized tool.
With the emergence of AI-based models and the development of new dedicated environments, today we can enjoy tools like ChaiNNer, which uses a node system very similar to ComfyUI. The first versions of ChaiNNer prioritized upscaling, and although it is still in alpha, it can now do much more than that.
How to Process Images with ChaiNNer
Running ChaiNNer is divided into three phases:
- First, download its latest version from GitHub and dump the contents of the zip into a temporary folder.
- Then, we load its executable and install the three main dependencies from within the interface: PyTorch, NCNN, and ONNX. Both NCNN and ONNX have small downloads, but PyTorch is a 2.2 GB brick. Patience.
- Finally, those interested in upscaling need to download one of the models available through OpenModelDB, a kind of Civitai dedicated exclusively to that function. Models can be saved in any folder, since we must select them manually within ChaiNNer.
The official ChaiNNer page presents a basic upscaling example, and in general it will be the main tool. However, it would be a mistake to stay at the tip of the iceberg. ChaiNNer includes batch processing, solid image and gradient creation, text-to-image transformation, edges, cropping, rotation, and even hash generation. Easy to learn, hard to master… but definitely worth it.
Official site and download: Click here.
